key: cord- -jcvrqeew authors: gelain, maria elena; bonsembiante, federico title: acute phase proteins in marine mammals: state of art, perspectives and challenges date: - - journal: front immunol doi: . /fimmu. . sha: doc_id: cord_uid: jcvrqeew the term “acute phase response” (apr) is referred to a nonspecific and complex reaction of an organism that occurs shortly after any tissue damage, such as infection, trauma, neoplasia, inflammation, and stress. the apr can be identified and monitored with some laboratory tests, such as the concentration of several plasma proteins, the acute phase proteins (apps). the apps are components of the non-specific innate immune response, and their plasma concentration is proportional to the severity and/or the extent of tissue damage. the evaluation of health status of marine mammals is difficult because the classical clinical signs of illness used for human and domestic animals are difficult to recognize and understand. for this reason, in the past years, several efforts were done to identify laboratory markers of disease in these animals. the apps have demonstrated their role as early markers of inflammation in veterinary medicine, thus several apps were tested in marine mammals, such as c-reactive protein (crp), serum amyloid-a (saa), and haptoglobin (hp). however, the difficulty to extrapolate the knowledge about apps in one species to another, the lack of specie-specific reagents, the absence of data about negative apps have hampered their extent use in marine mammals. herein, the state of art of apps in marine mammals is reviewed, with particular attention to pre-analytical and analytical factors that should be taken into account in validation and interpretation of apps assays. moreover, the current application, potential utility and the future developments of apps in marine mammals is highlighted and discussed. the mammalian immune system includes innate or nonspecific immunity as well as adaptive or specific immunity. the responses of these two different pathways are distinct, but highly interconnected. the first reaction of an organism to different pathological conditions is an innate, non-specific response ( ), a more conserved response during evolution which aim is the immediate reaction against pathological stimuli ( ) . after the initial recognition of pathogens or tissue damages by the tissue-resident macrophages, which express the pattern recognition receptors (prrs), a variety of different inflammatory mediators are produced by leukocytes, endothelial cells, tissue cells or are derived from plasma proteins. these mediators include different chemokine, cytokine, vasoactive amines and products of the arachidonic acid: their primary effect is to elicit inflammatory response locally and to recruit leukocytes and plasma proteins in the site of injury ( ) . glycoprotein (agp) ( ) , in ruminants is haptoglobin (hp) ( ) ; in horse is serum amyloid a (saa) ( ) and in pigs are hp, saa, and major acute phase protein (map) ( ) . recently, in veterinary medicine, studies on the role of apps as markers of infectious, inflammatory and neoplastic diseases have proliferated ( ) and at least different plasma proteins have been identified as apps ( ) . their use as marker of homeostasis perturbation provides some advantages compared with traditional parameters like the white blood cell (wbc) counts. compared to wbc count, the diagnostic sensitivity of apps is higher and the change in concentration is faster ( ) . moreover, their stability in serum/plasma is high, so it is possible to measure apps in frozen samples ( ) . one limitation of the apps is the poor diagnostic specificity, for this reason they cannot be used as primary diagnostic test for a specific disease, but they were successfully used to detect subclinical diseases and to monitor clinical evolution and to assess the response to treatment ( ) . additionally, the combined measurement of several apps provides more information than the evaluation of a single protein: in the "app value index, " proposed by gruys et al. ( ) , sensitivity and specificity are improved by combining response of both positive and negative major and minor apps. marine mammals are a group of around mammalian species which depend on water environment for most of their needs. they include orders: carnivora (pinnipeds, otters, and polar bear), cetacea (dolphins, whale, and porpoise) and sirenia (manatee and dugogons). marine mammals are differently adapted to life in water, with some species, which are fully aquatic (cetaceans and sirenia), and others that spend part of their life on land (pinniped and polar bears) ( ) . from an immunological point of view, aquatic adaptation caused few differences in distribution and function of immune system between marine and terrestrial mammals ( ) . however, nowadays the marine mammal' immune system is deeply exposed to environmental pollution because they are a long-lived animals placed at the top of food chain, thus they are exposed to a progressive bioaccumulation of fat-soluble pollutants, such as pcbs, which affect both innate and adaptive immune function ( ) . for these reasons, increasing knowledge in cellular and humoral immune response is continuously required to understand their immune system and in particular its relationship with infectious pathologies and the environmental pollution ( ) . furthermore, marine mammals live also in controlled environment like aquaria, rehabilitation facilities and research center where health assessment is fundamental to evaluate the correct management of animals and to monitor the response to therapy during rehabilitation. from this perspective, the availability of new markers to asses immune functions is fundamental both for medical care and research purpose ( ) . innate immune response represent the first line of response against pathological stimuli, it's very fast and it's based primarily on effector cells (e.g., mast cells, macrophages, neutrophils) and antimicrobial substances (e.g., complement, reactive oxygen, and nitrogen species). on the other hand, adaptive immune system is antigen-specific, it takes more days to be effective and it's based on different t-cells response and on b-lymphocytes which are responsible of humoral response mediated by the different subclass of immunoglobulin (igg, igm, iga) ( ) . several assays were proposed to evaluate both immune response in marine mammals, generally based on isolated leucocytes with the aim to evaluate the leukocytes response against in vitro stimuli ( ) . to assess the response pattern of cetaceans' cellular innate immune system, the phagocytosis and the generation of reactive oxygen species of polymorphonuclear leukocytes were investigated. in particular, in vitro ingestion of latex beads and hydrogen peroxide production have been evaluated in beluga whales (delphinapterus leucas) and in bottlenose dolphins (tursiops truncatus) ( , ) whereas phagocytosis and respiratory burst assay, using whole blood from bottlenose dolphins, were used to assess the antimicrobial activity ( ) . in addition, the investigation of apr by analyzing the cytokine expression gives important information on the functionality of lymphoid cells. the production of specie-specific antibodies allows the development of immunological assays for the quantification of cytokine expression useful to investigate the inflammatory response in whales and dolphins. the coding regions of il- , il- β, il- , and tnf-α gene of the beluga whale have been sequenced, and a cytokine-specific rabbit antisera have been produced ( ) ( ) ( ) . in harbor porpoise (phocoena phocoena), the quantification of mrna of il- β, il- , il- , il- , il- , and tnf-α have been performed by rt-pcr ( ) , and the increase of il- was seen in harbor porpoises suffering from long lasting infectious ( ) . also in bottlenose dolphins, pacific white-sided dolphins (lagenorhynchus obliquidens), and beluga whales, il- , il- , il- , il- , il- , il- , tnf-α, tgf-β, and interferon (ifn)-γ quantification was performed using rt-pcr ( ) . an il- receptor expression assay and an il- elisa were developed in bottlenose dolphins and killer whale (orcinus orca), respectively ( , ) . similarly, both innate and the cell-mediated response were studied in pinnipeds. to better understand the innate response, phagocytic activity of isolated peripheral blood leukocytes was evaluated in harbor seal (phoca vitulina), gray seal (halichoerus grypus), and harp seal (phoca groenlandica) pups, in harbor seal female during lactation and in harbor seal pups admitted to rescue center ( , ) . the authors found an age-related variation in both pups and adults: phagocytosis increased with age in gray and harbor seal pups, while in female harbor seals decreased from sub-adult to adulthood. at the same time, pups after rehabilitation showed a decreased phagocytic activity, probably due to the decreased stimulation of innate response after therapy. also cytokine response was evaluate in harbor seal. pro-inflammatory cytokine mrna (il- β, il- , il- , and il- ) in pups in a rehabilitation center were higher at admission whilst il- was higher before the release ( ), demonstrating the recovery from inflammation. recently, a multiplex canine cytokine assay was validate in harbor, gray and harp seal to measure proteins levels in cell culture supernatant of peripheral blood mononuclear cells (pbmc) ( ) . however, all these techniques are not generally applicable in a clinical setting in which the primary goal is a sensitive diagnostic tool with a rapid turnaround, even if give us important information on factors affecting cetaceans' immune system. for this reason, in the past years, several efforts were made to identify laboratory markers of disease in these animals. first parameters tested were wbc and erythrocyte sedimentation rate ( ) . however, even if they are inexpensive and rapid, they lack specificity and sensitivity. moreover, changes in wbc occur after several hours after inflammatory stimuli. thus, efforts were directed to identify inflammation at earlier stage ( ) . to examine the humoral response, species-specific antibodies against igg were produced and used to evaluate serum igg levels in killer whale by radial immunodiffusion assay ( ) and by competitive elisa in bottlenose dolphins ( , ) . the determination of igg baseline values in free-ranging and in managed dolphins revealed higher levels of immunoglobulin in the first group with several values over the accurate range of the assay, probably due to the higher parasitic load in free-ranging dolphins ( ) . serum total protein, albumin, globulin and albumin:globulin ratio (a:g) are undoubtedly among the most measured markers in basic health assessment in domestic animals as well as in marine mammals. serum protein electrophoresis is also broadly applied in veterinary medicine and it has the advantage to produce an accurate measurement of albumin and the visualization of globulin fractions ( ) . the interpretation of total proteins values and electrophoretic pattern of serum proteins is receiving increased attention also in marine mammals in which a typical pathologic pattern could be identified in inflammatory diseases ( ) . reference intervals for these markers are available for free-ranging bottlenose dolphins ( ) and, compared to these, recently data on managed dolphins showed slightly lower values of tp, α-globulins, and γ-globulins and higher albumin and albumin/globulins ratio ( ) . it's interesting to note that hp, α -antitrypsin, α antichymotripsin, and α -macroglobulin migrate in the α-globulins fraction, while the igg and crp migrate in the γ-globulins fraction. albumin acts as a negative acute phase protein since the synthesis of this protein is decreased during an inflammation ( ) . thus, the lower concentration of positive apps associated to a higher concentration of albumin and the consequent higher albumin/globulins ratio could reflect lower antigenic stimuli in managed population compared to the free-ranging populations ( ) . serum total protein analysis were used to assess health status in several cetaceans species such as pantropical spotted dolphins (stenella attenuata) ( ), beluga ( ) , minke whales (balaenoptera acutorostrata) ( ) and killer whales ( ) as well as in other marine mammals, like harbor seals (phoca vitulina) ( ) and walruses (odobenus rosmarus) ( ) . in all these species, serum total protein analysis was demonstrated to be one of the most used and commonly accepted marker of inflammation. however, specific apps have demonstrated their superior role as early markers of inflammation, so based on the results obtained in humans and companion animals, several positive apps were tested in marine mammals ( table ) . published works had the primary aims to evaluate the feasibility of the assays to measure the apps, to validate the antibody-based assay and to determine the ris. in bottlenose dolphins three apps (crp, saa, and hp) were tested, even if not always complete validation studies were performed ( , ) . for these apps, the authors established the ris in free-ranging and managed dolphins using automated assays ( ) and they found significantly lower saa and higher hp levels in free ranging animals. the only clinical significance of these alteration was a higher ability to detect chronic inflammation for hp. regarding hp, segawa and colleagues validated commercially available hp-elisa and hp-hemoglobin binding assay in bottlenose dolphins with ''acceptable" intra-and inter-assay imprecision (cv: . % healthy dolphins and . % inflamed dolphins; cv: . % healthy dolphins and . % inflamed dolphins) and demonstrated that hp levels in the serum increase under inflammatory conditions ( ) . positive apps were tested also in florida manatees (trichechus manatus latirostris) to define the more accurate marker of inflammation. five different apps were tested: agp, crp, hp, fibrinogen, and saa. saa showed the highest diagnostic sensitivity and specificity ( % for both sensitivity and specificity) in the detection of inflammatory diseases, the diagnostic specificity of hp and fibrinogen were and %, respectively, while their diagnostic sensitivity were and %, respectively, ( ) . when used in stranded manatee suffering from cold stress and trauma, saa showed % of sensitivity and % of specificity in detecting diseased animals ( ) . by contrast, the abs used for the determination of agp and crp did not cross-react in this species ( ) . in harbor seal, an ab anti-crp and a competitive immunoassay was produced ( ), but hp is probably the app most used in pinnipeds. a multispecies assay based on hemoglobin binding capacity was used to demonstrate as hp is a sensitive marker of the health vs. disease status in harbor seal ( ) . in seal pups admitted in a rescue center. hp, total protein, igg and globulin values correlated positively, but hp levels increased during the hospitalization, probably reflecting age-related changes ( ) . hp is considered a health marker also in steller sea lions (eumetopias jubatus): significantly higher levels of hp were found in declining population compared to more stable ones ( ) . however, also genetic differences between distant and isolated population of wild animals could be the causes of this difference, not only a pathological condition. if some data on marine mammals positive apps are available in literature, quite surprising are the lack of data available on negative apps. for these reasons, the possibility to evaluate the usefulness of an "app value index" is still far from being applied. the availability of sensitive markers of inflammation both for free-ranging and managed marine mammals is nowadays considered fundamental to evaluate the health status and, in rehabilitation setting, to monitor the response to therapy and to define the prognosis. as serum markers, the apps have several advantages: they have longer stability compared to other blood component such as wbc; they can be performed on frozen serum, thus the samples can be shipped to references laboratories; some assays can be automated to obtain results in an excellent turnaround time. however, is important to consider that the knowledge about apps in one species cannot be readily generalized to another species, in which healthy levels, response to inflammation or infection, and prognostic significance may be different ( ) . moreover, the evolution of marine mammals and their adaptation throughout the millennia to an aquatic environment had led to a different physiology and metabolism compared to terrestrial mammals. thus, the understanding of the genetic, phenotypical and biochemical properties of marine mammals apps are essential prior to using them as a new biomarker. an example of how the biochemical properties influence the analytic method is paroxonase- (pon ), a hdl-bound esterase which protects against organophosphate compounds, acts as negative app and as oxidative stress marker. pon is usually assessed by enzymatic method and, based on the different pon functions, several substrates have been identified to evaluate serum pon activities. nevertheless, both in humans as in some terrestrial mammals, pon gene polymorphisms highly influence the enzymatic activity toward different substrate: the single-nucleotide polymorphisms (snps) leu met and gln arg increase the paraoxonase activity ( ) in humans and different pon genotypes influence activities toward paraoxon and phenyl-acetate in rabbit ( ) . also in cows, some snps in the promotor region of pon gene are associated with serum pon activity ( ) . recently, a phylogenetic study on convergent functional losses across marine mammals, has identified a pon functional loss in marine mammals, probably related to their different lipid metabolism and fatty acid oxidation due to adaptation to the marine environment and a high concentration of ω- fatty acids on their diet. as a consequence, in several marine mammals species paroxonase activity is very low, while enzymatic activity against other pon substrates is still present, such as arylesterase activity ( ) . for these reasons, the use of classical enzymatic assays is hampered in these animals and further studies are needed to elucidate the role of pon as possible negative app, oxidative stress marker and the consequences of its inability to detoxify organophosphates compounds. from an analytical point of view, another challenge in the evaluation of apps in marine mammals is the need of speciesspecific assays, especially for the immunological assay, such as elisa or immunoturbidimetry. this means the development of a de-novo method, often a time-consuming and expensive approach, or the validation of a commercial available assay used in other species ( ) . the latter approach is surely the most used in veterinary medicine, in which some human assays were validated for dogs, cats and horses ( ) . however, even if some apps appeared highly conserved among species, an accurate validation of antibody cross reactivity is needed as well as species specific standards and control material ( ) . among positive apps, saa is the most used across different species: it appeared as the most conserved app in mammals even if some difference in circulating isoforms were reported ( ) and it's considered a major app in all the mammals in which it was investigated ( ) . some commercial saa assays showed good results also in marine mammals, such as bottlenose dolphin, manatee and striped dolphin (stenella ceoreloualba) ( , , ) and its use as diagnostic and prognostic marker appears nowadays the most promising. to obtain accurate data, all the pre-analytical factors that could influence the results should be taken into consideration. the effect of storage, temperature and different anticoagulant had to be evaluated in a correct validation process as well as the interference of hemolysis and lipemia, as done in other species ( ) . the application of a novel biomarker required a full evaluation of all the analytical performances and the clinical value. this process is usually divided in steps: the assessment of analytical features (precision, accuracy, detection limits), the overlap performance (the ability to detect difference between healthy and diseased animals), the assessment of diagnostic capacity (sensitivity, specificity, accuracy, positive, and negative predictive values) and, at the end, the evaluation of the outcome of the new methods (which is the advantage of the test and its influence in the patient management) ( ) . in veterinary medicine, the validation studies do not always follow all these steps, mainly due to the lack of resources or technical limitation ( , ) . also in marine mammals, the majority of studies had performed only some steps ( , , , , ) . this is mainly due to the limitation in species-specific reagents, the number of samples from animal with known health status and, last but not least, the capability to generate appropriate reference intervals, hampered the possibility to perform complete validation studies. population-based reference intervals derived from an appropriate group of reference individuals are usually required for diagnostic purpose ( ) . however, a number of biological factors have to be taken in consideration to select the appropriate reference population. surely, age, sex and pregnancy could be used for partitioning ( , ) , but in marine mammals greater attention should be given to the difference between wild and managed animals. serum protein electrophoresis values obtained in managed bottlenose dolphins showed lower total proteins and higher albumin levels compared to reference intervals derived from free-ranging ( ) while wild-caught manatees, apparently clinically healthy, had saa level above reference limit ( ) . these data could indicate a trend to an inflammatory status or the presence of subclinical inflammation in free-ranging animals which are more exposed to immunological stimuli. in any case, this highlights the need to define appropriate reference intervals for animals living in different environment to have an accurate toll for the evaluation of clinical condition. compared to human and companion animals, the use of apps in marine mammals is just getting started. the increasing need of knowledge on immune system and its response against infectious diseases or chemical pollutants and the request of more sensitive inflammation markers have increased the effort of researchers to study the apr and apps. even if apps are considered a sensitive, but non-specific marker of inflammation, some studies revealed that, in some infectious diseases, apps showed a specific behavior and biochemical features. one example is the modification of the glycan moiety of agp in feline infectious peritonitis, fiv and felv, influencing the host-pathogens interaction and the immune response ( ) ( ) ( ) . currently, some of the greatest threats for wild marine mammals is pathogens, like morbillivirus, herpesvirus, brucella ceti, and toxoplasma gondii ( ): the evaluation of apr and apps patterns during these infectious diseases could lead to the identification of a distinctive response of the immune system and increase the understanding of hostpathogen interaction. secondly, for managed or rescued animals, the forthcoming needs are the increase of automated assays, the standardization of procedures across laboratories and the discovery of new markers, for example negative apps, to generate an app index also in marine mammals. these new tools will certainly increase the diagnostic and prognostic skills for health assessment and, especially for stranded animals, the development of new "health status" markers will provide valuable resources in evaluating the response to treatment and rehabilitation prior to the release into the wild. mg and fb analyzed the literature review, designed, and wrote the review. monitoring health by values of acute phase proteins inflammation and factors that may regulate inflammatory response origin and physiological roles of inflammation inflammatory mechanisms: the molecular basis of inflammation and disease acute phase proteins in dogs and cats: current knowledge and future perspectives initiation of acute phase response and synthesis of cytokines acute phase proteins: biomarkers of infection and inflammation in veterinary medicine current research on acute phase proteins in veterinary diagnosis: an overview the feline acute phase reaction application of acute phase protein measurements in veterinary clinical chemistry acute phase proteins as promising biomarkers: perspectives and limitations for human and veterinary medicine applications of acute phase reactants in infectious diseases an immunoturbidimetric assay for canine c-reactive protein bovine acute phase response following turpentine injection serum amyloid a protein (saa) in horses: objective measurement of the acute phase response the final hurdles for acute phase protein analysis in small animal practice study on biological variability of five acute-phase reactants in dogs haptoglobin and ceruloplasmin as determinants of inflammation in dogs chapter -introduction marine mammal immunology immunotoxic effects of environmental pollutants in marine mammals cetacean host-pathogen interaction(s): critical knowledge gaps immunology of whales and dolphins evaluation of the polymorphonuclear cell functions of bottlenose dolphins immune functions in beluga whales (delphinapterus leucas): evaluation of mitogen-induced blastic transformation of lymphocytes from peripheral blood, spleen and thymus simultaneous measurement of phagocytosis and respiratory burst of leukocytes in whole blood from bottlenose dolphins (tursiops truncatus) utilizing flow cytometry molecular cloning and characterization of beluga whale (delphinapterus leucas) interleukin- beta and tumor necrosis factoralpha molecular cloning, phylogenetic analysis and expression of beluga whale (delphinapterus leucas) interleukin molecular cloning and phylogenetic analysis of beluga whale (delphinapterus leucas) and grey seal (halichoerus grypus) interleukin development of a lymphocyte-transformation-assay for peripheral blood lymphocytes of the harbor porpoise and detection of cytokines using the reverse-transcription polymerase chain reaction increased blood interleukin- mrna levels in diseased free-ranging harbor porpoises (phocoena phocoena) quantitation of leukocyte gene expression in cetaceans development of an interleukin- receptor expression assay and its use in evaluation of cellular immune responses in bottlenose dolphin (tursiops truncatus) expression and functional characterization of killer whale (orcinus orca) interleukin- (il- ) and development of a competitive immunoassay immune status and function in harbor seal pups during the course of rehabilitation phagocytosis in pup and adult harbour, grey and harp seals cytokine and acute phase protein expression in blood samples of harbour seal pups validation of a commercial canine assay kit to measure pinniped cytokines clinical pathology crc handbook of marine mammal medicine measurement of serum immunoglobulin concentration in killer whales and sea otters by radial immunodiffusion development and validation of monoclonal and polyclonal antibodies for the detection of immunoglobulin g of bottlenose dolphins (tursiops truncatus) baseline circulating immunoglobulin g levels in managed collection and freeranging bottlenose dolphins (tursiops truncatus) acute phase response in animals: a review hematologic and serum biochemical reference intervals for freeranging common bottlenose dolphins (tursiops truncatus) and variation in the distributions of clinicopathologic values related to geographic sampling site detection of hereditary bisalbuminemia in bottlenose dolphins (tursiops truncatus fundamentals of veterinary clinical pathology hematological, serum, and plasma chemical constituents in pantropical spotted dolphins (stenella attenuata) following chase, encirclement, and tagging serum chemistry of freeranging white whales (delphinapterus leucas) in svalbard serum chemistry of the minke whale from the northeastern atlantic hematological and serum biochemical analytes reflect physiological challenges during gestation and lactation in killer whales (orcinus orca) hematology and serum chemistry in stranded and wild-caught harbor seals in central california: reference intervals, predictors of survival, and parameters affecting blood variables serum chemistry reference values in free-ranging north atlantic male walruses (odobenus rosmarus rosmarus) from the svalbard archipelago acute phase protein quantitation in serum samples from healthy atlantic bottlenose dolphins (tursiops truncatus) evaluation of immune and stress status in harbour porpoises (phocoena phocoena): can hormones and mrna expression levels serve as indicators to assess stress? comparison of methods used to diagnose generalized inflammatory disease in manatees (trichechus manatus latirostris) assessement of serum amyloid a levels in the rehabilitation setting in the florida manatee (trichechus manatus latirostris) plasma haptoglobin levels in threatened alaskan pinniped populations characterization of haptoglobin in the blood plasma of harbor seals (phoca vitulina) fibrinogen concentrations in captive bottlenose dolphins during pregnancy characterization of the circulating serum amyloid a in bottlenose dolphins molecular characterization and validation of commercially available methods for haptoglobin measurement in bottlenose dolphin harbor seal (phoca vitulina) c-reactive protein (c-rp): purification, characterization of specific monoclonal antibodies and development of an immuno-assay to measure serum c-rp concentrations acute phase protein haptoglobin in blood plasma samples of harbour seals (phoca vitulina) of the wadden sea and of the isle helgoland assay validation and diagnostic applications of major acute-phase protein testing in companion animals human paraoxonase- (pon ): gene structure and expression, promiscuous activities and multiple physiological roles rabbits possess a serum paraoxonase polymorphism similar to the human q r characterization of single nucleotide polymorphisms in the promoter region of the bovine paraoxonase (pon ) gene affecting serum enzyme activity in dairy cows ancient convergent losses of paraoxonase yield potential risks for modern marine mammals clinico-pathological findings in a striped dolphin (stenella coeruleoalba) affected by rhabdomyolysis and myoglobinuric nephrosis effects of age and sex on clinicopathologic reference ranges in a healthy managed atlantic bottlenose dolphin population association between faecal shedding of feline coronavirus and serum alpha -acid glycoprotein sialylation hyposialylated α -acid glycoprotein inhibits phagocytosis of feline neutrophils glycan moiety modifications of feline alpha -acid glycoprotein in retrovirus (fiv, felv) affected cats emerging infectious diseases in cetaceans worldwide and the possible role of environmental stressors clinico-pathological investigation of serum proteins in odontocetes the literature review presented in the manuscript is partially included in ph.d. thesis of bonsembiante ( ), supported by a ph.d. grant from the university of padova. key: cord- -r m lij authors: sanchez-rodriguez, dolores; annweiler, c.; gillain, s.; vellas, b. title: implementation of the integrated care of older people (icope) app in primary care: new technologies in geriatric care during quarantine of covid- and beyond date: - - journal: j frailty aging doi: . /jfa. . sha: doc_id: cord_uid: r m lij nan the covid- pandemic due to a novel coronavirus (sars-cov- ) in december has rapidly spread worldwide. the mortality rate is about . % in general population, with high human-to-human transmission of . (credible interval [ . , . ]), and nasopharyngeal asymptomatic carriers act as vectors within the population ( ). the world health organization (who) declared the pandemic on march , and established objectives and action plan. first, who aimed at limiting the transmission of sars-cov- , which required large isolation actions (country borders lockdown and individual quarantine). second, who aimed at guiding and supporting the different health care systems across countries. finally, developing therapeutic interventions appeared as a global priority as available evidence were still scarce ( ). more than clinical trials are ongoing worldwide. despite all age groups are similarly affected by covid- , older adults experience a higher risk of developing severe complications, with mortality raising up to % in patients - -year-old, and up to . % in those aged ≥ years. europe is particularly concerned by this problem due to the aging of the population. in , the european union counted million people aged years or more, compared to only million children aged years or less (https://ec.europa. eu/eurostat/statistics-explained/index.php). moreover, it is noticeable that the isolation measures have reduced the availability of the primary care givers and led to delayed/ cancelled outpatients' clinics, which increases the risk of acute decompensation of chronic diseases and frailty process in shortand mid-terms, independently of the covid- infection. in this context, implementing new technologies based on quantified-self and on-line resources may help optimizing older adults' health. the integrated care of older people (icope) project is an international program ( , ), based on the measurement of "intrinsic capacity", a new concept defined by the world health organization plan of action - , as "the composite of all the physical, functional, and mental capacities of an individual" ( - ) intrinsic capacity changes the focus from a negative paradigm of aging (diseases, disability) towards a positive focus related to "optimal aging" ( , ). intrinsic capacity is focused on disease prevention, it has been related to the ability of individuals to overcome stressful or acute events ( , ) . intrinsic capacity has been recently associated with the onset of autonomy decline, falls, and death during a -year follow-up ( ) . in a second step in the long-term, the icope project is intended to collect large population data for the development of "nomograms" of optimal aging in older adults. several actions to implementation of the icope program in the community have been scheduled by who ( ), and one is the icope app, a free-of-charge app (android/apple store https://www.youtube.com/watch?v=glva rev ka. the first version of the icope app is available and allows a selfadministered screening of individuals' intrinsic capacity. new functions of the app are developed and include comprehensive geriatric assessment, sharing on-line information, and receive individual guide and support from primary care. the implementation of the icope app would mitigate the current gap between geriatricians and family physicians and those between the family physicians and the patients. moreover, the decline in intrinsic capacity measured with the app may be useful to anticipate clinical adverse outcomes, such as prefrailty, missteps, falls, disability (i.e., a -point barthel score), institutionalization, decreased quality of life, and death) and to share professional advices to community-dwelling older adults. some research teams are already following who guidelines in the implementation of the icope app ( , ), which needs: ) partnership with primary care physicians to share the app and ensure the feasibility of the app in each environment (e.g. by the technology, economics, legal, operational, scheduling -telos-model). ) evaluation of the association between the components of intrinsic capacity at baseline and the incidence of adverse health consequences in prospective cohort studies. it would be interesting to evaluate the relationship among intrinsic capacity at baseline and incidence of frailty, prefrailty, falls, disability, and decreased quality of life, hospitalization, institutionalization and particularly, if the score at baseline measured by the app is associated with all-cause mortality including mortality related to covid- . ) assessment of the impact of the icope project and app on community-dwelling older adults and revisiting strategies towards geriatricians and primary care physicians. the results expected from the implementation of the icope app would be i) to support primary care, and indirectly create a long-term clinical and research partnership; ii) for the older patients, to better identify those at risk of developing frailty and adverse health consequences, and to be able to make reasoned decisions in clinical routine ; and iii) for the health system, to invest in tools likely to prevent decline in intrinsic capacity and maintain function late in life. the icope project has a number of strengths. the first one is that the project is timely, as there is an increasing, crucial need of providing support to primary care physicians and community-dwelling older patients; moreover, on-line health resources had raised in interest in the latest years ( ) and now even more due to the covid- pandemic. the implementation of the icope project involves economic advantages because no additional materials are required to deliver support, with the exception of the human cost of the healthcare providers engaged. in addition, the icope app is free-of-charge, suitable for all brands (android/phone) and devices (telephones, computers, etc.). of course, online communication and technical devices required by the icope app could be challenging for older adults. however, this limitation could be counteracted by taking the window of opportunity of the quarantine, which makes older people and their families more likely to get engaged in online communication. the new-acquired technical skills by the users are effective in controlling the environment e.g. receiving advice and professional support, therefore, the behavior is reinforced ( ) . finally, it is noticeable that using the icope app may be slightly modified during the covid- pandemic compared to the original aim, which was to measure individuals' intrinsic capacity in a non-pandemic "normal" situation. this is also a great opportunity to unleash the potential of academic discoveries ( ) . part of the inspire program ( - ), the apps icope-monitor is also free available both with android and iphone, and give the possibility to monitor intrinsic capacities overtime. connected to a secure medical database, the icope step is performed every - months by professionals or seniors themselves. if a deterioration in one or more domains of intrinsic capacity is identified, an alert is generated by an algorithm which allows the health professionals to intervene quickly ( ) . in conclusion, the icope app applies to the "actionresearch philosophy" ( ) to bridge the gap between research and clinical practice, and to provide better care for older community-dwellers, especially during this exceptional and dramatic period of pandemic and containment, after which the world of tomorrow will no longer look like the one before. clinical course and risk factors for mortality of adult inpatients with covid- in wuhan, china: a retrospective cohort study actions required to implement integrated care for older people in the community using the world health organization's icope approach: a global delphi consensus study redesigning care for older people to preserve physical and mental capacity: who guidelines on community-level interventions in integrated care the world report on ageing and health: a policy framework for healthy ageing prediction of adverse outcomes in nursing home residents according to intrinsic capacity proposed by the world health organization. newman a, editor evidence for the domains supporting the construct of intrinsic capacity the international registry of patients with sarcopenia : applying research in sarcopenia to clinical practice i control therefore i do: judgments of agency influence action selection the inspire research initiative: a program for geroscience and healthy aging research going from animal models to humans and the healthcare system original research integrated care for older people and the implementation in the inspire study. j prev alzheimer's dis framework implementation of the inspire icope-care program in collaboration with the world health organization (who) the 'action-research' philosophy: from bedside to bench, to bedside again authors' contribution: dsr and ca conceived the manuscript; dsr, ca, and sg wrote the manuscript; dsr and sg did literature review; bv corrected the manuscript. all co-authors read and approved the final version of the manuscript. key: cord- -gtdmqi l authors: lehman, blair; gu, lin; zhao, jing; tsuprun, eugene; kurzum, christopher; schiano, michael; liu, yulin; tanner jackson, g. title: use of adaptive feedback in an app for english language spontaneous speech date: - - journal: artificial intelligence in education doi: . / - - - - _ sha: doc_id: cord_uid: gtdmqi l language learning apps have become increasingly popular. however, most of these apps target the first stages of learning a new language and are limited in the type of feedback that can be provided to users’ spontaneous spoken responses. the english language artificial intelligence (elai) app was developed to address this gap by providing users with a variety of prompts for spontaneous speech and adaptive, targeted feedback based on the automatic evaluation of spoken responses. feedback in the elai app was presented across multiple pages such that users could choose the amount and depth of feedback that they wanted to receive. the present work evaluates how english language learners interacted with the app. we focused on participants’ use of the feedback pages and whether or not performance on spontaneous speech improved over the course of using the app. the findings revealed that users were most likely to access the most shallow feedback page, but use of the feedback pages differed based on the total number of sessions that users completed with the app. users showed improvement in their response performance over the course of using the app, which suggests that the design of repeated practice and adaptive, targeted feedback in the elai app is promising. patterns of feedback page use are discussed further as well as potential design modifications that could increase the use of feedback and maximize improvement in english language spontaneous speech. language learning has moved from the traditional classroom-only model to computerassisted language learning to mobile-assisted language learning (mall) [ , ] . mall apps provide users with flexibility, autonomy, and personalized learning experiences [ ] . there are currently over language learning apps in the ios app store, and apps have even expanded to smart watches that incorporate exercise into language learning [ ] . mall apps have shown to be an effective method [ , ] . duolingo, for example, claims to be as effective as college-level language courses [ ] , but others report more mixed findings [ ] . in this abundance of mall apps, many have a similar focus in that they target (a) general language learning and (b) learners at an initially low proficiency level. thus, there is still a need for the development of apps that provide support to learners at other proficiency levels and with differing goals. for example, recent efforts in mall app development have focused on the particular language needs of migrants and refugees [ , ] and low literacy adults [ ] . one of the main advantages of mall apps (and attractions to users) is that they provide immediate, targeted feedback about users' performance on learning activities. this is consistent with years of research that has shown that simply providing feedback is not enough, it must be delivered in a way that is optimally useful for learners [ , ] . mall apps are typically able to provide targeted feedback on the quality of selectedresponse items, grammar and spelling for written responses, and word pronunciation for constrained speaking tasks. however, many mall apps are limited in the level of detail that can be provided for feedback on speaking tasks [ , ] . duolingo, for example, identifies whether or not a user has correctly pronounced a word, but it does not provide feedback about how the user could more accurately pronounce the target word. given that speaking is often one of the more challenging aspects of learning a language [ ] [ ] [ ] , it is important for mall apps to provide feedback in such a way that users feel confident that they can improve their speaking skills. given the challenges of providing targeted feedback in real time for speaking tasks, most mall apps focus only on constrained speaking tasks in which users are provided with a text to read aloud verbatim because automated feedback can be more easily provided. however, our recent user interviews suggested that many language learners would like to practice and receive feedback for spontaneous speaking tasks. spontaneous speaking tasks involve learners responding to an open-ended prompt (e.g., tell me about your favorite vacation.). this type of task is utilized on many standardized assessments of language skills (e.g., toefl ® , ielts™) as it shows an advanced level of speaking proficiency and spontaneous speaking skills are viewed as an important aspect of effective communication [ , ] . this type of task is often not included in mall apps because it is difficult to provide immediate, targeted feedback. spontaneous speaking tasks are typically evaluated by human raters in standardized assessments, which limits the ability to provide feedback to users immediately after responding. to address the apparent lack of spontaneous speaking practice with immediate feedback for language learners, we have developed the english language artificial intelligence (elai) app. the elai app was designed to provide users with an opportunity to practice spontaneous speech and receive detailed feedback about the quality of their responses. this learning model is consistent with languaging [ ] [ ] [ ] as students are asked to engage in effortful language production that can draw attention to their current weaknesses, but with the added benefit of targeted feedback to help focus efforts for improvement. we utilized an automated speech analysis tool that evaluates spontaneous speech on delivery, language use, and topic development to provide targeted, detailed feedback. however, it is not enough to simply provide feedback [ , ] . a recent review of research on oral feedback for spoken responses, for example, found that there is a limited understanding of how learners make use of feedback [ ] . it is then important that we understand how users interact with the feedback provided. the present work is the first evaluation of the elai app and was guided by three research questions: ( ) how do users interact with the app features?, ( ) what do users do after viewing feedback?, and ( ) does users' performance improve during app use? we investigated these three research questions with native mandarin speakers who are learning english for the purpose of attending university in an english-speaking country. the elai app was developed to provide an easily accessible resource for english language learners at an intermediate or advanced level, with the goal of attending university in an english-speaking country, to practice spontaneous speech and receive feedback. the development was guided by interviews with potential users from the target audience, which revealed that users were often practicing spontaneous speech on their mobile phone but were unable to receive feedback in the same medium [ ] . users were most interested in feedback that corresponded to standardized english language assessment evaluations. users also revealed their desire for access to sample responses to compare to their own responses both in terms of delivery and content. the elai app was then developed to address the needs of these real-world users. users began with the elai app by browsing the many prompt options available. figure shows (from left to right) screenshots of the app splash page as well as the process of selecting a prompt category, responding to a specific prompt (e.g., do you think the use of smart watches will increase or decrease in the future? why?), and the feedback overview. after completing a new response, users were notified when feedback was available (latency was equivalent to the response length). user responses were evaluated with an automated speech analysis tool that used acoustic and language models to allow for the extraction of acoustic characteristics and creation of a response transcript. the models were based on nonnative english speakers to account for pronunciation differences due to accents. the automated speech analysis tool then evaluated the response on over raw speech features from the acoustic characteristics and transcript. a subset of these features was selected based on their potential for learning feedback and were then combined to provide feedback on six key speech features (filler words, pauses, repeated words, speaking rate, stressed words, vocabulary diversity) to help users improve speaking skills. users could access feedback on four pages within the app, which allowed for self-selection of the type and amount of feedback provided. the first feedback page was my history (fig. , rightmost panel), which provided a feedback overview for each response at a relatively shallow level in that it only identified two speech features that needed improvement (weightlifter icon) and one feature that was done well (thumbs up icon). this was the first instance of feedback that was adaptive to the individual user. for example, in fig. the user needed to improve on repeated words and vocabulary diversity, whereas filler words was done well in the technology response (top card). needs work was defined separately for each speech feature with some defined as overuse (filler words, pauses, repeated words, vocabulary diversity), whereas other features had an inverted u-shaped relationship in which too much or too little was problematic (speaking rate, stressed words). however, users were not provided with any explanations or resources to improve future responses on overview. thus, overview provided minimal feedback on the quality of a response and minimal support for improving future responses but did serve as an organized resource for users to access all of the feedback they had received. figure shows the next feedback page that users could access by selecting a specific response card on overview or directly through the feedback ready notification. this next page was feedback summary report and was designed to be the main source of feedback for users. on summary report users could listen to their own response, review explanations for those three speech features that were shown on overview, access additional ideas for how to develop a response to that prompt, and listen to sample responses from both native and nonnative english speakers (from left to right in fig. ). the design of summary report allowed the user to quickly develop an understanding of the quality of their response by focusing on two features that needed improvement and ensured that this feedback was actionable by providing users with additional information and resources to improve their future responses. users could view more detailed feedback on feedback full report and feedback details (see fig. ). the full report provided explanations for all six speech features. for example, in fig. the two leftmost panels show that the user did well on filler words but needed to improve on repeated words and speaking rate. details provided even more detailed information about four of the six speech features (pauses, repeated words, filler words, vocabulary diversity). details provided a transcript of the response (see second from the right panel in fig. ) , which highlighted the problematic aspects of the speech feature (e.g., repeated words). details for vocabulary diversity provided suggestions of additional words that could be used to respond to the prompt (see rightmost panel in fig. ) . full report and details provided users with a greater amount of and more indepth feedback, which can be beneficial if users dedicate the time and effort needed to process and apply the information provided [ ] . users were also able to view information about their app use metrics through the me screen. users could see the total amount of time they had recorded responses, total number of responses, the amount of time for recorded responses in the current week, and how many days in a row they had recorded responses. the me screen also allowed users to view badges that they earned. users could earn a variety of badges that targeted engagement and performance. engagement-based badges were designed to encourage persistence and regular practice (e.g., multi-day streaks of recording), whereas performancebased badges allowed users to track their progress over time on a single speech feature (e.g., received "good job" on filler words three times in a row). participants were students from an english language learning program in china that primarily focused on preparation for standardized english language learning assessments. gender information was obtained from participants: % female, % male, and % preferred not to respond. participants completed from to sessions with the elai app over a one-month period (m = . , sd = . ). sessions were a little over five minutes on average (sd = . ) and included an average of . user-initiated actions (sd = . ). users completed an average of . spoken responses over the course of using the elai app (sd = . ). participants were recruited through their english language learning program. those participants who were interested then completed an informed consent and were provided with the information needed to access the elai app. participants were free to use the app as they wanted for one month. there were no direct instructions about how users should interact with the app; however, participants were told that they would receive a certificate of participation if they recorded at least five spoken responses. first, we investigated the use of app features in four ways (see table ): feature access (proportion of participants), average feature time use (in seconds, avg time per access), proportion of total session time (proportion of time), and proportion of total session actions (proportion of actions) [ ] . the proportion of participants that accessed each feature at least once revealed a generally high rate of feature access, with the exception that % or less of users accessed the more in depth feedback pages (full report, details) and listened to their own or samples responses, which were features that users specifically requested. this contradiction between what users say they want and how they interact with a mall app has been found in other apps as well [ ] . overall the feature use analyses revealed that users spent the majority of their time interacting with the elai app browsing for a prompt, responding to prompts, viewing the shallowest level of feedback, and viewing their overall app usage data. this pattern is both consistent and inconsistent with user requests. users were frequently practicing their spontaneous speech, but they were not typically utilizing the more detailed feedback and learning resources that they requested. it is important to note, however, that the more detailed feedback and learning resources were embedded in the app, meaning that users could only access them via another feedback page. feedback overview and summary report, on the other hand, could be accessed directly. thus, the lack of access to the more detailed feedback (full report, details) could represent a lack of user interest or lack of feature awareness. in an effort to consider this dependence between actions, we repeated the proportion of actions analysis with instances in which less detailed feedback page views were removed if they immediately preceded a more detailed feedback page view. this was an overly conservative analysis as it assumed that all of these less detailed feedback page views were only in service of accessing more detailed feedback. the pattern of findings remained the same, which suggests that although we cannot target the exact reason for infrequent access of more detailed feedback, we can feel confident that those pages were accessed less frequently. the previous analyses considered the sample as a whole; however, there was a wide range in the degree to which users engaged with the elai app ( to sessions), which suggests a potential for different use patterns. users were divided into low (five or less sessions, n = ) and high engagement groups (more than five sessions, n = ) based on a median split to explore potential feature use differences. table shows the descriptive statistics for each engagement group. particularly large differences can be seen for accessing more detailed feedback and learning resources, with high engagement users accessing those features at least once at a higher rate than low engagement users. the two engagement groups were compared with independent samples t-tests for average time spent on each feature, which revealed that the high engagement group spent more time on all features, except summary report, full report, and me screen. despite this difference in time spent on features, there were no differences in how users in each engagement group distributed their time (proportion of time, p's > . ) and actions within a session (proportion of actions, p's > . ). the comparison of engagement groups revealed that users who had greater engagement with the elai app accessed more features and spent more time on those features, particularly those features that provided more in-depth feedback and support for improving future performance. the previous findings led us to question if there were particular patterns of behavior after viewing feedback that were indicative of more or less productive behavior. for example, a productive behavior after viewing fb overview would be to access fb summary report to better understand why certain speech features need improvement and access resources for improvement. thus, we investigated the next action taken after viewing each type of feedback. we combined several actions into action categories that consisted of browse behavior (view category, view prompt), feedback viewed, me screen viewed (me screen, view badge), and exit app for these analyses. repeated measures anovas compared the prevalence of post-feedback actions for each feedback page (see table ) and were significant [overview: f( , ) = . , p < . , mse = . , partial η = . ; summary report: f( , ) = , p < . , mse = . , partial η = . ; full report: f( , ) = , p < . , mse = . , partial η = . ; details: f( , ) = , p < . , mse = . , partial η = . ]. bonferroni corrections were applied to all post hoc comparisons. the pattern for overview revealed that all action categories were more likely to occur after viewing overview than exiting the app. a different pattern emerged for the remaining feedback pages. specifically, viewing feedback was the most likely action to occur after viewing summary report, full report, and details, with at least % of next actions involving viewing one of the feedback pages. exit app and browse behavior were the next most likely to occur and me screen viewed was the least likely action to occur after viewing those three feedback pages. these findings suggest that if users can go deeper into the feedback than overview, they may get into a potentially beneficial feedback loop. last, we investigated changes in spoken response performance over the course of app interaction. user sessions (visit to app) were divided into thirds (first, middle, last) and we investigated changes in performance from the first third to the last third. performance was measured as the proportion of spoken responses that received "needs work" feedback on each speech feature in each third of sessions. this investigation reduced the number of users to as users were required to have at least three sessions and to have at least one speech in both the first and last third of sessions. all users included in this analysis were in the high engagement group, which means that they made greater use of the app features, in particular the more detailed feedback and resources for response improvement. table shows the descriptive statistics and paired samples t-test comparisons for each of the six speech features. the comparisons revealed a reduction in the proportion of speech features that needed work, which suggests an overall improvement in performance across use of the elai app. the effect size differences between the first and last third of sessions were all large (d > . ) [ ] , with the exception of a medium effect size (. < d < . ) for pauses. these findings are very promising as they show large improvements in a variety of speech features over a relatively short period of time. however, the findings should be interpreted with a modicum of caution as a small number of participants were included in these analyses ( % of sample), time on task varied across participants, and we were not able to consider additional resources that users may have accessed during this same time period (e.g., language courses, other mall apps). it is also important to note that this investigation was limited to performance within the app and a more formal investigation of changes in speaking skills is needed (e.g., pre/posttest design) to determine the true effectiveness of the elai app as a learning tool [ ] . there is currently a plethora of language learning apps available to users. however, these apps are often designed for beginning language learners and are limited in their ability to provide feedback to spoken responses. the elai app was developed to provide an easily accessible english language learning app for users that want to receive detailed feedback about their speaking skills during spontaneous speech. the present work was the first evaluation of the elai app. overall, the findings revealed that users spent the majority of their time browsing for prompts, completing new responses, and viewing shallow level feedback. this suggests that users are generally not taking advantage of the more in-depth feedback and resources to facilitate improvement, which were requested by users during interviews [ ] . although the prominence of viewing shallow feedback is disappointing, it could represent productive behavior. feedback overview is the only page in which users can compare their performance on multiple speech features across individual responses, which could reveal patterns of improvement or persistent issues [ ] by leveraging the benefits of open learner models [ ] . future research is needed to determine if this cross-response comparison is occurring and to explore designs to facilitate these comparisons [ , ] as language learners may not engage in self-regulated learning behaviors on their own in mall apps [ ] . we also investigated changes in user performance. our preliminary findings were promising in that more engaged users improved their performance on all six speech features from the beginning to the end of their interaction with the elai app. however, these findings are only preliminary and a more rigorous investigation of the impact of the elai app on speaking skills is needed. overall our initial findings suggest that the elai app is a promising mall, but there is still room for improvement. the feedback summary report, for example, could be improved by requiring less scrolling for users to access learning resources and explicitly highlighting the availability of more in-depth feedback to reduce any lack of feature awareness. tailoring the feedback to user characteristics (e.g., cultural background) could also benefit learning [ ] . new in-app incentives (e.g., badges) could encourage more frequent use (e.g., more than five sessions) and use of the in-depth feedback pages and learning resources. users could also benefit from being shown their improvement over time to implicitly reward continued use of the app. overall, the elai app shows initial promise at creating an easily accessible resource for practicing and receiving feedback on spontaneous speaking tasks, but more research is needed to understand how this app can be the most beneficial to users. mobile-assisted language learning a theory of learning for the mobile age mobile devices for language learning? a meta-analysis smart watches for making efl learning effective, healthy, and happy enhancing l learning through a mobile assisted spacedrepetition tool: an effective but bitter pill? mobile-assisted esl/efl vocabulary learning: a systematic review and meta-analysis duolingo effectiveness study final report mobile-assisted language learning: a duolingo case study migrants and mobile technology use: gaps in the support provided by current tools mobile language learning innovation inspired by migrants hidden in plain sight: low-literacy adults in a developed country overcoming social and educational challenges through mobile learning support tools the effects of feedback interventions on performance: a historical review, a meta-analysis, and a preliminary feedback intervention theory focus on formative feedback user experience of a mobile speaking application with automatic speech recognition for efl learning an overview of mobile assisted language learning: can mobile devices support collaborative practice in speaking and listening? paper presented at eurocall language anxiety and achievement investigating the dynamic nature of l willingness to communicate students' perspectives on foreign language anxiety why do many students appear reluctant to participate in classroom learning discourse? system challenges experienced by japanese students with oral communication skills in australian universities communicative competence: some roles of comprehensive input and comprehensible output in its development three functions of output in second language learning languaging, agency, and collaboration in advanced second language proficiency understanding linguistic, individual and contextual factors in oral feedback research: a review of empirical studies in l classrooms the appropriation of interactive technologies: some lessons from placeless documents levels of processing: a framework for memory research exploring mobile tool integration: design activities carefully or students may not learn self-directed language learning in a mobile-assisted, out-of-class context: do students walk the talk? a power primer protutor: historic open learner models for pronunciation tutoring open learner models mobile-assisted language learning (mall) mobile learning as 'microlearning': conceptual considerations towards enhancements of didactic thinking the role of self-regulation and structuration in mobile learning learners' oral corrective feedback preferences in relation to their cultural background, proficiency level and types of errors key: cord- -dx bbeqm authors: simmhan, yogesh; rambha, tarun; khochare, aakash; ramesh, shriram; baranawal, animesh; george, john varghese; bhope, rahul atul; namtirtha, amrita; sundararajan, amritha; bhargav, sharath suresh; thakkar, nihar; kiran, raj title: gocoronago: privacy respecting contact tracing for covid- management date: - - journal: j indian inst sci doi: . /s - - - sha: doc_id: cord_uid: dx bbeqm the covid- pandemic is imposing enormous global challenges in managing the spread of the virus. a key pillar to mitigation is contact tracing, which complements testing and isolation. digital apps for contact tracing using bluetooth technology available in smartphones have gained prevalence globally. in this article, we discuss various capabilities of such digital contact tracing, and its implication on community safety and individual privacy, among others. we further describe the gocoronago institutional contact tracing app that we have developed, and the conscious and sometimes contrarian design choices we have made. we offer a detailed overview of the app, backend platform and analytics, and our early experiences with deploying the app to over users within the indian institute of science campus in bangalore. we also highlight research opportunities and open challenges for digital contact tracing and analytics over temporal networks constructed from them. contagious viral diseases such as the sars-cov ( ), h n ( ), mers-cov ( ), and sars-cov- ( ) have resulted in global epidemic outbreaks and placed a massive burden on public health systems around the world. these pandemics have cascading effects that result in irreparable consequences to economies and quality of life. the recent sars-cov- or covid- pandemic has triggered national and regional lockdowns across the world to curb the spread of the virus. with incubation periods that last days and with a significant fraction of asymptomatic carriers, the proliferation of the disease has been hard to detect and localize. further, testing of populations at a large-scale has proved challenging due to limited testing kits, well-trained health-care professionals, and funds in emerging economies . to tackle this problem, governments and health workers use contacttracing of infected social distancing: social distancing is the practice of maintaining physical distance between individuals to prevent the spread of face-to-face communicable diseases. a . - m distance is recommended for covid- . tracing is the process of identifying people might be at risk due to physical interactions with a disease carrier. individuals to identify those who may have come in contact with them, also called primary contacts. these primary contacts are then quarantined and/ or tested depending on their symptoms. testing, tracing, and isolation form essential components of covid- management, besides preventive measures like wearing masks, practising social distancing , and washing hands . traditional methods of contact tracing are often laborious and may be erroneous due to recall biases , . also, human activity patterns often involve interactions with strangers, especially when travelling, which makes it difficult to identify contacts using traditional methods. as a large fraction of the population owns smartphones, countries around the world, including india, have attempted to use digital contact tracing , , . mobile apps that use bluetooth technology are deployed to record close interactions between users. these bluetooth low-energy (ble) apps typically advertise a unique device id, j. indian inst. sci. | vol xxx:x | xxx-xxx | journal.iisc.ernet.in which can be recognized by other nearby devices with the app that scan for and save these advertised ids, also called contacts. this information is typically stored on the local device; if a user tests positive, their bluetooth contacts are uploaded to a central database and their contacts are alerted. this can dramatically reduce the time required for contact tracing from days to potentially hours, thereby mitigating the spread of the virus . examples of such national-scale apps include aarogya setu in india, tracetogether in singapore, covidsafe in australia, covid alert in canada, corona-warn-app in germany, etc. however, there are limitations to digital contact tracing. these constraints include the low reliability and asymmetry of bluetooth technology in detecting nearby users , , , ; low accuracy of the proximity distance between users to help distinguish nearby and farther off users , ; high degree of adoption required for digital contact tracing to be effective , ; and the inability to locate secondary and tertiary contacts until the primary and secondary contacts test positive, respectively. it is hence still important to use complementary digital contact tracing with manual methods. in this article, we describe gocoronago (gcg), a digital contact tracing app for institutions, which attempts to address these limitations. a key distinction of our approach is to collect the contact trace data of devices into a centralized database, continuously, irrespective of if or when a person is diagnosed as covid positive. this proximity data of all app users are used to build a temporal contact graph, where vertices are devices, and edges indicate proximity between devices for a certain time period and with a certain bluetooth signal strength. this approach has several benefits. when a gcg user is tested positive for covid- , we use graph algorithms to rapidly identify primary, secondary, and other higher-order contacts, based on who guidelines . further, even if the bluetooth scans were missed by the infected user, successful scans by other proximate devices can be used to alert the relevant contacts, increasing the reliability of detection. in addition, centralized digital contact tracing has the potential to estimate the state of the population using network-based seir models, which can be used to assign risk scores and prioritize testing , , . of course, centralized contact data collection has its downsides, primarily, the privacy implications of tracking the interactions between a large number of individuals. we take several precautions to mitigate this. one, the app is designed for deployment only within institutions and closed campuses, and not at a city, regional, or national scale. the data collected are owned by the host institution and not by a central authority. two, users do not have to share any personal information, and devices are identified using a randomly generated id. sharing gps location or their phone number is voluntary and through opt-in. last, deanonymization of data is limited to covid- contact tracing and, by design, requires multiple entities to cooperate, and is overseen by an advisory board with a broad representation from the institution. we discuss these pros and cons in more detail later. besides a centralized data collection approach, we also conduct experiments to understand the impact of various smartphone devices and the environment on the bluetooth signal strength to better ascertain the proximity between devices. we also send proactive messages for users to enable custom bluetooth settings in their smartphones to improve reliability. the use of the gcg app within an institutional setting, with data collection and usage governed by the organization, may lead to higher adoption of the app and enhance its effectiveness in contact tracing. this article examines the design rationale, architecture, and our experience in deploying the gocoronago digital contact tracing app as part of a pilot at the indian institute of science (iisc). it also discusses the challenges and opportunities in improving the utility of digital contact tracing. the rest of the article is organized as follows: in sect. , we review digital contact tracing and provide an overview of a few popular covid- apps. section provides details of the app design and the backend architecture. in sect. , we describe various analytics, including temporal contact network algorithms, for contact tracing, and for providing feedback to app users. finally, sect. summarizes our experience with deploying the app at iisc and highlights some of the opportunities and challenges of digital contact tracing. j. indian inst. sci. | vol xxx:x | xxx-xxx | journal.iisc.ernet.in background and related work . contact tracing infectious diseases, that spread through personto-person interactions, can be contained by tracking their sources and quarantining the individuals who are or may be affected. this is typically done using physical interviews, which try to determine the places visited and the people met by the patient . in some cases, the location history of the patients is shared by cities and public health agencies on websites and mobile apps to allow others who were in the vicinity at that time to take precautions. this form of contact tracing relies heavily on one's memory and collecting such data manually is cumbersome. contact tracing is crucial, especially for viruses such as the sars-cov- that exhibit high transmission rates, low testing rates, long incubation times, and a significant fraction of asymptomatic carriers, who could infect other susceptible individuals , , . digital contact tracing, on the other hand, involves the use of technology to keep track of the individuals who came in close proximity with each other. it has been shown to be effective in preventing the spread of communicable diseases in livestock , , but experiments involving human populations have been limited . the scale at which covid- has spread has led to the use of bluetooth and gps-based contact tracing applications on mobile phones. such apps help individuals automatically keep a record of the places they visited and the people they met, along with the timestamps. this permits us to build contact neighborhoods that can be used to alert or quarantine the concerned individuals and identify potentially risky interactions. most digital contact tracing (dct) apps for covid- rely on bluetooth technology available on smartphones. in addition, a few apps collect the gps location of users. the rapid spread of the covid- virus has led to the development of a variety of smartphone apps around the world, which are variants on this theme. examples include both national apps like aarogya setu (india), nhsx (uk), and covid safe (australia), as well as those proposed by institutions, like novid (cmu) and safepaths (mit). a review of contact tracing apps can be found in , , , , and their features are contrasted in table . at a broad level, these apps scan and advertise for bluetooth signals and record the timestamp, along with the signal strength or the received signal strength indicator (rssi), reported in decibel-milliwatts (dbm) in android. the rssi values are negative and higher when the devices are close to each other. translating the bluetooth rssi to proximity distances for contact tracing is not straightforward since it depends on numerous factors such as the phone hardware, drivers, operating system, ability to run continuously in the background, and interference due to surfaces. yet, they have been widely attempted and deployed because of its potential advantages over manual contact tracing. in fact, to address some of the interoperability issues across android phones and iphones, google and apple have even introduced an exposure notifications (gaen) protocol into their os as part of their covid- response . the bluetrace protocol used by apps in singapore and australia is another popular standard. europe has two competing contact tracing standards that are being refined, decentralized privacy-preserving proximity tracing ( dp t) and pan-european privacy-preserving proximity tracing (pepp-pt) . the bluetooth special interest group (sig) is also working on a contact tracing standard for wearables . such protocols help with mobility across national boundaries, avoid having to install multiple apps, and in the development of custom, yet interoperable, apps. besides smartphone-based apps, others have also developed hardware devices such as the tracetogether token that uses bluetooth, but operates independently of a phone, or wearables like wristwatches that can track the location using gps . in addition to bluetooth, a few apps like novid also broadcast ultrasound signals using a phone's speakers and other apps in the vicinity detect them using their microphone . there have also been other digital apps such as the nz covid tracer that use qr codes for users to check-in when they enter specific locations . besides contact tracing, digital tools have also been used to track symptoms among populations to identify emerging "hotspots" and for health professionals and volunteers to coordinate their response . however, the global adoption of contact tracing apps is low. the percentage of the population who have installed such apps has struggled to go j. indian inst. sci. | vol xxx:x | xxx-xxx | journal.iisc.ernet.in table table comparing gcg features with other covid contact tracing apps, as on sep past %, even among developed countries where a majority of the individuals have smartphones . while there is debate on the minimal adoption rate required for contact tracing apps to have a tangible effect, some use is better than none and more is better , , . in particular, higher adoption rates in dense neighborhoods can highlight the effectiveness of tracing effective since the risk of spreading the infection is greater in closelyknit communities. there are a number of ways in which one can design such digital contact tracing apps. these offer different trade-offs in terms of individual privacy and the health and safety of the community. the target of the app may be for national/regional use or institutional use. while national-scale contact tracing apps potentially offer greater ability to manage the pandemic, they also carry greater risks of data leaks and misuse . further, a high degree of adoption at such large scales is challenging, limiting the usefulness of the app for contact tracing. apps deployed at an institutional scale can be better targeted to the audience and offer better uptake due to the fact that the data are managed at the organizational level. institutions can also respond more rapidly based on insights provided by the app. but they are less effective when users are moving outside the confines of campuses and interacting with the broader population, e.g., apps like aarogya setu and tracetogether are national apps, while goc-oronago, novid, and covid watch are designed for institutions. the use of the app may be voluntary or mandatory. some countries like china have made such apps mandatory for all residents, or for those meeting certain requirements such as travelers. even organizations may make such national or institutional apps mandatory within their premises. but most countries and institutions tend to keep the use of such apps voluntary. further, the use of the collected data for contact tracing may also be voluntary or mandatory. if voluntary, there is an explicit opt-in by the individual who is tested covid positive or is quarantined, before contact tracing using their data can be initiated. alternatively, there may be rules in place that allow the government or institutions to use any proximity data that are available with them, without additional consent from infected users. an explicit consent helps address concerns of social stigma around covid patients. the use of gcg is strictly voluntary, and there is an additional consent required by a user who is infected with covid- before their data can be used for contact tracing-this, despite their data already being available centrally in the backend. apps may collect identifiable, strictly anonymous, or pseudo-anonymous information as part of contact tracing. some apps like singapore's tracetogether compulsorily require the contact details and/or a national identification number to be shared when installing the app. this makes it quicker to reach-out to users during contact tracing, but also heightens the risk of misusing the data for the surveillance of specific individuals and can lead to a significant loss of privacy if the data arre breached. in a strictly anonymous setting, no personal information of the user is collected, and they are only identified by a random id, which itself may also be changed (or "rotated") periodically. a set of such ids may be provided by a central server (tracetogether) or generated locally by the app. during contact tracing, the user's app is alerted and they have the option of voluntarily responding by contacting the health center or a government agency. if the user uninstalls the app, it may be impossible to do contact tracing. a hybrid approach of pseudoanonymization ensures that the contact trace data themselves are anonymous, but the information required for de-anonymization is available with a trusted independent authority whose consent is required (optionally, with a consent from the infected individual) to identify the users relevant for contact tracing. gcg adopts this hybrid model that balances the privacy of users while also enabling rapid and reliable outreach during contact tracing. the contact tracing data may be kept de-centralized, semi-centralized, or centralized. if decentralized, the bluetooth device ids observed by a user's app are stored locally on the device. when a user tests positive for covid- , they can inform a backend service of their device id (potentially, multiple ids, in case of id rotation) and their status. the backend periodically relays a list of device ids associated with covid positive individuals to all apps, which is then used by the user to verify if they came in contact with a covid positive person. this is used by pact and google-apple exposure notification (gaen) framework . in a semi-centralized approach, a mapping between an app and its device id is maintained centrally, but the contact trace data remains locally on the device. on testing positive, a user may choose to (or be required to) upload the contact trace data for the recent past to a backend service, which then sends notifications to these primary contact devices asking them to quarantine or get tested. examples of this approach include bluetrace and aarogya setu . however, aarogya setu also allows users to voluntarily upload their bluetooth contact data to central servers at any time to get an estimate of other high-risk users in the vicinity. last, in a centralized approach, both the mapping of apps to device ids as well as their contacts are sent to a backend service periodically. when a user reports themselves as covid positive, contact tracing can be initiated on the centralized data already available, optionally after an additional consent. gcg adopts this model. this variant is relatively intrusive, but arguably has advantages that may justify its use. one, contact data from both the infected and the proximate users can be combined to increase the reliability of contact tracing. two, even if users uninstall the app, if the data collected are personalized or is de-anonymizable, then contact tracing can still happen over the backend data for the period during which the app was kept installed. three, not just primary but even secondary and tertiary contact tracing, can be performed rapidly. and four, having a centralized model allows us to perform temporal analytics on a global contact network. this can help identify high-risk individuals for prioritizing preventive, testing and (future) vaccination strategies, and infer the health of the user population. bluetooth data provide the relative interaction between proximate users but in itself does not reveal the spatial location of users. while this may disclose interaction patterns between (anonymous) users, which is necessary for contact tracing, correlating this with particular individuals is not possible without additional out-of-band knowledge about them. some contact tracing apps may also collect gps data (covid safepaths) and data from beacons or qr codes (novid) that may reveal the absolute spatial location of the users. collecting spatial location has some benefits. the coronavirus may be transmitted through surfaces or be suspended in the air and thereby be passed on to others who are not near an infected user but in the same location soon after . bluetooth based proximity will miss such users. also, gps data collection may be more reliable than bluetooth. however, gps is not precise enough to be useful for identifying proximity between users. furthermore, tracking the spatial movements of users continuously can have serious privacy consequences , . bluetooth beacons and scanning qr codes present at well-known locations can also provide such spatial information, but will be limited to places where the beacons or codes are deployed. gcg allows users to optionally share their gps data through an explicit opt-in and also allows the selective use of beacons deployed by institutions. last, we need to consider the duration for which the centralized or de-centralized data that are collected retained. this needs to be explicitly stated by the apps for transparency. more the data that are collected and more personalized it is, the greater are the consequences for retaining it longer, especially in a centralized or semicentralized setting. typically, the contact trace data themselves are useful only for roughly days after they are collected since this duration is typically the outer time-window of transmission of the virus. also, there should be clarity on how long the data are retained after a user uninstalls the app. gcg deletes a user's phone number, the only personal data they may share, from its backend within months of them uninstalling the app. the anonymized contact trace data are retained for future research purposes, as per the rules set out by the institute human ethics committee (ihec). the gocoronago (gcg) contact tracing platform consists of a smartphone app and backend services for data collection, management, and analysis. the app is designed for covid- operations and management within an institution and is also proposed as a research project governed by the institute human ethics committee (ihec). the design and technical details of the app and qr code: quick response (qr) code is a -d barcode standard which serves as a machine or device readable label that encodes information. smartphones can use their cameras to take a picture of the qr code and apps or libraries can extract the information present in them. examples of such information include some identifier, the physical location or a url to a website. beacon is a compact device that can be configured to continuously broadcast an identifier and some custom data as part of a bluetooth signal. other bluetooth-enabled devices can detect these signals to get information, typically specific to the location of the beacon. the backend services are described in this section. a high-level design is illustrated in fig. . the gcg app is limited for use by authorized institutions. since not all institutions may have a private/enterprise app store for their organizations, hosting the app in the public google play or apple app store is convenient. users at authorized institutions are provided with individual invitation codes by a separate entity within the institution, typically the information technology (it) office. the it office also maintains a mapping from the user's unique invite code to the actual individual to whom the code was provided, along with their contact details, as shown in fig. . this mapping from the individual to their invitation code is later used by the it office during contact tracing, as described in sect. . . the user can download the gcg app from the google play store or from an institutional download link. during installation, users enter this invite code into the app, which submits and validates it with the gcg backend servers and is returned a unique id, a device id, and a pin. the gcg backend maintains the mapping from the invite code to the unique id for the installed device. the invitation code can only be used once by the user for the first installation. to allow future re-installations, a pin is generated for this invitation code and is shared with the user. optionally, the user may provide their one-time password (otp)-verified phone number during installation, which is recorded in the backend. this number can be used along with the pin to reinstall the app in the future, in place of the one-time-use invite code. last, a device id in the form of a random bit uuid is generated by the backend for each re/installation on a phone, and a mapping is maintained from the unique id to the device id, along with the creation timestamp. this device id will be broadcast as part of the bluetooth advertisement (fig. ) . both the invite code to unique id and unique id to device id mappings are used during contact tracing (sect. . ) . a final piece of information collected from the app during re/installation is the make and model of the phone. as we discuss later, this is vital for interpreting the bluetooth signal strength and translating it into a distance estimate. these identifiers are designed to maintain the anonymity of users from the gcg app and backend, enable de-anonymization of contact users upon an authorized request for contact tracing, and ensure that the app can be re/installed by authorized users. such sandboxing and identifierindirection ensures that no single entity -the it office, a gcg user, or the gcg backend-can independently find the identity of any (other) user and their trace. a key tenet of gcg is transparency. the installation process in the gcg app has disclosures on the legal terms and conditions for the use of the app, and on how the data collected will be used. in addition, there is also a multi-lingual informed consent, as required by ihec, which clearly documents the scope of the research project, potential benefits and downsides, voluntary participation, etc. the gcg app uses bluetooth low energy (ble) signals to detect other proximate phones running the app. the ble wireless protocol is ubiquitous among smartphones sold within the last years. it enables low-power, short-range wireless communication and is intended for applications in fitness, smart homes, healthcare, beacons, etc. its maximum range is < m though this is affected by environmental conditions and transmitting power, and ≈ m is the typical range . ble devices use an advertising and scanning protocol to discover each other and establish a connection. when acting as a server, the devices advertise one or more services that they support, which are identified by service assigned numbers; when acting as a client, they find servers to connect, to based on the advertised service assigned numbers. a single device may advertise multiple services, and it can include a custom payload such as a service name. also, the ble advertisement is broadcast in an open channel, which any nearby ble client can discover. besides standard bit service numbers that are registered and pre-defined for specific types of services, applications can also generate and use bit uuids for custom services they provide. once discovered, clients can establish a network connection with the service to perform additional operations such as data exchange. the gcg app acts as both a client and a server when using the scanning and advertising capabilities of ble, respectively. specifically, it advertises two service assigned numbers, x , which represents a generic access service, and another custom service whose assigned number is the unique device id for a particular app installation. this advertisement is broadcast continuously. as a client, the gcg app scans for secs every minute for advertisements that contain the service number x . if found, it extracts and records the device id that is sent as a secondary service number in the same advertisement. piggy-backing the device id as a service assigned number rather than a custom payload takes fewer bytes, which in turn can reduce the power consumption for the advertisement. as part of the scanning, the gcg app also retrieves the received signal strength indicator (rssi), which is the strength of the ble signal that is received by the app. as we discuss later, this can be used to estimate the proximity distance. the gcg android app uses the default ble settings for broadcasting its advertisements, which translates to ble broadcasts every sec at a medium transmission power level. also, the app consciously does not establish a connection with apps on another device; the device id is broadcast to any ble device that is in the vicinity. in fact, we explicitly set the connectable flag in the advertisement to false. this enhances security by avoiding malicious content from being transferred. while such proximity tracking is helpful for contact tracing of individuals who were spatiotemporally co-located, this does not address situations where two users shared the same space, such as an atm, mess dining hall, or campus grocery, but for a short time apart. since covid- can be transmitted through surfaces and can linger in the air for some time , it is beneficial to identify users who were in the same location but not at the same time, especially for locations with a lot of footfall. the gcg app allows users to voluntarily share their gps location information with the backend. this is disabled by default. if enabled by the user, the gps location is retrieved and uploaded to the backend every mins, and buffered for retries. since the sharing of gps location is strictly voluntary, gcg supports the selective use of beacons installed by institutions at such highrisk spaces. these beacons behave like a gcg app that passively advertises its device id, and the smartphone app can scan for and record the beacon's id, just as it would detect another gcg smartphone's device id. specifically, we use the ibeacon protocol from apple. the beacon transmits a static gcg uuid as its service number, x c, as the manufacturer id for the protocol, and a major and minor version number to uniquely identify itself. the gcg app scans for the static service number, filters results based on the manufacturer id, and retrieves the major and minor version numbers. the app encodes these version numbers into a template uuid to form a unique device id for that beacon and adds it to its proximity trace. during each scan, the proximity data collected consist of zero or more device id(s) and the corresponding rssi values that were discovered at that timestamp. performing a service call to send these data to the backend servers consumes power and bandwidth on the phone. instead of sending these data after each scan, we buffer it to a sqlite database on the phone and periodically send the buffered data to the backend in a single batch. this transmission interval is set to mins. this type of batching amortizes the power and network costs across scans, while ensuring the freshness of the data available at the backend. buffering is also beneficial when internet connectivity is intermittent. if the proximity data cannot be sent to the backend, the buffered data are retained on the device and a resend attempt is made in the next transmission interval. given that this is the most frequent service call to the backend, we use a compact binary serialization to represent the proximity data sent to the backend, unlike the other services which use json. the gcg app needs to run in the background all the time for effective bluetooth advertising, scanning, and proximity data collection. however, the heterogeneity of smartphone models and the limitations of their os means that this advertising and scanning may not be reliable. to identify issues with specific device models and app installations, and verify if the app is running, we collect and report liveliness telemetry statistics to the backend every hour. these include a count of ble scans performed, ble scans failed, gps scans, gcg users and beacons detected, and contact buffer size; bluetooth and gps enabled status, bluetooth and gps permission flags, battery level, app version, etc. these statistics also help us in understanding the aggregate usage of the gcg app within an institution. besides tracking bluetooth contact data, the gcg app offers several features to inform the users about covid- and engage them in preventing its spread. screenshots of these ui elements are shown in fig. . key among these is a proximity alert, wherein a notification is triggered on the smartphone if or more users (configurable) were detected within a ≈ m distance during the last bluetooth scan. this acts as a warning to users in case they inadvertently overlook social distancing. as discussed later, the m distance threshold is just an estimate based on the rssi. the alert is also triggered only once an hour (configurable) to avoid saturating the user. in addition, users can visualize a plot of the hourly count of contacts segregated by the duration of contact within the hour, e.g., < mins , − mins and > mins (fig. b) . this gives them a sense of their interaction pattern for the past hours. similarly, we also display the number of scans performed each hour for the past h (fig. c) . this can help identify issues with bluetooth scanning on specific phones, and prompts the user to take corrective measures. a summary of the number of scans completed per day is also shown as a progress bar to motivate users to hit or more of the possible min scans (fig. a) . j. indian inst. sci. | vol xxx:x | xxx-xxx | journal.iisc.ernet.in these local analytics within the app are complemented by aggregate analytics performed in the backend and are shared through the app each day. these include the social distancing score, user density heatmap for neighboring locations, and a visualization of the contact network neighborhood. these are described later in sect. . a unique aspect of the app is that the set of remote analytics available can be dynamically changed without having to update the app. in the future, this can also be used to push forms and conduct surveys from within the app. importantly, none of the analytics provided to users reveals the identity of other users or even their device ids, to protect their privacy. for example, the hourly contact bars only report the aggregate counts of nearby devices and cumulative duration of interaction at different distances, while the proximity alert is triggered only if at least three users are nearby to prevent fine-grained estimates of the number of gcg users from being revealed. last, we also provide helpful information to educate users about covid- . these include a plot of the positive, recovered, and deceased cases across time in india, and in the local state, and a map of the current positive cases at the state and district level. in addition, we also share let's control covid and curious about covid? infographics as app alerts each day, which suggest precautions, debunk myths, and offer scientific information (fig. f) . these are sourced from public health and science resources such as who, the covid gyan initiative from iisc-tifr, and indian scientists' response to covid- the features described here are largely applicable to gocoronago v . on android smartphones. gocoronago v . is a lighter version available for ios with features limited to advertising, scanning, and receiving alerts. this is due to the limited numbers of iphone users on the academic campus. there are other os and device-specific issues as well that we encountered and addressed in various iterations of the app. while we were initially using wildcard filters when performing bluetooth scans for service numbers on the android app, we noticed that certain phone models such as samsung did not reliably perform such scans. this led us to adopt the x approach. continuous bluetooth advertisement and scanning in the background is challenging in android, and virtually impossible in ios. smartphone brands with custom android builds, such as xiaomi, oppo, vivo, etc. do not always support the recommended practise of executing such applications as a foreground service with a persistent, ongoing notification. as a result, users are forced to change the android battery usage settings and/or autostart permissions for the gcg app, which are brand and even model specific. absence of reliable scanning and advertising defeats the key purpose of the app. we provide local analytics and alerts to help users address such issues. further, android requires users to enable gps to even perform continuous bluetooth scanning, as a way to indicate to users that their location may be revealed indirectly, say, through beacons at well-known locations. but requiring gps to be on even though the app does not collect the gps location without opt-in confuses users, and may lead to privacy concerns. on ios, the problems with background bluetooth advertisement and scanning is well documented due to apple's restrictive policies , , . the ios gcg app is effective when in the foreground and when the user is viewing the app. however, when the user is not actively using the app or the phone is locked, the app can scan for other devices that are advertising, but it cannot advertise. as a result, there needs to be other android or active ios gcg devices nearby for contacts to be recorded, colloquially referred to as "android herd immunity" . besides technical challenges, there are also policy challenges in deploying covid- related android and ios apps to google play and apple app stores. certification from an official government of india agency with specific verbiage was required before the gcg android app would even be reviewed for hosting on the play store, and the subsequent reviews of the app's update takes weeks. given the restrictions that apple imposes on apps posted on its app store, the ios gcg app is only viable for an ad hoc or enterprise license deployment. gcg web services, data management, and analytics are hosted on the microsoft azure public cloud. as shown in fig. , these are present on different virtual machines (vms) that are segregated based on their workload (service endpoint, data management, analytics), and their security zone (internet, intranet, and internal). we describe these backend capabilities next. a suite of rest service application programming interface (api) is defined for the gcg app virtual machines (vms): a virtual machine (vm) is a computing environment that provides all the functionalities of a full computer, but executes within another computer. a vm is the typical unit of renting a computer in public clouds. vms help divide a single large computer or server in the cloud into multiple smaller computers, and the vms are independently rented to different users. public cloud: public cloud is an internet-based service that allows users to rent and access remote computation, storage and software capabilities that are hosted at large data centers offered managed by service providers like microsoft, amazon, and google. it reduces the cost and effort in managing physical computing infrastructure at an organization, and at a higher reliability and scalability. to interface with the backend, to upload data and to download analytics and alerts. the rest services are implemented using java servlets running on apache tomcat web server, and their service endpoints are accessible on the internet. these apis include register device, add proximity contacts, add gps, add liveliness, get notifications, and fetch analytics. most use json as the rest body, except add contacts which uses a binary protocol. the register device api accepts an invitation code from the app, checks a mariadb table if the code is present, not expired and not yet used, and if so, generates a random device uuid, a random pin and a unique id for the user, which are returned back to the app. these mappings, as described earlier, are maintained in mariadb. the phone number, if provided, is salted, hashed, and stored in the database for comparison in the future if a user reinstalls the app. the number is also asymmetrically encrypted and stored in the database, so that it can be decrypted upon authorization by the institution's advisory board, if needed. the decryption key is store securely off-cloud to prevent accidental breaches. the add contact api is most frequently invoked, once every mins by potentially 's of users. to avoid the power, compute, and network overheads of de/serializing json, we use an alternative binary format. it starts with bytes of the source device id, followed by a series of scan records, one per scan. each record starts with bytes of unix epoch time in seconds with the scan record's timestamp. the next byte indicates the number of device contacts 'n' in that scan, followed by × n bytes having the byte device id and byte rssi value for the n proximate devices. if more than n = devices are found in one scan, the app creates multiple scan records. records are created and sent by the app even if there are no proximate devices, since this information is also useful. as mentioned before, beacons are also encoded as device ids following a standard uuid template. intuitively, each record forms an adjacency list for the contact graph. the binary records from service calls from all users are appended to a file and every h, a pre-processing service fetches these binary files and generates a corresponding csv file with an edge list consisting of the timestamp, source device id, sink device id, and rssi. this csv file is backed up to azure blob store and, as discussed later, stored on hdfs for further analytics. add gps is the next frequently called api, every mins, for users who choose to share their gps location. these data are used to generate a device density heatmap of the user's neighborhood for the recent past, and potentially for contact tracing. to support such spatio-temporal queries, we use the influxdb temporal database to store the gps data. one copy of the latitude and longitude is asymmetrically encrypted and stored in influxdb, along with the timestamp, to support authorized contact tracing. another copy is transformed using a geohash of characters, which reduces the precision of the location to a m × m grid. when generating the heatmap for the app user's current location, we query over this geocode. the app communicates hourly device health data using the add liveliness api, as a set of keyvalue pairs that has evolved over app versions. as a result, we store these data within azure cosmos db, which is a nosql database. these data are later queried for identifying devices that are not reporting bluetooth data reliably for sending alerts with possible fixes, and also for monitoring the overall status of the gcg deployment at an institution. alerts are sent to the app using a custom notification service in the backend that the app polls every mins. this approach was initially chosen over google or apple's push notifications to reduce the dependence on external services. alerts that are generated by various analytics are inserted into a mariadb table with the device id, title, content, type, and validity time range. when an app polls the service, any pending alerts for that device are returned. besides displaying alerts to the user, they may also have a special payload that triggers changes to the ui, such as updating the social distancing score on the main screen. user-level analytics such as displaying their contact network and other analytics such as the user density are sent to the app as html that is locally rendered. the app invokes a get analytics api, which returns a json containing a list of current endpoints that serve the analytics. the plots and maps are served off an apache instance. separately, we also run our own open street maps tileserver for serving the map tiles. these external-facing services are hosted on a separate set of vms over which the services are distributed based on their workload and to avoid performance interference. these vms are shown in orange in fig. . we use one azure d s v vms to host the register device, add gps, and add liveliness endpoints, a second one that exclusively runs the add contact, and another to run the get geohash: geohash is a mechanism to encode a location in the form of a compact sequence of alphabets and numbers that are easy to remember, compared to latitude and longitude. typically, longer hashes offer a higher precision of the location. programming interface (api) is a description of the input and output parameters that are received and returned when accessing a capability offered by an application. j. indian inst. sci. | vol xxx:x | xxx-xxx | journal.iisc.ernet.in notifications service; the latter two see a higher load. the tileserver for displaying open street maps, which is only occasionally used, runs off an azure b s vm, while the analytics are served from an azure d s v vm. a separate azure d s v vm hosts mariadb and influxdb used by these services. besides the internet-facing services, there are internal services to support the gcg platform. these are used to host an operations portal to oversee the health of the system, on-boarding of devices, and visualize the contact network. the portal does not directly access any user database or files to prevent accidental access to or modifications of the raw data. instead, a separate routing service offers a limited set of well-defined services to access authorized data. these apis are periodically called and the results are cached in a separate mariadb instance used by the portal. the portal and its database are also hosted on separate vms, shown in yellow in fig. . this sandboxing also extends to the analytics services, which too do not directly access the user databases for sending alerts or generating visualizations, but operate through this routing api. for example, the liveliness data are fetched every mins through this routing service from cosmos db and into mariadb for the portal to visualize the number of scan records received and scans failed among the apps, while the device registration summary is fetched through the api to plot the users on-boarded over time, distribution of their device make and models, etc. ensuring the security of the services and the data collected by the gcg platform is of paramount importance and is intrinsic to various design and deployment choices. all the rest endpoints use http/ with http strict transport security (hsts), which forces the use of a transport layer security (tls . /ssl) encrypted channel between the gcg app and the backend and prevents man-in-the-middle attacks. further, all service calls are authenticated based on a device key that is returned to the app during registration. to ensure that this service call authentication is light-weight, we use a digital signature protocol, which ensures that each call can be locally validated, without the need for any database (fig. ) . specifically, the device key is generated by the backend service as key = base (sha (device id, salt)), where salt is a secret phrase known only to the service. the gcg app encrypts and stores this device key on the phone. subsequently, when invoking any backend service, the app sends its device key, the current timestamp, and a signature, which consists of sign = base (sha (device id, timestamp, device key)) as part of its https header or body. the service then uses the received device id to generate the device key on the fly, and additionally uses the timestamp to generate the signature. it also verifies if the timestamp rest: representational state transfer (rest) is a software architecture that allows desktop and mobile clients to interact with internet services by passing requests and receiving responses, using web standards such as http and data models like json. passed is recent, for mitigating replay attacks. if the generated signature matches the received signature, the request is valid and is executed. note that all of these are flowing over an encrypted https channel. various other best security practises are used. the register device service takes measures to mitigate brute-force attacks using random invitation codes and pins by limiting the number of daily attempts. internal services such as the portal are only accessible from the institution's private network, over vpn, and are additionally secured using authentication. firewall rules are used to restrict access to unused ports. direct ssh access is not available to any vms running services or the database. the internet-facing vms are in a separate subnet from the ones hosting the databases and internal services on azure to keep the networks in different security domains. data flows between the services and databases/storage are tightly controlled and a routing service used for internal services. we run the latest stable release of all software and the latest security patches to protect against known security flaws. the mariadb sql database follows the principle of least privileges for access, and only minimal permissions for select or select/ insert are given to user accounts. user-defined functions are disabled. all queries are templatized to avoid sql code injection. sensitive data such as phone number and location are kept hashed and/or encrypted when stored. this prevents privacy from being compromised even if there is a cloud security breach and the data are leaked. we use asymmetric public-private keys so that only public keys are hosted on the vm for encryption and private keys for decryption are kept securely offline. contact data are backed up to azure encrypted blob storage. the backend services have undergone professional vulnerability and penetration testing by crossbow labs. the gcg app is designed to provide feedback to users on their daily interactions using simple metrics and contact neighborhoods. additionally, to improve user engagement, the app also provides heatmaps of user density and charts and maps that show the covid- situation in various states and districts around the country. in this section, we describe these features along with the contact tracing protocols that are in place if an app user tests positive. we receive contact records from various devices that contain the contact timestamp and associated bluetooth signal value. for efficient primary and secondary contact tracing, we periodically stitch these contact records to create a global contact network graph. further, we annotate the edges with the contact timestamps and signal values to creating a temporal contact network or a temporal graph. we use apache spark to perform this stitching from the csv edge file, as a pre-processing step. specifically, we create an interval graph for scans received during a specific time interval. the spark application takes a start and end time for the interval, and then filters in all the edge list entries in the input csv file whose timestamp falls within this time interval. it then groups all edges by their source and sink vertices to create an adjacency list for each vertex that includes all scan entries from either source or sink edges. every edge is characterised by a time interval [t s , t e ) , where t s is the earliest scan timestamp and t e is the latest scan timestamp between the connecting devices, during that interval. scans on an edge that fall on adjacent time points with the same rssi value are combined to form longer intervals on the edge annotations. this gives a set of disjoint sub-intervals on the edge with an associated bluetooth signal strength. the output is stored in hdfs for future analysis. temporal graph: like a regular graph, a temporal graph (or temporal network) is a collection of vertices and edges between vertices that indicate a relationship between them. but the vertices and edges that exist at different points in time may vary, and their attributes may also change over time. e.g., temporal graphs model interactions in a social network, traffic flow in a road network and proximity contacts in a contact tracing network. the social distancing score provides users with a measure of their extent of social distancing, on a daily basis. unlike the local bluetooth data used to plot the contact counts on an hourly basis within the app, the social distancing score uses more global knowledge from a device and its neighbors. in particular, it accounts for "background devices" that are often or always in the vicinity, such as family members or hostel room neighbors, and which are subtracted from this score as their sustained presence does not pose any additional risk. these scores are calculated using apache giraph once a day, over the interval graph created for the preceding -h period. the score calculation depends on three parameters: signal threshold (δ) , minimum contact duration (φ m ) , and background contact duration (φ b ) . for each device id, we first identify those neighboring devices that could detect each other for at least φ b mins , cumulatively, during the -h period. these neighbors form the background devices and are eliminated from further analysis. currently, we use φ b = mins. next, from the remaining neighbors, we retain only the rssi entries which exceed a value of δ on their edge sub-intervals. this helps identify the duration of nearby contacts with them. based on experiments described in the next section, we set δ = − , which approximates a distance of m. we sum up the duration of nearby contacts for each edge, and those whose duration is greater than φ m mins form the proximate contacts, p. we set φ m = mins by default. intuitively, this means that the user has interacted with p other devices in close physical proximity of about ≤ m for a cumulative of mins or more in the past h, but who are not part of the sustained background presence. from this, the social distancing score for a device is calculated as max{ , − p} . this normalization offers a higher score for users who practise social distancing and a lower score for the others. in the example snapshot, assume that δ = − , φ m = mins and φ b = min . for the device c, devices b and d are proximate contacts since their close contact durations are h and h, respectively. however, a is not a proximate neighbor of c since it is a part of its background, having been detected for a total of h. so the social distancing score of c is . measures the sars-cov- virus is currently assumed to spread by 'contact and droplet' as well as airborne transmission . who and various countries have provided social distancing advisories that emphasize a minimum spacing of - m for curbing the spread of the virus , , , , . being able to nudge users to maintain such distancing is one of the goals of the gcg app. however, inferring distances accurately from bluetooth rssi values is non-trivial. factors such as smartphone hardware variations, body interference, and multi-path interference lead to both false-positives and false-negatives while estimating the distance from rssi values , . researchers elsewhere have conducted experiments to understand if contact tracing apps can estimate if two users are close to each other, i.e., within a distance of m for mins or longer . these were performed with google pixel and samsung galaxy a devices using the open-trace app, an open-source version of singapore's tracetogether app . they used different environmental conditions such as signal attenuation by the human body, a handbag, walls, etc. and also by enacting real-world scenarios. the measured rssi and the distance are plotted over time to understand the variability for different configurations and their relationship to the ground truth. another smart contract tracing (sct) system uses machine learning classifiers to classify the contacts as high/low risk using the bluetooth rssi values. they perform experiments to collect rssi from a nokia . with android and htc m with android . for distances ranging from . - m, and for random device orientations, and at different locations such as hand, pocket, and backpack. the collected data are labeled as + (high-risk, ≤ m ) or − (low-risk) according to the ground truth. they filter the data using a moving average filter before training using machine learning classifiers like decision tree, linear discriminant analysis, naïve bayes, k nearest neighbors, and support vector machine. the google-apple exposure notification api in android also applies ble calibration corrections based on manual measurement of the signal strength under standard conditions. given the hardware diversity we observe among our campus population, we conduct similar lab-scale experiments, as described, using a more diverse number of smartphones and beacons. we evaluate the effect of rssi at , , and m distances to help us determine whether two phones are within m. we use a debug version of the gocoronago android and ios apps that log the bluetooth scan information to a local file on the smartphone in our experiments. the experiment was performed in an open room measuring about × m with few furniture, mimicking a real-world environment. our experiment uses android devices, iphones, and all the devices were used at a high battery level, with power-saving modes disabled and screen set to stay on for as long as possible while performing the bluetooth scans. each experiment configuration was performed for a period of mins to give ≈ rssi measurements per device pair in that configuration. given the technical limitations of ios, android devices can detect other android devices and the beacons, and iphones can detect the android devices. considering these factors, two experimental setups were designed to collect the rssi data as illustrated in fig. . for the distance a = m , we use a hexagonal placement, as shown in fig. a , with pairs of devices at the vertices, a, b, c, d, e, f, and the center, g. these give us devices at distances of m (same vertex); m, between adjacent vertices, e.g., a-b; m, between vertices at diagonal corners, e.g., a-d; and √ m for vertices that are two hops away, e.g., a-c. three runs with the hexagonal setup are required to ensure that every pair of devices is measured at a m distance. for distances a = m and m the devices were arranged in three clusters, a, b, c, at the corners on an equilateral triangle with a side of length a (fig. b) . in each cluster, the devices are placed vertically and adjacent to each other, in a row. devices across clusters are separated by a distance a while those within a cluster have a distance of ≈ m . three runs of the triangular setup with different clusters are performed to ensure that we get the rssi for each pair of devices at m and m. a key rationale for this study is to understand if two devices are within m of each other or not, as we use the m distance as the proximity threshold in our platform. a total of rssi data points at m, data points at m, and data points for m are collected. we focus our analysis on just the android phones, which form the bulk of our deployment. there are , , and data points for , , and m between the android devices, respectively. for each distance and a device pair, we drop the maximum and minimum rssi values to eliminate outliers. an empirical cumulative distribution function (cdf) of the rssi values at , , and m are shown in fig. a . the x-axis shows the rssi values, while the y-axis lists the corresponding percentiles for different distance configurations. we see that there is a substantial overlap between data points at the three different distances for a given rssi. for example, for an rssi of ≤ − , we have % of the m data points, % of the m data points, and % of the m data points fall within that signal strength. so, using any single threshold value of rssi as an estimate for a m distance is liable to result in both false positives and false negatives. for this preliminary study, we wish to determine an rssi value that is the most discriminating with regard to the ≤ m and > m proximity. so for each rssi value, we plot the difference in the percentile of data points that are at m and at m distances, and this is shown in fig. b . the peak difference is observed at an rssi value of − , i.e., the difference between the true positive of m ( %) and false positive of m ( %) is the highest. hence, we use an rssi of − as the proximity threshold in our gcg app and the backend analytics. in the future, we propose to study the effect on rssi from different pairs of phone models and in different environmental conditions in order to develop a more customized proximity threshold, instead of using a single global value that is currently adopted. when an app user tests positive for covid or is under mandatory quarantine, the current protocol at iisc requires the campus health center to check if the user is willing to share their contact data for tracing. if so, they are asked to enter their phone number within the gcg app, if not done so. the health center collects and enters the gcg unique id, device id suffix, and phone number from the user into a portal. this initiates a call to the gcg backend and triggers an otp to the user's phone number, if the details match with an existing user. the user may share this otp with the health center and this serves as their informed consent for contact tracing. the health center enters the otp and any additional details about the subject, such as symptoms, start and end dates for contact tracing, and test information. the gcg backend confirms if the otp is accurate, and if so, the request is forwarded to the advisory board to get the primary and secondary contacts for this user. the advisory board has representatives from the institute, including faculty, staff, students, doctors, and a bio-ethicist. if the board approves the request through their portal, the gcg backend is notified and it will perform a time-respecting breadth first search (t-bfs), which is a variant of breadth first search (bfs) performed over the temporal contact graph. the t-bfs will be initiated from the device id corresponding to the given user's unique id and for the time duration in the past indicated by the health center. if the user's unique id is associated with multiple devices during this period, the search will be initiated from each of these ids. the output is a list of device ids for the primary and secondary contacts. we then use the invitation code, unique id and device id mappings maintained in the gcg backend to get the list of invitation codes used by the primary and secondary contacts. these invitation codes are shared with the it staff, who then use their mapping table to deanonymize them and provide the health center with a list of email ids and/or phone numbers of these contacts. the gcg backend also provides the duration of contacts for each of the invite codes. the health center can then choose to initiate their relevant protocols for reaching out to these contacts, and quarantine or test them. if mandated by law, the health center may share the contact trace data with the local government agency responsible for covid- surveillance. engagement besides the local analytics within the app, we also provide additional analytics to the gcg user based on aggregation in the backend. figure d shows a heatmap of gcg user count in a . × . km area around the current location of an app user, if they share their gps location. it is aggregated over the past h from users who share their gps data. these data are queried from the timestamp and geohashes present in the influxdb backend. in order to respect privacy, the location data are spatially coarsened into tiles of approximately m × m , and temporally coarsened over h, and only the aggregate count of users in each tile is shown. also, when few users are present in a tile, we display these data in a categorical manner, e.g., < . the contact graphs that are constructed in the backend can be visualized using tools such as gephi. figure shows a subset of the temporal graph generated for a single day. here, the size of a node depends on its degree centrality measure across the entire time duration. the thickness of the links depends on the duration of their contact. while such a graph is instructive for backend analytics, we use it to generate a neighbourhood tree for each user, as shown in fig. e . the tree is based on the last h of data and contains contacts up to two hops. importantly, this is a tree and not a neighborhood sub-graph to preserve privacy, i.e., edges between the -hop and -hop neighbors are not shown to avoid revealing contact patterns between them. these trees are generated on a daily basis. it helps the users get a sense of not just their primary contacts, but also their secondary contacts, which could be much larger, and in-turn motivate users to take greater precautions by socially distancing. the gcg app is currently deployed at the indian institute of science (iisc), bangalore. the iisc campus is an access-controlled residential campus with close to students, over faculty, centrality measure: centrality measure is a graph-theoretic score that measures the relative importance of vertices in their ability to spread or influence other vertices in the network. examples of these measures include degree, betweenness, eigenvalue, closeness centrality, page rank, etc. they are used to identify important or critical vertices in contact networks, social networks, www graphs, road networks, etc. and over research and administrative staff. a majority of the students and faculty live on campus. however, iisc entered a full shutdown in march, , a few days ahead of a nation-wide lockdown in india, and the students on campus were instructed to leave for their homes. initial versions of the app were tested among faculty volunteers during the lockdown period. the gcg app was first rolled out to students in june, after a subset of them were allowed to re-enter campus, and subsequently to other faculty and staff. at the time of writing this paper, the gcg app has been installed by over users at iisc. a plot of the number of installations of the gcg app over time is shown in fig. . sharp jumps in installations correspond to new invitations or reminders sent to students, faculty, and staff for installing the app. the app is yet to be rolled out to essential workers such as hostel cooks, cleaning staff, and security personnel, and noticeably, some of the early cases of covid- on campus have been initiated through them. this is understandable since many of them stay off-campus and possibly have a larger mobility footprint, increasing their risk of acquiring the coronavirus. while the gcg android app was initially hosted on the iisc website due to restrictions by google and apple in hosting covid-related apps on their online app stores, it has recently received approval to be hosted on the google play store, with v . currently available there since early august, . an ad hoc ios version is also being tested since the last week of august, . while gcg is designed for institutional use, contact tracing for users from the same institutions who interact outside the campus is also captured. this benefit can be further enhanced through a federated deployment for institutions that are spatially close to each other, such as a cluster of college campuses and software tech-parks in the same neighborhood. here, the chances of physical interaction between users from different organizations are high, e.g., visiting the same local cafeteria or grocery store. in this federated deployment (fig. ) , individual institutions would maintain their independent gcg deployments. but in addition, they would share the strictly anonymized contact graph for their institution with a trusted data broker, such as a non-profit agency or a neutral university. this data broker would then stitch these graphs together based on contacts between unique device ids that span graphs from different institutions. this can then be used to trigger "glocal" analytics-a global combination of local clusters that are near each other-and share more accurate proximity scores with the users of individual institutions, as well as perform more effective contact tracing across institutions in the same community. a key requirement to preserving privacy is that no personal data should be shared with this trusted broker, and any de-anonymization for contact tracing should strictly be handled at the local institution. this can further be complemented through the use of national or regional-scale contact tracing apps, even if used by a smaller fraction of users who are mobile. this can help link clusters of gcg contacts within institutions, and allow with contact tracing beyond the institutional premises as well. however, care should be taken to sandbox the regional and institutional datasets to avoid privacy loss. the availability of fine-grained contact tracing data has opened opportunities for new research on infection spreading. classic epidemiological models are compartmentalized formulations that classify the population into different states such as s (susceptible), e (exposed), i (infected), and r (removed/recovered). based on the progression patterns of a disease, different models such as si, sis, sir, and seir models , , , have been proposed. these models are applicable to large populations and can estimate the time evolution of the fraction of individuals in different states over time and can identify the peak number of infections for different reproduction numbers. the assumptions in these models are, however, coarse and their utility is hence limited. they can be used to take higherlevel policy decisions such as deciding the duration of lockdowns, planning hospital bed-capacity over time, etc. however, the input data for these models are tightly related to the testing rates, which in the case of covid- was very low during the initial few months. research in the past two decades has extended such compartmentalized models to static or timevarying contact networks , , , . in a static network, a node, if infected, can potentially infect any other nodes that it comes in contact with, regardless of the time of contact. but in dynamic networks, temporal ordering is preserved. that is, if an individual a comes in contact with a person b before b and c interacted, then a faces no risk from c. this can correct for the over-prediction of infection rates from static models. with bluetooth-based mobile contact tracing, it is possible to include both duration of contact and the signal strength, which is a proxy for the distance between the phone users during their interaction, to make better predictions of the transmission rates. results from simulated experiments by kretzschmar et al. , indicate reduced reproduction numbers when contact tracing is performed using mobile apps as the delay in alerting vulnerable individuals is reduced to a minimum. apart from identifying primary and higher-order contacts quickly, contact data allow us to identify the most vulnerable users through either simulations of network models assuming hypothetical initial conditions or centrality measures. most centrality scores from network science are defined on static graphs, and it would be interesting to develop better centrality measures that can be used to find the nodes with higher spreading capabilities in a temporal network. identifying such individuals can in-turn be used to device adaptive testing and vaccination strategies, which can help improve the estimates of the health states of the population, especially when testing is expensive, or its availability is limited. another major opportunity with centralized contact tracing is the ability to influence social distancing behavior using alerts and scores. creating control groups and providing such information to one of them and observing their contact patterns for a limited subsequent period can throw light on the effect of such scores. such randomized control trials can help quantify the effectiveness of contact tracing apps even in the absence of covid- case data. one of the key challenges with digital contact tracing is user adoption. as highlighted in sect. , digital contact tracing requires a large fraction of users within the community to use it before it becomes effective. having only a small sample of individuals use the app makes it difficult to identify the true sources of infection, because of which paths between infected individuals and their primary and higher-order contacts may go undetected. however, our experience with institutionallevel contact tracing appears more promising than that employed by governments at a national level in terms of the fraction of users installing an app and the duration for which they had it installed on their phones. in fact, recent reports indicate that even % of user adoption of contact tracing apps can have a meaningful impact of - % reduction in covid infections and death . that said, not all workplaces are captive environments. in such cases, neighborhood or regional deployments of contact tracing apps may be required since they are more likely to interact with people outside their cluster. further, people may also interact during activities outside workplaces and their institutional contact tracing app can be ineffective during these periods. we frequently observe app users turn off their bluetooth or gps, because of which the contact trace data collected are curtailed. users may do so to save battery-even though our experience shows that the android app consumes less than % of batter in an entire day-or when they perceive a lower risk based on their current activity and environmental conditions. these factors can dramatically offset the promises offered by network-based epidemiological models in identifying risk-prone individuals and in contact tracing to contain the spread of infection. it is also extremely difficult to impute such missing data and no assumption can be confidently justified. although digital contact tracing apps have several potential advantages, validating its usefulness is tough. the difference between the two approaches can be best demonstrated when there are covid positive app users who have shared data for continuous periods. in practice, it is wise to use data from such tools in conjunction with manual contact tracing since there would be gaps in data due to user behavior or technology limitations. building robust epidemiological models is all the more challenging because they contain several parameters that have to be calibrated from sparse and missing data. heavy reliance on digital contact tracing apps can also exclude fractions of the community who use feature phones. visitors to institutions such as delivery providers can also be missed out but can contribute to virus spreading. digital contact tracing is still in its infancy. it is important that individuals understand the data shared, risks, and benefits before fully using such apps. communicating these details to a lay audience can be challenging and misconceptions about what such apps collect and can do are not uncommon. in this article, we have described the various dimensions of digital contact tracing for managing the covid- pandemic. we have highlighted the approaches taken by diverse apps globally and their pros and cons. we have proposed gocoronago as an institutional contact tracing app, whose design choices attempt to balance the privacy of individuals with the safety of the community in performing rapid multi-hop contact tracing. we have offered a detailed technical description of the gcg app, its backend services, and analytics. this platform is currently being validated at the iisc university campus, with additional campus deployments underway. we have shared our early experiences with the deployment over the past few months, in the midst of the covid- epidemic, and the opportunities and challenges that lie ahead. given the evolving nature of covid- , our continued experience with this contact tracing platform at iisc and other campuses can serve as a role model, or a cautionary tale, in managing the pandemic in the ensuing months and years. springer nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. using the data collected from the app is currently under review by ihec. the authors are also glad for valuable inputs from dr. olinda timms from st. johns research institute and prof. mukund thattai from ncbs on the design of the contact tracing protocol to balance safety and privacy. a special thanks to crossbow labs for their pro bono security testing services. received: august accepted: september advisory on social distancing measure in view of spread of covid- disease. tech rep . world health organization (who) ( ) contact tracing in the context of covid- : interim guidance world health organization (who) ( ) coronavirus disease (covid- ) advice for the public coronavirus: people-tracking wristbands tested to enforce lockdown ) digital tools for covid- contact tracing: annex: contact tracing in the context of covid- . tech rep . google and apple ( ) exposure notifications: using technology to help public health authorities fight covid- centers for disease control and prevention (cdc) ( ) social distancing modeling the combined effect of digital exposure notification and non-pharmaceutical interventions on the covid- epidemic in washington state a survey of covid- contact tracing apps infectious diseases of humans: dynamics and control incubation period of novel coronavirus ( -ncov) infections among travellers from wuhan network science an overview of mobile applications (apps) to support the coronavirus disease- response in india bluetrace: a privacy-preserving protocol for community-driven contact tracing across borders indoor distance estimated from bluetooth low energy signal strength: comparison of regression models bluetooth sig to extend reach of covid- exposure notification systems automated and partly automated contact tracing: a systematic review to inform the control of covid- why the nhs covid- contact tracing app failed covid- contact tracing apps reach % adoption in most populous countries physical distancing, face masks, and eye protection to prevent person-to-person transmission of sars-cov- and covid- : a systematic review and meta-analysis social distancing: the science behind reducing from two metres to one metre. independent trace together token: teardown and design overview inferring distance from bluetooth signal strength: a deep dive editorial board ( ) much-hyped contact-tracing app a terrible failure. the sydney morning herald quantifying sars-cov- transmission suggests epidemic control with digital contact the effect of network topology on the spread of epidemics temporal dynamics in viral shedding and transmissibility of covid- critical mass of android users crucial for nhs contact-tracing app. the guardian demographic structure and pathogen dynamics on the network of livestock movements in great britain a contribution to the mathematical theory of epidemics mathematics of epidemics on networks contactbased model for epidemic spreading on temporal networks impact of delays on effectiveness of contact tracing strategies for covid- : a modelling study effectiveness of isolation, testing, contact tracing, and physical distancing on reducing transmission of sars-cov- in different settings: a mathematical modelling study coronavirus contact tracing: evaluating the potential of using bluetooth received signal strength for proximity detection decentralized is not risk-free: understanding public perceptions of privacy-utility trade-offs in covid- contact-tracing apps covid- mortality is negatively associated with test number and government effectiveness accuracy of bluetooth-ultrasound contact tracing: experimental results from novid ios version . using -year-old phones a computer oriented geodetic data base and a new technique in file sequencing covid- and your smartphone: ble-based smart contact tracing no, coronavirus apps don't need % adoption to be effective use of social network analysis to characterize the pattern of animal movements in the initial phases of the foot and mouth disease (fmd) epidemic in the uk uber removes racy blog posts on prostitution, one-night stands the pact protocol specification. private automated contact tracing team mobile location data and covid- : q&a the epi info viral hemorrhagic fever (vhf) application: a resource for outbreak data management and contact tracing in the - west africa ebola epidemic covid- digital contact tracing: apple and google work together as mit tests validity simulation of an seir infectious disease model on the dynamic contact network of conference attendees aerosol and surface stability of sars-cov- as compared with sars-cov- virus spread in networks epidemic spreading in real networks: an eigenvalue an individual-based approach to sir epidemics in contact networks china's virus apps may outlast the outbreak. stirring privacy fears the authors acknowledge a research grant from the department of science and technology (dst), government of india, to partly sponsor this work (grant no. dst/icps/ rakshak/ ). they also recognize the support offered by the rakshak review committee. yogesh simmhan was supported by the swarna jayanti fellowship (grant no. dst/sjf/ eta- / - ). the authors thank the administration of iisc for assistance with the development and deployment of gcg, the members of the institute who volunteered to test early versions of the app, and prof. y. narahari who offered valuable guidance to the project. the authors are grateful for the detailed feedback offered by the institute human ethics committee (ihec) at iisc in designing the operations and the research study. a proposal for research key: cord- - dm asen authors: joo, jaehun; shin, matthew minsuk title: resolving the tension between full utilization of contact tracing app services and user stress as an effort to control the covid- pandemic date: - - journal: serv bus doi: . /s - - - sha: doc_id: cord_uid: dm asen although contact tracing apps can be effective for controlling covid- , the app usage can be stressful for users. this study identifies countermeasures for users’ stress while maximizing full utilization of the apps. this study presents the relationships among the stress factors, users’ appraisal, users’ emotion focus coping, and the infusion to exert the full potential of the app through a structural equation model. the research model is validated by surveying health code app users. given the results of the study, the contact tracing apps could become a valuable tool to control covid- by removing app users’ privacy concerns. overcoming coronavirus disease is the largest pending global issue. despite worldwide efforts, there currently is no specific vaccine against covid- (cdc ) . meanwhile, health authorities are suggesting preventative measures such as tracing and isolating early-stage confirmed covid- patients and those who contacted the patients, along with social distancing, mask wearing, and hand washing (un ) . authorities around the world are temporally permitting implementations of digital contact tracing apps on smartphones to find confirmed cases and trace their travel logs (kelion ; servick ) . a recent research showed the effectiveness of the digital contact tracing apps for epidemic control using mathematical simulations (ferretti et al. ) . the simulation based on a mathematical model proposed by ferretti et al. ( ) provides evidence that digital contract tracing apps could stop the covid- epidemic by using the reproductive number, r = . calculated from data collected from the early covid- stages in china. some of the successful examples of contact tracing apps in usage are self-quarantine safety protection app required for south korean residents by its government and health qr code apps (hereafter called health code) mandated by the chinese government (cdschq a; gan and culver ) . there are two types of contact tracing technologies: a centralized system and a distributed system (ferretti et al. ; servick ). the centralized system stores all contact tracing information on a central server using the global positioning system (gps) to trace travel logs of smartphone users (servick ). the distributed system uses bluetooth low energy (ble) technology to trace travel information and store data in individuals' smartphones (ferretti et al. ; servick ) . in the centralized system, collected data include information on when, where, and with whom an individual met, what the individual bought, and the activities the individual conducted. thus, while the centralized system is efficient as the quarantine authority can integrate and manage all relevant data, users are concerned about authority's invasive surveillance powers (gallagher ) . in other words, while users do understand that contact tracing apps are needed during the pandemic, they also exhibit increased stress and anxiety over security issues related to their personal information (mozur et al. ) . in general, technology users feel more stressed under mandatory settings of technology acceptance than from voluntary acceptance (marakhimov and joo ) . despite users' stress, the mandatory acceptance of contact tracing apps results in greater effectiveness of reducing the spread of covid- than voluntary acceptance. thus, for being required by both governments, self-quarantine safety protection app of south korea and health code app of china are representative technology for examining the relationship between the effectiveness of mandatory centralized contract tracing apps and user stress. technology related stress (called technostress) results in a negative emotional state and a severe state of anxiety (la paglia et al. ) . coping theory refers to a process of conscious and unconscious efforts to overcome stress (lazarus and folkman ) . it is necessary to examining the tension between the diffusion of contact tracing apps and user stress by applying coping theory. to exert full potential of contact tracing technology to control the spread of covid- , it is critical to resolve the tension between benefits of the contact tracing apps and users' stress. moreover, it is necessary to find the causes of contact tracing app user stress which restricts the utility of the app. first, this study aims to examine users' accuracy concerns arising from potential problems of using contact tracing apps and privacy concerns from privacy infringement as the potential causes of user stress. based on the coping theory (beaudry and pinsonneault ; lazarus and folkman ) , this study proposes a structural equation model that shows the relationships between contact tracing app users' stress and how they accept such stress through a process called challenge appraisal. once users appraise the stress as an opportunity they emotionally cope with the stress and they may engage in the infusion behavior of using the contact tracing app to its fullest potential (jones et al. ) . to test the research hypotheses derived from the proposed structural equation model, survey data were collected from the users of health code, which is the mandatory and centralized contact tracing app with the largest user base. contact tracing apps can be used for a variety of purposes even when the vaccine for covid- is developed. vaccines as a means of disease prevention have limitations that new vaccines should be developed when a new infectious disease emerges, and their development takes a long time. however, the contact tracing apps have the advantage that they can be applied quickly without major modification even for new infectious diseases. since this study can shed insight on maximum benefits of the contact tracing apps while protecting individual privacy, it can contribute to preventing the spread of new infectious diseases including covid- . south korea is one of the few countries that are successfully dealing with covid- . as shown in table , diverse information communication technologies (icts) have been applied to prevent the spread of covid- in south korea. icts such as artificial intelligence and big data are applied to the support for the treatment of covid- . the covid- epidemiological investigation support system (eiss) combined with physical interviews plays a great role in tracking confirmed cases and contacts of covid- (the government of the republic of korea ). this system identifies the travel logs of only those patients who have been determined as confirmed cases by the korea centers for disease control (kcdc) on the map and supports quick responses to covid- control teams using the relevant statistical information (park et al. ) . the eiss integrates data in conjunction with the smart city data hub, which collects and processes data from large cities, developed by the ministry of land, infrastructure and transport. the eiss analyzes data including location information and credit card usage details of confirmed cases, in real time using with the support diverse statistical methods to automatically identify travel logs and points-of-stay by time zones, and provides routes of infection and hot spots to identify the source of infection in each area (park et al. ) . by using the eiss, the travel routes of confirmed cases can be identified and analyzed within ten minutes (the government of the republic of korea ). contact tracing is critical in epidemiological investigation (the government of the republic of korea ). identification of those who have been in contact with a confirmed case in early stages plays an important role in preventing the spread of covid- (park et al. ) . therefore, many countries have adopted mitigation and suppression strategies that trace the travel routes of confirmed cases to identify and isolate contacts, thereby reducing the overall scale of incidence (walker et al. ) . diverse apps for digital contact tracing have been developed and used in various countries (kelion ; servick ) . successful examples include self-quarantine safety protection app (south korea) and health code app (china), both of which are based on the centralized system (cdschq a; gan and culver ) . the south korean self-quarantine safety protection app offers services such as self-diagnosis of the health conditions of individuals in self-quarantine, guidance for living rules, and emergency contact networks to effectively control individuals in self-quarantine (cdschq a). individuals in self-quarantine are required to icts for contact tracing self-diagnosis app: an application that supports the self-diagnosis of entrants from overseas countries with fever, cough, sore throat, and breathing difficulties, etc. self-quarantine safety protection app: an application for persons in self-quarantine to enter their health condition twice a day, and for notification of breakaway from the quarantine area an app for self-quarantine managers and for notification of breakaway: an application for management of persons in self-quarantine and for notification of breakaway from the designated place epidemiological investigation support system: kcdc's confirmed cases' travel route tracking system linked with smart city's data server self-report their health conditions, such as fever, cough, sore throat, and dyspnea symptoms and the report results are automatically sent to the kcdc twice a day. in addition, the location information of individuals in self-quarantine is automatically reported to the kcdc in real time. when the individual leaves his/her designated quarantine location, an alarm notification is sent to both the quarantined individual and the kcdc. then, the kcdc official who is responsible for the location immediately takes necessary actions (cdschq a; the government of the republic of korea ). health code is a contact tracing app with the largest user base in china since early february (mozur et al. ) . the app displays a green, yellow, or red qr code according to each user's health status thereby acting as a pass permit. users with a green code are allowed to visit others, but those with a yellow code should undergo self-quarantine for days, and those with a red code must self-isolate for days. for digital contact tracing, south korea uses mobile phone location tracking (location information at the communication base station), credit card usage details, and cctv records (cdschq a). china uses mobile phone location tracking, facial recognition, cctv, drones, and qr codes (gan and culver ) . users of these contact tracing apps report significant stress over the invasive surveillance functions that they are required to abide by (davidson ; mozur et al. ). coping theory explains individuals' conscious or unconscious endeavors to solve problems and reduce stress. lazarus and folkman ( , p. ) defined coping as "constantly changing cognitive and behavioral efforts to manage specific external and/or internal demands that are appraised as taxing or exceeding the resources of the person." in other words, it is the process of actively adapting to events that are happening and may happen in the future. coping theory has been applied in the fields of psychology, sociology, medicine, and social welfare. coping theory is also used in ict research. beaudry and pinsonneault ( ) proposed the coping model of user adaptation (cmua), which is a model describing users' adaptation to icts based on coping theory. cmua includes the process that users experience when using a new ict, which consists of awareness, appraisal, adaptation, and outcome (beaudry and pinsonneault ) . while contact tracing apps can be an effective covid- control system, the app as a new ict is causing users stress because they must unwillingly adapt and accept the novel invasive surveillant technology. the european parliament (ep) insists on preventing the abuse of personal information by legislating that it is not stored in a central database, and applying sunset clauses. the ep recommends that personal information is deleted as soon as covid- is no longer a threat, and a decentralized system applying ble technology (ep ). in the cases of south korea and china, user stress may be higher than in europe since these two countries have adopted the centralized system for their contact tracing apps. in the technological diffusion approach, new icts become standard means by which individuals or organizations undertake the processes of initiation, adoption, adaptation, acceptance, routinization, and infusion (cooper and zumd ; jones et al. ; zmud and apple ) . from an organizational perspective, zmud and apple ( ) defined infusion as "the extent to which the full potential of the innovation has been embedded within an organization's operational or managerial work systems." jones et al. ( ) defined infusion at the individual level as "the extent to which a person uses technology to its fullest extent to enhance his or her productivity." in other words, infusion is the process through which technology is accepted by individuals or organizations and used so that its full potential functions are realized. contact tracing apps such as self-quarantine safety protection app and health code app are actually initiated and accepted by users in mandatory settings for controlling covid- pandemic. however, the process of adaptation to infusion partially depends on users' emotion and ability because a variety of factors such as their stress and attitude affect adaptive efforts. mason ( ) categorized ethical issues of information age and among them are ( ) accuracy and ( ) privacy. accuracy is associated with the authenticity, fidelity, and precision of the information (mason ). privacy asks the fundamental question of where the borderline is between the information that should or should not be shared with others (mason ). concerns about inaccuracy and privacy infringement are raised with regards to contact tracing apps to control the spread of covid- (davidson ; mozur et al. ) . health code users in china reported concerns about the lack of transparency related to the app's operations, scope of data storage, inability to change an erroneous "red" code, excessive dependence on the internet, and reliance on private companies such as alipay and wechat monitoring their travel routes (davidson ; mozur et al. ) . as previously reviewed, users of contact tracing apps may experience stress mainly due to concerns over the following two issues: being mistakenly subjected to self-quarantine due to incorrect data input or technical errors in the contact tracing app; privacy infringement due to the system storing their personal information in the centralized system. in this context, this study proposes accuracy and privacy concerns as the main causes of user stress for contact tracing apps. users who perceive stress when using contact tracing the apps appraise each stressful situation as a threat or challenge (fadel and brown ; lazarus and folkman ) . based on the above discussion on concerns and challenge appraisal, this study proposes the following hypotheses: hypothesis accuracy concerns about contact tracing apps affect the challenge appraisal. hypothesis privacy concerns about contact tracing apps affect the challenge appraisal. in the appraisal stage of contact tracing apps, users assess whether contact tracing apps are an opportunity to prevent the spread of covid- or a threat to their individual liberty. in the cases of the self-quarantine safety protection app in south korea and health code app in china, users cannot refuse their use since they are mandatory. thus, users try to appraise the perceived consequences of the apps as a new opportunity and then undertake adaptation efforts to satisfy these expected benefits, which were termed as emotion-focused behaviors by beaudry and pinsonneault ( ) . fadel ( b) conducted an empirical study using a survey of electronic medical systems at university health departments to validate the cmca (beaudry and pinsonneault, ) . according to fadel ( b)'s study, appraisal of electronic medical systems as a challenge resulted in increased engagement in adaptation behaviors. in the similar sense, marakhimov and joo ( ) reported positive relationship between users' challenge appraisal of wearable devices and their emotion-focused coping behaviors toward the wearable devices (marakhimov and joo ). moreover, according to a study by joo ( ) regarding infusion of smart grid technology, challenge appraisal of technology significantly influences positive reappraisal. positive reappraisal as a kind of emotion-focused coping behavior refers to efforts to create or ascribe positive meaning to the technology. thus, the present study posits the following hypothesis: hypothesis challenge appraisal of contact tracing apps affects the emotionfocused coping behavior. there have been a few studies on the relationship between emotion-focused coping behaviors and infusion of information systems. in the individual level, emotionfocused coping behaviors positively influence work efficiency and effectiveness (beaudry and pinsonneault ) . emotion-focused coping behaviors are significantly associate with infusion of information systems at the individual level (fadel a) . joo ( ) reported that users of smart grid technology achieved the fullest potential of the smart system by utilizing positive reappraisal based on emotionfocused coping behaviors. thus, the following hypothesis is proposed: hypothesis emotion-focused coping behavior toward contact tracing apps affects user infusion. figure shows the relationships among accuracy and privacy concerns as factors affecting stress, challenge appraisal, emotion-focused coping behavior, and infusion as a structural equations model. an individual that feels stress due to concerns about contact tracing apps appraise the stressful situation (fadel and brown ; lazarus and folkman ) . when users experience a new it service, the new features constitute a challenge that they must evaluate (fadel and brown ) . during challenge appraisal, users of apps undertake emotion-focused coping behavior in an attempt to identify positives and strengths or to avoid/tolerate negative aspects or risks. through emotion-focused coping behavior, users of contact tracing apps adapt to attain the full potential benefits of the app. the research model in fig. is based on the stress-coping-adaptation model of lazarus and folkman ( ) , cmua of beaudry and pinsonneault ( ) , and stress-coping model of joo ( ). table shows measurement items for the five constructs in the proposed research model. the measurement items were modified and adopted from the studies conducted by fadel ( b) , marakhimov and joo ( ) , and joo ( ) to fit the purpose of the current study. each of the questions for the five constructs was measured on a five-point likert scale. the questionnaire was developed in korean, and three graduate students who were bilingual in korean and chinese translated the questionnaire into chinese and mutually reviewed the translations. finally, editing of the translation was commissioned to an agency specializing in korean and chinese. the survey was conducted targeting health code app users in china. although both self-quarantine safety protection app of south korea and health code app of china are good samples of centralized contact tracing apps, korean sample is inefficient for the data collection purpose of the research. as kcdc requires those who are infected or had contacted the infected to install self-quarantine safety protection app, only kcdc has the full list of those who have installed the app. however, health code of china is required for all its residents regardless accuracy concern the degree of concern about the possibility of errors in data input or unintended data usage on the contact tracing app i am concerned about getting red or yellow code by mistakenly inputting health status in the health code (health qr code) app developed i am concerned about getting red or yellow code due to operational error by the health code app service provider i am concerned about being penalized for wrongful data entry even though i enter accurate health status in the health code app the degree of concern about the possibility of abuse of personal information of contact tracing app users and consequential risks i am concerned about the possibility of my health in order to reach respondents with experiences using the app, the current study employed snowball sampling using wechat. excluding missing data, error responses, and inadequate answers, a total of valid responses were used for the analyses. characteristics of the samples are organized in table . male respondents outnumbered female respondents, as the percentage of male respondents was %. the proportion of respondents in their s and s was high at %, and % of respondents have been using the health code more than two months. alipay's app was shown to be the most widely used, followed by wechat and local government apps in order of precedence. the reliability, validity, and research hypothesis of the research model were tested using smart pls (version . . ). common method bias (cmb) may occur in cases where independent and dependent variables are measured in the same way during data collection (kock ) . harman single factor tests and variance inflation factor (vif) were used to check cmb. in the exploratory factor analysis of the harman single factor test, it is unlikely that cmb is present when the total variance of the unrotated first factor is less than % (podsakoff et al. ) . in the case of this study, since the total variance of the first factor was . %, cmb was determined as being unlikely. in a structural equation model, cmb may exist when the vif of a potential variable is . or higher (kock ) . in the present structural equation model, vifs of all potential variables were found to be between . and . , which demonstrates that the possibility of cmb is very low. cronbach's alpha, an indicator of internal consistency of variables, was below the standard of . (hair et al. ) in the challenge appraisal and emotion-focused coping behavior but was found to be reliable at a significance level of . as a result of instances of bootstrapping. therefore, there is no conflict regarding the reliability of the variables from the perspective of internal consistency (table ) . composite reliability (cr) and average variance extracted (ave) are used for the evaluation of convergent validity. as shown in table , the cr values of all variables at least satisfied the reference value of . and the ave values exceeded the reference value of . (fornell and larcker ) . therefore, each variable in this research model shows convergent validity. variables have discriminant validity when the square root of ave is greater than the correlation coefficients of the relevant variables (fornell and larcker ) . in table , the value of the diagonal column is the square root of the ave, and since it is larger than the correlation coefficients of the individual variables, the variables have discriminant validity. in general, there is no multicollinearity, which explains correlations between independent variables, when the vif is below the reference value of . (hair et al. ) . since all vifs were found to be are . or less, as shown in table , multicollinearity is unlikely to exist. the standardized root mean square residual (srmr) is used for the goodnessof-fit of the structural equations model using pls (garson ) . the goodnessof-fit is regarded to be high when the srmr is not greater than the reference value of . (hu and bentler ). the srmr of this research model was shown to be . , which is not too beyond the standard. path coefficients are used to test research hypotheses using smartpls. table shows the results of test of the research hypotheses. the hypothesis (h ) that accuracy concern for the contact tracing app affects the challenge appraisal was not supported. the hypothesis (h ) that privacy concern affects the challenge appraisal was supported at a significance level of . . the two hypotheses (h and h ) that challenge appraisal affects emotional coping behavior and that emotion-focused coping behavior affects infusion were supported at a significance level of . , respectively. table shows the results of the path analysis. the path for challenge appraisal, emotion-focused coping behavior, and infusion (ca → ec → in) demonstrates the significant impact of challenge appraisal on infusion. on the other hand, the path for privacy concern, challenge appraisal, emotion-focused coping behavior, and infusion (pc → ca → ec → in) shows the significant negative impact of privacy concern on infusion. eventually, if concerns about privacy infringement are resolved, app users can more actively engage with apps and maximize their potential benefits through emotion-focused coping behavior (table ) . the r-squared values of emotion-focused coping behavior (ec) and infusion (in) were shown to be satisfactory at . and . , respectively (garson ). in particular, the emotion-focused coping behavior of users of contact tracing apps accounted for . % of the infusion. according to the results of the current study conducted on mandatory centralized contact tracing app users, accuracy concerns about the apps did not significantly affect challenge appraisal. on the other hand, concerns about privacy infringement by contact tracing apps had significant negative effects on challenge appraisal. in addition, challenge appraisal had positive effects on emotion-focused coping behavior, through which app users effectively transitioned to the stage of infusion to maximize the potential benefits of the app. with regard to research hypothesis h , users who appraise contact tracing apps in terms of a challenges more actively conducted emotion-focused coping behavior when using the app. that is, users who appraise contact tracing app as providing a new opportunity to prevent and end the spread of covid- showed efforts to highlight and magnify the strengths and benefits. regarding to research hypothesis h , users attempted to enhance the app's strengths and benefits through emotionfocused coping behaviors, even if they recognized negative aspects in the early stages of use. in a study conducted by marakhimov and joo ( ) on users of wearable devices for healthcare, challenge appraisal was shown to positively affect the extended usage of the wearable devices through emotion-focused coping behavior. in a study conducted by joo ( ) on users of smart grid technologies, the more that users engaged in challenge appraisal of the technology, the more that they conducted emotion-focused coping behavior to actively maximize the potential of smart grid technologies. these studies support the results of the current study on the significant relationships among challenge appraisal, emotion-focused coping behavior, and infusion. with regard to research hypothesis h , the more concerned users are about privacy infringement with regard to the contact tracing app, the less users conduct challenge appraisal of the app. therefore, reducing concerns about privacy infringement may help users to reframe issues related to the app as instead challenges to overcome as well as recognize the strengths and benefits of the app. a previous study also reported that users' concerns about inputting personal health information into the wearable devices had a negative impact on users' challenge appraisal of the devises (marakhimov and joo ). with regard to research hypothesis h , concerns about problems related to inaccuracy, such as input errors or incorrect results, did not significantly affect app users' challenge appraisal. the reason why research hypothesis h was not supported is related to the sociopolitical and cultural systems of china. although private companies and local governments provide health code services, in reality, the central government forces people to mandatorily install and use the app. users of health code apps have a strong belief that they should trust and conform to government orders during these extraordinary circumstances of the pandemic. in such crisis, users tend not to think about errors in data operations or even be skeptical about the possibility of wrongful government operations. in addition, an online discussion was conducted with five graduate students in china who responded to the questionnaire in order to determine why research hypothesis h was not supported. three out of the five students argued that accuracy concerns about incorrect or wrongful information usage did not affect their challenge appraisal of the apps, and they further stated that the reason is that most chinese people trust app services in which the government is involved. given the results of the path analysis, relieving concerns about privacy infringement regarding the use of apps will enable users to realize the full potential benefits of contact tracing apps and make a greater contribution to preventing the spread of covid- . therefore, if the distributed system is used for contract tracing apps rather than the current centralized system, more effective covid- control can be expected. in collaboration with each other, google and apple have decided to provide bluetooth-based distributed contact tracing technology to all quarantine authorities (dumbrava ; apple ) . quarantine authorities in each country will be able to use these open apis (application programming interfaces) to develop customized contact tracing apps to reduce the concerns about privacy infringement to some extent. however, even for the distributed system, users' active participation and trust in operating authorities are paramount. in cases in which the centralized contact tracing system is used, it is necessary to be transparent and disclose how personal information will be safely and expediently deleted when the treat of covid- is over. south korea also decided to introduce an electronic entry and exit registration (qr code) system (called korea internet-pass) for facilities at risk of mass infection from june (cdschq b). users who visit designated facilities must present a personalized encrypted one-time qr code to the facility manager. the facility manager then scans the user's qr code and automatically transmits it to the social security information service (ssis), which is a public institution. the ssis manages facility information and qr code visit records, whereas the qr code-issuing company manages personal information such as the name and phone number of the person to whom the qr code has been issued. when a confirmed case of covid- occurs and the kcdc needs information, the quarantine authority can request information from the ssis, which keeps records of visits to the facility, and the qr code-issuing company, which keeps personal identifying information, to find out who (names and contact information) visited where and when. by separating and encrypting visit records and personal information before storage, only quarantine authorities are enabled to view personal information as a preventative measure to reduce privacy infringement. in addition, the electronic entry and exit registrations are stored only for four weeks, after which the facility visit records are automatically deleted. however, since this is also a centralized system, personal information is not protected if the government forcibly links related systems to track information. therefore, trust and transparency of the government operations are important. based on the coping theory, this study proposed a research model that shows the path from contact tracing app users' stress to their full utilization of the app. this research model can be applied to various fields of technology diffusion and expands the scope of coping theory applications. to date, no vaccine or treatment for covid- has been developed. the findings of this study can be used as a guide to maximize the potential benefits of contact tracing apps with the goal of preventing the spread of covid- . quarantine authorities in each country can improve the utilization of contact tracing apps by reducing the possibility of privacy infringement through establishing transparency and trust. even once a vaccine or treatment for covid- is developed, the findings of this study can be used to prevent the spread of future infectious diseases. given the findings of this study, the distributed system may be more effective than the centralized system for adoption of contact tracing apps, as the distributed system can relieve concerns about privacy infringement. therefore, the findings of this study can provide insights to quarantine authorities or app developers who want to deeply understand the tensions that arise when applying icts to prevent spread of infectious diseases and find innovative solutions. this study verifies the important role of icts in this current pandemic climate of covid- and contributes to maximization of the potential benefits of contact tracing apps. in this context, specific and unique implications of the current study are as follows: first, the current study shows that users are less likely to appraise the strengths and benefits of the app in terms of challenging opportunities as they have more privacy concerns over contact tracing apps. thus, those countries which utilize mandatory centralized contact tracing apps should implement policies that place more emphasis on relieving user's privacy concerns. second, the authorities should offer promotions and training programs that elucidate benefits and strengths of contact tracing apps in controlling the spread of covid- . for example, untraceable covid- cases had increased from . % (may ) to % (august ) in south korea where only infected or those who contacted the infected are required to install the app. on the other hand, the chinese case shows significantly lower rate of untraceable cases as the chinese government requires all residents, regardless of their infection status, to install the contract tracing app. such case can serve as a circumstantial evidence for the strength of contact tracing apps for ending the spread of coivd- . therefore, south korean government should also implement internet-pass qr code not only in facilities at risk of mass infection but also more universally as the case of china. another possible solution, if there is a high public concern for such mandatory installation of centralized apps, decentralized apps could be required for all residents of south korea. finally, authorities and social activities should focus on users to have positive conviction toward contact tracing apps in order for users to understand the strength of contact tracing apps and take benefits of the apps. this study has the following limitations. first, this study did not account for the differences in values as well as cultural and political aspects of the countries in which users of contact tracing apps live. second, the survey conducted on chinese app users cannot be said to represent all contact tracing apps. third, an actual comparison study between the central and distributed systems regarding users' concerns is recommended. privacy-preserving contact tracing understanding user responses to information technology: a coping model of user adaptation how to protect yourself & others guide on the installation of self-quarantine safety protection app mandatory introduction of a new qr code in high-risk entertainment facilities for korea internet-pass information technology implementation research: a technological diffusion approach china's coronavirus health code apps raise concerns over privacy tracking mobile devices to fight coronavirus covid- tracing apps: ensuring privacy and data protection the role of appraisal in adapting to information systems user adaptation and infusion of information systems information systems appraisal and coping: the role of user perceptions quantifying sars-cov- transmission suggests epidemic control with digital contact tracing evaluating structural equation models with unobservable variables and measurement error surveillance technology will only get more intense after covid china is fighting the coronavirus with a digital qr code. here's how it works partial least squares: regression & structural equation models thousand oaks hu l, bentler pm ( ) cutoff criteria for fit indexes in covariance structure analysis: conventional criteria versus new alternatives factors leading to sales force automation use: a longitudinal analysis infusion process of smart grid-related technology based on coping theory coronavirus: german contact-tracing app takes different path to nhs common method bias in pls-sem: a full collinearity assessment approach technostress: a research about computer self-efficacy, internet attitude and computer anxiety consumer adaptation and infusion of wearable devices for healthcare four ethical issues of the information age a ( ) in coronavirus fight, china gives citizens a color code development and utilization of a rapid and accurate epidemic investigation support system for covid- common method biases in behavioral research: a critical review of the literature and recommended remedies covid- contact tracing apps are coming to a phone near you. how will we know whether they work? the government of the republic of korea ( ) flattening the curve on covid- : how korea responded to a pandemic using ict covid- : guide on home-based care, screening & isolation ward set up the global impact of covid- and strategies for mitigation and suppression measuring technology incorporation/infusion acknowledgement this work was supported by the dongguk university research fund. this paper was written as part of konkuk university's research support program for its faculty on sabbatical leave in . conflict of interest the authors declare that they have no conflicts of interest. key: cord- -sp o h authors: raskar, ramesh; nadeau, greg; werner, john; barbar, rachel; mehra, ashley; harp, gabriel; leopoldseder, markus; wilson, bryan; flakoll, derrick; vepakomma, praneeth; pahwa, deepti; beaudry, robson; flores, emelin; popielarz, maciej; bhatia, akanksha; nuzzo, andrea; gee, matt; summet, jay; surati, rajeev; khastgir, bikram; benedetti, francesco maria; vilcans, kristen; leis, sienna; louisy, khahlil title: covid- contact-tracing mobile apps: evaluation and assessment for decision makers date: - - journal: nan doi: nan sha: doc_id: cord_uid: sp o h a number of groups, from governments to non-profits, have quickly acted to innovate the contact-tracing process: they are designing, building, and launching contact-tracing apps in response to the covid- crisis. a diverse range of approaches exist, creating challenging choices for officials looking to implement contact-tracing technology in their community and raising concerns about these choices among citizens asked to participate in contact tracing. we are frequently asked how to evaluate and differentiate between the options for contact-tracing applications. here, we share the questions we ask about app features and plans when reviewing the many contact-tracing apps appearing on the global stage. more than , deaths are now attributed to the global covid- pandemic. many thousands more lives are expected to be lost before we have brought the disease under control and are capable of managing future spikes in the number of cases. in an effort to both slow and stop the disease, communities across the world have halted everyday life, requesting or requiring their residents to close non-essential businesses, stop going to school, and stay home. digital initiatives hope to support safe and wellconsidered approaches to the reopening of our societies while simultaneously reducing the human loss of life by giving frontline officials modern tools with which to control this pandemic. one particular set of modern digital tools aims to upgrade contact-tracing capacity, typically a lengthy and laborious process. in addition to increasing the speed with which contact-tracers can reach those who have been exposed to the disease, these tools can increase the accuracy of contact tracing. however, many first-generation digital contact-tracing tools have paved the way for a post-pandemic surveillance state and the mistreatment of private, personal information. privacy must remain at the forefront of the global response, lest short-term pandemic interventions enable long-term surveillance and abuse. the design and development of the next generation of contact-tracing tools offers an opportunity to sharply pivot to solutions using privacy-first principles and collaborative, open-source designs. these tools present an opportunity to save lives by flattening the curve of the pandemic and to provide economic relief without allowing privacy infringements now or in the future. covid- virus transmission occurs for several days before a person shows any symptoms. during this time, a person going about their daily life may interact with, and possibly pass the infection to, as many as a thousand people. without knowing they are infected, an individual who has only mild symptoms or is asymptomatic may continue to interact with others, further spreading the virus. this creates an exponential rise in infections. stopping the spread of covid- with pharmaceutical treatments and vaccines remains at least - months away from widespread availability. therefore, public health countermeasures, such as social distancing, offer the only possibility of stopping virus proliferation in the near future. when applied broadly, such measures disrupt every aspect of society and risk economic collapse. already, unemployment rates have skyrocketed, tenants are struggling to pay rent, and critical supply chains, including the food supply chain, have been interrupted. the longer strict social distancing measures remain in place, the more severe the consequences for economies and societies will be. however, if social distancing measures are lifted too quickly, the virus will spread once again, claiming many additional lives. the contact-tracing process evaluates the recent location history and social connections of those who become infected and notifies the people they have interacted with of their exposure to the virus. in this way, contacttracing methods allow targeted measures (e.g., quarantining, virus testing) to be applied only to exposed individuals. traditionally, public health officials perform contact tracing manually, by interviewing patients diagnosed with a disease about their activity over the past days or weeks. then, officials reach out to people who crossed paths with the patient during the time the patient was contagious and recommend targeted interventions to prevent further spread of the disease. widespread, rapid transmission of a virus by respiratory droplets, as in the case of covid- , challenges the practicality of the traditional contact-tracing process. manual tracing is resource intensive, is time consuming, and will, at best, be limited to contacts within the social circles of the infected-and thus cannot trace strangers effectively. furthermore, the patient being interviewed is often extremely ill and at risk for memory errors during the interview. digital contact-tracing tools may help mitigate these challenges. today, almost half of the world's population carries a device, such as a smartphone, capable of gps tracking and bluetooth communication with nearby devices. each device is able to create a location trail-a timestamped log of the locations of an individual, as well as a list of anonymous id tokens that are collected when the device user crosses near another device. by comparing the device users' location trails or the anonymous id tokens they have collected with those from people who have covid- , one can identify others who have been near the person who is infected; this facilitates contact tracing in a more accurate and timely manner than the traditional manual approach. several pilot programs, particularly in china and south korea, have demonstrated the technical feasibility of contact-tracing applications as tools to help contain the covid- outbreak within a large population. however, these programs also highlight the very real risks that exist with the use of such technologies. a location trail and list of nearby device ids contains highly sensitive, private information about a person: everything from where they live and work and which businesses they support, to which friends and family members they visit. location data can be used to identify people who are infected and might then be targeted by their community. for example, data sent out by the south korean government to inform residents about the movements of persons recently diagnosed with covid- sparked speculations about the individuals' personal lives, from rumors of plastic surgery to infidelity and prostitution. more frightening still, enabling access to a person's location data by a third party, particularly a government, opens a path to potentially unrestrained state surveillance. in china, users suspect that an app developed to help citizens identify symptoms and their risk of carrying a pathogen was used to spy on them and share personal data with the police. care must be taken in the design of such apps. a number of groups, from governments to non-profits, have quickly acted to innovate the contact-tracing process: they are designing, building, and launching contact-tracing apps in response to the covid- crisis. a diverse range of approaches exist, creating challenging choices for officials looking to implement contact-tracing technology in their community and raising concerns about these choices among citizens asked to participate in contact tracing. we are frequently asked how to evaluate and differentiate between the options for contact-tracing applications. here, we share the questions we ask about app features and plans when reviewing the many contact-tracing apps appearing on the global stage. we are asking an open-source approach lets programmers and other experts outside the app development team review the code for a project. these outside programmers can make improvements, copy the code, or use it to create something entirely new. open source offers a layer of trustworthiness. because the code is publicly available, it can be reviewed by experts around the world to confirm it works the way the development team says it should. there are, at times, valid reasons to not use an open-source approach, such as when a business is seeking to develop a proprietary technology. during the covid- crisis, we believe that open-source projects promote collaboration and foster community. contact-tracing apps require the use of a data source to infer contact between two people: two of the most useful data sources are gps location data and bluetooth broadcasting. gps-based apps create a "location trail" for each user by recording their time-stamped gps location. if a person catches covid- , they can share their location trail with the responsible authority-the health worker, public health official, government official, or app creator. the authority then releases some or all of the location trail for other users to compare to. in some applications, the person who is infected might be able to directly share their location trail with other users. other apps rely on bluetooth to determine who the person who is infected has crossed paths. such apps create a unique identifier, a number or token, which the app broadcasts to nearby devices. the user's phone then records the identifiers of other phones it has been near. if a person becomes infected, their unique identifiers can be compared to those stored by other users to determine who the infected person has crossed paths with. in some cases, such as the singapore tracetogether app, the central authority stores user information and can determine the user's phone number and identity from an identifier. in others, such as covid watch and coepi, the identifiers provided by the person who declares themselves to be infected cannot be used by the central authority to determine the person's real world identity. both approaches offer distinct advantages and challenges: gps-based approach • allows for estimation of exposure related to surface transmission of disease. unlike bluetooth, gps-based systems can notify users if they were in a location shortly after a person infected with covid- , when the chance for exposure to the virus through commonly touched surfaces is high. • enables users to import historical data. other applications on the users' phones, such as google maps, are already collecting the potential user's location histories before they install the contacttracing app. when users import these historical data, the app can alert the user to potential exposures from their location history, even before they downloaded the app. • provides redacted, anonymized gps data to help public health officials follow the spread of disease within a community. • is able to record the user's location history using a small amount of data, making scaling and implementation in regions with high data costs more likely. bluetooth-based approach • uses signal strength, which is reduced by walls and other barriers, to estimate the distance between users. in some places, such as a large, multi-floor building, this estimate more accurately reflects the chance of exposure to disease than a gps-based approach. • uses time-range-dependent, randomly generated numbers as ids to ideally achieve relative anonymity. • requires the use of a compatible app by other users to record possible exposures. if an app is not widely adopted, the potential utility is limited. • no potential to collect historical data from before the user downloaded the app. in the near-future, some solutions, including covid safe paths, will integrate both approaches, allowing the user to harness the advantages of each while mitigating some challenges. both gps and bluetooth: aarogya setu (india) bluetooth: trace together (singapore) some bluetooth-based apps use a fixed identifier, meaning the unique number assigned to the device does not change and is permanently associated with the user. time-variable identifiers change on a set time interval, such as an hour, so each user is associated with many different identifiers. the use of time-variable identifiers adds a layer of privacy protection by making it difficult for a third party to track a particular phone over time based upon a single identifier. in a centralized version of contact tracing, location and contact data are collected and consolidated centrally by a single authority, often a government entity. china utilized a centralized approach with its app. other information about the user, such as mobile telecommunication service provider or payment data, may be collected and paired with the location data. the central authority identifies people who are infected, determines their contacts, and requests specific actions by those who may have been exposed to the virus. centralized systems create powerful tools for analysis and public health decision making. however, such systems also expose a person's data to a central authority, creating an opportunity to undermine the person's privacy. in a decentralized approach, the healthy user's data never goes to a central server. location data are stored and processed on the phone of the user. only the location data of people confirmed to be infected need to be shared. tools, such as redaction and blurring of the infected person's data, can be used to help preserve their privacy. an israeli app, track virus, is an example of a decentralized approach, as is covid safe paths. decentralized systems typically offer greater privacy protection and are, therefore, more in line with privacy requirements and regulations such as gdpr. some utility may be lost compared to centralized systems as collection and aggregation of large data sets from users can be used for beneficial public health research. however, as we consider the various approaches, the grave privacy risks associated with centralized systems far outweigh the limited additional benefits, leading us to highly value decentralized approaches. when checking if a healthy user has been exposed to covid- , contact-tracing apps may either push the healthy user's data to the authority (centralized processing) or pull a list of locations and/or contact ids of those who have been infected from the authority (decentralized processing). with a push, the healthy user's data is pushed (shared) off of the user's device and is compared by the authority to the data of people who have been infected. this exposes a large amount of data to the authority. in a pull model, an anonymized history of location data or identifiers from people who have been infected are pulled onto the healthy user's device so that the comparison can take place locally without compromising the privacy of healthy individuals. given what is known to date about person-to-person transmission of covid- , contact-tracing apps can properly assess users' potential exposure to the virus if they take four important factors into consideration: • the distance between the person who is infected and the user. • the length of time the person who is infected and the user occupied the same space. • how many days prior to becoming infected the person interacted with the user. • whether or not the user may have had contact with contaminated surfaces after interaction with the person who is infected. a location history must be collected from a person who has been diagnosed with covid- in order for contact tracing to occur. several approaches are being piloted. in general, these approaches fall into two categories: • an authority (public health official, healthcare provider, government official) collects the location history from the person who is infected and makes it available to users of the app. • the patient self-reports symptoms and directly shares their data with other users of the app. use of an authority offers the advantage of confirmation that the person has covid- . the overlap of symptoms between covid- and other common respiratory illnesses might cause someone to suspect they have covid- when they actually have the flu or a common cold. systems where people self-report themselves as infected pose the risk that people with symptoms, but without a confirmed diagnosis, share their location trail. self-reporting approaches are also at risk from bad actors who may misreport their status as infected in order to create chaos and fear. however, self-reporting systems have the advantage of fuller consent of the infected person as the person definitively decides to share their location trail without influence from an authority figure. when evaluating contact-tracing solutions, we seek to understand how data will be collected from the person who is infected and how the solution will confirm that the person truly has covid- . at the base of every contact-tracing app lies an algorithm that determines whether the app user has been exposed to people who are infected and might have an increased chance of being infected themselves. the algorithm integrates many factors, such as the distance between the users, the length of time the users were in the same location, or the amount of time between the contact and the start of symptoms. two apps with different algorithms will potentially give a different likelihood of exposure to the same user. understanding the algorithm used is necessary for public health officials and healthcare providers to provide appropriate guidance to users who receive an exposure notification. contact-tracing app developers must clearly communicate their algorithm with all stakeholders and failure to do so will be a significant red flag. location data may potentially be repurposed to achieve additional objectives beyond contact tracing. we believe these data should be used only for response to an ongoing pandemic and that other uses should be strictly forbidden. turning app data over to law enforcement or other non-health actors, such as commercial entities seeking to target ads to potential customers, threatens users' rights and privacy. critically, this undermines public trust. without trust, citizens will not adopt contact-tracing apps at a wide enough scale to effectively control the spread of the epidemic. therefore, access to location-tracking data should be tightly limited to specific public health initiatives working on pandemic response. users should be able to confirm how their data is used. promises by the app's developers to delete data are insufficient. users should be able to check exactly what location data has been collected and stored and to confirm that their data is no longer there after the deadline for deletion (the disease's incubation period, to days for coronavirus). apps must obtain users' unforced and informed consent for any disclosure of their data. recently, the a teleom austria group shared aggregated user location data from an app not regularly used for public health purposes with the austrian government's covid- emergency management team for reasons that were not initially specified. observers believe that a 's data was most likely being used to forecast disease spread or to monitor the population for large gatherings that might transmit the virus. however, the sharing of location data with government agencies for unspecified purposes attracted the criticism of privacy rights activists and created suspicions that weakened user trust, threatening long-term success. an opportunity for misuse and privacy violations arises whenever a third party, a government, a corporation, or any other entity is able to access the data of healthy users. a decentralized approach prevents privacy compromise for healthy users because they are doing all the calculations on their own phones. time-limited storage of location data also protects user privacy, such as only storing days of data with deletion of everything beyond this point. all contact-tracing app development teams should clearly articulate how they protect the privacy of all users -whether healthy or infected. as an example, a preliminary draft of the privacy principles of the covid safe paths team can be accessed in covid- contact tracing privacy principles. this overview of model privacy practices explains how the application embraces principles such as privacy by design, the fair information practice principles (fipps), and legal protection by design. historical location data and nearby device ids must be collected from a person who is infected to enable contact tracing. however, both the collection and release of that information have broad implications for the privacy rights of the individual. as the most vulnerable stakeholder, several efforts must be undertaken to protect, to the highest degree possible, the privacy of the person who is infected. app development teams may design for privacy by utilizing a variety of approaches: • providing users with the ability to correct incorrect information. • notifying individuals about what data is collected, how long it is stored, and who will have access to it during each stage of use. • enabling people to obtain access to information about potential exposures to covid- without requiring that they consent to share their data with other parties. • deleting user location data after it is no longer necessary to perform contact tracing. • alignment with the fair information practice principles. • using open-source software to foster trust in the app's privacy protection claims. • limiting the amount of data published publicly. • providing tools that allow the person who has been diagnosed and their healthcare providers to redact any sensitive locations, such as a home or workplace. • end-to-end encryption of location data before sensitive locations are redacted. • eliminating the risk of third-party access to information by enabling voluntary selfreporting by the person who is infected. • supporting strict regulation around access to and usage of the data by any entity that collects it, particularly governments. • obtaining targeted, affirmative, informed consent for each use of the person's data. • providing users with the ability to see how their data is being used and revoke consent for usage of their information. requiring people who are infected or potentially infected to track their movements and disclose their contacts achieves the highest degree of efficacy in contact tracing within a community. however, if residents cannot choose to at least selectively withhold their information, they may be stigmatized, persecuted, or exploited by malicious actors on the basis of their data. voluntary reporting respects users' rights to privacy and to informed consent. it encourages app developers to include safeguards that reduce the risk for abuse of sensitive data. however, when individuals who become infected refuse to share their contact-tracing data, the accuracy of contact tracing declines, potentially contributing to misinformation and a false sense of security. we believe that no one should be forced to relinquish highly sensitive personal data. we dislike solutions that require potential users to consent to share their data if they become infected in order to access information about whether or not they have crossed paths with someone who was infected. incentives such as those outlined in the following sections should be implemented to encourage users who become infected to share their data. people who are healthy should also proactively choose to use a contact-tracing app rather than being mandated to do so. potential users should be encouraged to do so by incentives, such as the opportunity to take control of their information to benefit their health, strong privacy protection policies, trust in the app's developers, clear communication, and informed consent. in order to roll out a contact-tracing app on a global scale, three groups must work together: a substantial team to create and promote the app; large, trusted institutions to support development and deployment of the app; and local, onthe-ground partners in the various communities in which the app is deployed. contact-tracing apps are tools, not complete solutions. disease containment utilizing these tools requires multidisciplinary collaborations across the technology, healthcare, public health, and government sectors. we are working hard to create these partnerships for covid safe paths and look for such partnerships in other apps we evaluate. among those partnerships teams should be seeking to build are: • cloud players (aws, azure, gcp, etc.) • mobile carriers and local telecommunications providers. • partnerships with health authorities; these partnerships are particularly important in light of app store requirements for all apps addressing the covid- pandemic to have the support of a health organization • government agencies • local public health workers and healthcare providers: contact-tracing apps will only succeed if those who crossed paths with someone who became infected can receive guidance and support from local providers on what steps to take to protect themselves and their families. • current contact tracers; integrating into the current contact-tracing protocol increases the effectiveness of a contact-tracing app within a community • non-profit organizations and academic institutions we see apps aiming to deploy at a variety of levels, from a single city to an entire nation to those aiming for a global reach. regardless of the level at which they are deployed, contact-tracing apps must be paired with existing infrastructure in order to support a successful containment strategy. public health officials and healthcare providers must be ready to answer user questions, offer testing, or provide advice about what to do if someone has been exposed to a person with covid- . the resources and support necessary to follow this advice must also be made available. we look for well-considered deployment strategies with aggressive outreach to local partners. for this reason, we are building not only a contact-tracing app, but also safe places, a web-based tool for public health officials working to contain the covid- pandemic. it is also worth noting that as global travel resumes, cross-communication between apps operating in different regions will be necessary to achieve global containment of covid- . we look for teams that are thinking ahead and building the technological foundation for this collaboration into their application. taking any software tool from idea to widespread solution requires the team to think creatively. contact-tracing apps gain value with each additional user. many approaches to encouraging user adoption exist, and good teams will use a variety of them. a few steps we encourage are: • fostering trust • developing key partnerships, including with community officials who can help drive local support for the solution • creating solutions that meet the needs of public health officials responding to the pandemic • focusing on the needs of the users • providing value to the user during a contact-tracing interview even if they choose not to download the app before they have been diagnosed with covid- contact-tracing apps need a strong value proposition for each stakeholder-the healthy user, the person who is infected, the public health worker responsible for contact tracing, the public health authority responsible for the community's response to the pandemic, and government officials tasked with coordinating the local or national response to covid- . as an example, the incentives for each stakeholder from the safe paths solution are presented here. offers an opportunity to take control and gain information. the user is able to make decisions about where they should be going and what activities are safe for their families and themselves. users are more confident and more informed about their actual risk of spreading the disease. gives the ability to quickly and accurately share location history with public health contact tracers. sharing their history offers an opportunity to help protect their community. gives immediate relief to contact tracers. provides a tool to more efficiently conduct interviews and gather information from patients. increases data accuracy over current methods (e.g., remembering). enables them to work with infected patients to quickly remove information that the patient asserts is personal, private, and/or confidential. allows more efficient and more accurate data collection and analysis about the spread of covid- within their jurisdiction. provides data to make better, more targeted recommendations for intervention to their community and to utilize limited testing resources most constructively. offers an opportunity to communicate a personalized risk profile to each citizen, answering the question "should i be concerned or not?" for every individual in their constituency and to closely monitor those who have the highest chance of experiencing complications from covid- . faster and more accurate contact tracing allows officials to catch up with the virus and more effectively deploy resources. rather than undifferentiated application of lockdown measures risking economic and subsequent financial collapse, officials are able to implement a differentiated approach with targeted measures as recommended by the who. the utmost care must be taken when notifying users of a potential exposure to covid- given the serious health, economic, and social consequences of a notification. during this stressful time, clear, easy-to-understand communication reduces the possibility for the user to misjudge their situation. high-quality translations should be available for all users. transparency about how the decision to notify the user was made helps the user and their public health officials make decisions about whether and which containment measures the user needs to undertake. notifications should evolve to reflect advances in the understanding of disease transmission as scientists around the world continue to clarify how covid- passes from person to person. contact-tracing apps, particularly those that allow individuals to self-report themselves as infected, must address the risk that some people will make fraudulent reports. in some instances, a false report may be done in good faith-the person truly suspects they have covid- , but they have not undergone definitive testing and actually have a different virus. in other cases, bad actors may report themselves as infected with covid- in order to create chaos. storing sensitive information in an anonymized, redacted, and aggregated manner minimizes the risk of data-tampering, yet it does not eliminate the chance for human error or malicious intervention. one approach to reducing fraud requires the diagnosis to be confirmed by a healthcare provider. however, creative teams may find other ways to prevent false reports of illness. with large-scale deployment, most apps will experience an occasional false report or find an error in an otherwise correct report. each app should develop a protocol for its response when an incorrect report is identified. easy-to-use tools should allow all involved in reporting to quickly mark and remove errors as soon as the false report is identified. most often, users should be notified of the change in their exposure history. while most apps aim to obscure the identity of the person who is infected, accidental release of information sufficient to identify the person can occur on rare occasions, similar to accidental release of protected health information. these low risks should be communicated to the users during the consent process. a process for quickly removing identifiable information from public access should be in place. notification of a potential exposure to covid- will be frightening to many, particularly those at increased risk for serious complications, and may lead to panic among users. large groups of people seeking medical evaluation or demanding testing could quickly overwhelm an already strained healthcare system. we have seen panic related to the pandemic lead to hoarding and vigilantism. conversely, users who are not notified of a potential exposure may assume they are at no risk to catch covid- and disregard critical social distancing and hygiene recommendations. any contact-tracing solution will need to provide users with accurate information to reduce the chance for panic or risky behavior. when reviewing an app, we look for the following: • clear, easy-to-understand, culturally appropriate communication with the user • engagement of epidemiologists, public health officials, and healthcare providers, both as core members of the decision-making team and as local partners within the community to which the app is deployed, in order to provide assessment and recommendations to people who may have been exposed to covid- • measures to prevent individuals from falsely reporting themselves infected and thoughtful consideration of how a person reported to be infected is confirmed to have covid- • use of both gps and bluetooth systems, utilizing the strengths of each technology • creative algorithms that reduce the chance that insignificant exposures are flagged contact-tracing apps should be viewed as a tool to be utilized by experts in infectious disease control. epidemiologists, public health officials, and healthcare providers must be core members of any team designing and implementing a contact-tracing app. we look to see that such experts are included as team members, mentors, and strategic partners. ideally, contact-tracing apps should fit into the current care pathway. one of the leaders in this area is tracetogether in singapore, which supports a contact-tracing process put in place long before the app was ready. tracetogether uses bluetooth to identify nearby phones with the app installed and tracks both proximity and timestamps. if a person is diagnosed with covid- , they can choose to allow the ministry of health to access their tracetogether data, which is then used by the manual contact-tracing team to alert those who may have been exposed. the manual contact-tracing team then alerts those who may have been exposed. we also aim to lead in this area with the development of covid safe places, a web-tool allowing public health officials to work more quickly, collect better data, and better respond to what is happening in their community. we are partnering with public health workers around the world to deploy covid safe places. the success of any contact-tracing program should be measured in lives saved. lives are saved both by a reduction in the spread of disease and by a reduction in the psychosocial and economic consequences of widespread quarantine actions. quantitative analysis of the effect of this new technology should be undertaken-not only to allow for further improvements during the current covid- pandemic, but also to better address the next outbreak of infectious disease. in addition to collecting real-world data about the impact of contact-tracing apps, teams should work to communicate their success to the public. if the apps are effective in helping to control the pandemic, the public may fail to notice the extent to which their use was critical to the community's ability to control the spread of disease. the covid- pandemic will not last forever. if we falter in our response and choose digital contact-tracing tools that compromise individual privacy for efficacy, the consequences will extend long after the last store has reopened and the last child has returned to school. we believe privacy does not have to be compromised in order to reduce new infections and slow the spread of disease. we are building covid safe paths with privacy protection at the forefront for this pandemic and the next. here, we have begun to detail the key questions that should be asked as we evaluate contact-tracing apps developed and deployed against the covid- pandemic. we plan to continue this discussion and are committed to serving as a resource for countries, states, cities, and individuals throughout the world. we welcome additions to and modifications of this report and analysis. to submit a change please email info@pathcheck.org assessing disease exposure risk with location data: a proposal for cryptographic reservation of privacy how europe manages to keep a lid on coronavirus unemployment while it spikes in the u.s. the washington post privacy by design: the foundational principles. implementation and mapping of fair information practices covid- dashboard recommendation regarding the use of cloth face coverings singapore says it will make its contact tracing tech freely available to developers % can't pay the rent: 'it's only going to get worse fair information practice principles clever cryptography could protect privacy in covid- contact tracing apps covid- contact tracing privacy principles centre for the mathematical modelling of infectious diseases covid- working group coronavirus: the korean clusters. reuters graphics legal by design' or 'legal protection by design'? in law for computer scientists coronavirus disease vs. the flu the efficacy of contact tracing for the containment of the novel coronavirus (covid- ) more scary than coronavirus': south korea's health alerts expose private lives. the guardian the incubation period of coronavirus disease (covid- ) from publicly reported confirmed cases: estimation and application how the coronavirus is disrupting the global food supply in coronavirus fight, china gives citizens a color code, with red flags. the new york times coronavirus vaccine in months? experts urge reality check coronavirus has disrupted supply chains for nearly % of u austria: telco a gives government location data to test movement restrictions apps gone rogue: maintaining personal privacy in an epidemic. arxiv don't believe the covid- models: that's not what they're for people tattle on neighbors flouting covid- shutdown orders contact tracing covid- virtual press conference jobs carnage mounts: million file for unemployment in weeks. national public radio key: cord- -cl caw q authors: casagrande, marco; conti, mauro; losiouk, eleonora title: contact tracing made un-relay-able date: - - journal: nan doi: nan sha: doc_id: cord_uid: cl caw q automated contact tracing is a key solution to control the spread of airborne transmittable diseases: it traces contacts among individuals in order to alert people about their potential risk of being infected. the current sars-cov- pandemic put a heavy strain on the healthcare system of many countries. governments chose different approaches to face the spread of the virus and the contact tracing apps were considered the most effective ones. in particular, by leveraging on the bluetooth low-energy technology, mobile apps allow to achieve a privacy-preserving contact tracing of citizens. while researchers proposed several contact tracing approaches, each government developed its own national contact tracing app. in this paper, we demonstrate that many popular contact tracing apps (e.g., the ones promoted by the italian, french, swiss government) are vulnerable to relay attacks. through such attacks people might get misleadingly diagnosed as positive to sars-cov- , thus being enforced to quarantine and eventually leading to a breakdown of the healthcare system. to tackle this vulnerability, we propose a novel and lightweight solution that prevents relay attacks, while providing the same privacy-preserving features as the current approaches. to evaluate the feasibility of both the relay attack and our novel defence mechanism, we developed a proof of concept against the italian contact tracing app (i.e., immuni). the design of our defence allows it to be integrated into any contact tracing app. the sars-cov- pandemic caught governments and healthcare systems unprepared. one of the main issues of the virus concerns the speed of diffusion, which is very high, especially in comparison to the time required to find all the people that have been in contact with an infected person. thus, the very first strategy adopted by some governments was to put citizens in a strict lockdown. however, the lockdown was only a temporary solution and governments started looking for alternative approaches aimed at the containment of the virus, which means: (i) rapid identification of the infected individuals, with the consequent quarantine, (ii) identification of the people that have been close to an infected person in the previous days and weeks, (iii) decontamination of places where the infected individual has been. guaranteeing the containment is a difficult task, which might lead to errors, especially if performed manually, as it happened at the start of the virus spread . as a result, governments expressed their need for automatic contact tracing (act) solutions, which can help with better monitoring the spread of the virus if applied together with social distancing. several act solutions were proposed and adopted by different countries according to their economical, technological and cultural status. the most promising tool for collecting contact tracing data that concern the citizens was the smartphone, since most people own one. tracing citizens contacts through the smartphone location (i.e., gps signal, wi-fi routers, cellular networks) seemed a feasible solution [ ] , which, however, introduced significant privacy issues: citizens had to partially give up on their privacy to protect others. as illustrated in [ ] , such privacy issues could even lead to public identification of diagnosed patients or to mass surveillance. thus, researchers proposed an alternative approach based on the bluetooth low-energy (ble) technology, which is already available on most smartphones. ble allows two devices that are close to each other to exchange their identifiers, while requiring a limited amount of battery power. several communication protocols have been proposed to support act on mobile devices: bluetrace [ ] , robust and privacy-preserving proximity tracing (robert) [ ] and decentralized privacy-preserving proximity tracing (dp- t) [ ] . in addition, google and apple designed the google apple exposure notification (gaen) [ ] protocol. gaen enables the interoperability among devices running android and ios, by providing a common api in the underlying platform to be used by contact tracing apps. all the above-mentioned protocols follow the same workflow: each smartphone has its own pseudonym, shared with other smartphones when they get close to each other. after a while, the smartphone updates its pseudonym with a new one, which is seemingly independent. thus, each smartphone will soon have a database of the announced pseudonyms and a database of the received pseudonyms. when a person is found infected, all the smartphones that have been in contact with the smartphone of the infected person should be notified, also receiving a risk score. by proving the unlinkability of pseudonyms, the above approach guarantees a good level of privacy. despite its privacy-preserving nature, the ble-based act approach has several limitations which reduce the citizens consensus to adopt it. among such limitations, there is the resilience to security attacks. in particular, the absence of an authentication procedure in act apps paves the way for replay and relay attacks. in this paper, we focus on the vulnerability to relay attacks of act apps and we prove that many popular ones (e.g., those promoted by the italian, french, swiss government) are not able to defend against such attacks. to tackle this security issue, we designed a solution called actguard that effectively prevents relay attacks by acquiring the location data concerning contacts between pairs of people. in particular, actguard saves on the mobile device only a hash of the contact, generated by providing as the input of a hash function the pseudonyms of the two smartphones, the timestamp of the contact and the location of the contact. whenever a person is found infected, the hashes of the contacts over the previous days are shared with a remote server and locally downloaded by other smartphones. if the pseudonym of the infected person has been spoofed and used in a relay attack (i.e., being retransmitted by the attacker in a different location than the person real one), the hashes of the relay attack victims will mismatch the ones of the infected person due to the different location. this way, actguard will prevent the relay attack victims from receiving a false positive alert. moreover, by saving and sharing only the hashes of the contacts, actguard provides a privacy-preserving solution. to assess the feasibility of the relay attack against act apps and the reliability of actguard, we developed two proofs of concept focusing on immuni, the italian contact tracing app based on gaen. contributions. the contributions of the paper are as follows: • we analyze the design of popular act apps based on gaen and identified a vulnerability against relay attacks; • we design the first solution that is compliant with the current gaen internal architecture and prevents relay attacks, called actguard; • we implemented a proof of concept of a relay attack and of actguard against immuni, the italian act app; • we release a video demonstration of the attack and of the defence . organization the rest of the paper is organized as follows: section ii provides background knowledge about the ble technology, and the existing proximity tracing solutions for act apps; section iii introduces the system model and the threat model of the relay attack we designed against gaenbased act apps; section iv outlines the design of actguard; section v describes our implementation of the relay attack against immuni and our proof of concept of actguard, i.e., immuniguard; section vi illustrates the related work concerning security and privacy issues of act apps; finally, in section vii, we conclude the paper with a discussion about act apps. the purpose of this section is to illustrate the ble protocol (i.e., section ii-a) and the proximity tracing protocols (i.e., section ii-b). the description of the ble protocol focuses on the android operating system (os), since both the attack and the defence target this platform. bluetooth stack. the bluetooth stack involves four different layers: physical, link, middleware and application layer. the physical layer and the link layer include physical and hardware components, such as chips, that communicate with the os through the host controller interface (hci). the middleware layer includes the protocols implemented by the host, among which the logical link control and adaptation protocol (l cap), the radio frequency communication (rf-comm) and the session description protocol (sdp). l cap is responsible for the bluetooth data-flow control, rfcomm generates serial data stream, while sdp broadcasts the device services to other devices in order to establish a connection. finally, the application layer defines the different functionalities offered to users. from version . , the bluetooth specification includes also the ble version, supporting the transmission of discrete data to reduce power consumption. in ble, discrete data are stored in attributes, which access is regulated by the attribute protocol (att) and the generic attribute profile (gatt) protocols. ble devices expose their data through services, that include attributes and characteristics (i.e., special types of attributes). bluetooth protocol. the bluetooth protocol always involves two devices, i.e., a client and a server. usually, the server keeps running in discoverable mode until a client sends it an inquiry. as soon as the inquiry is received, the server sends its information to the client (i.e., device name, device class, list of services, technical information, a userfriendly name defined by the manufacturer). being aware of the server's services, the client attempts to read such data. however, the access is regulated by the set of permissions associated to the data: no permission (the server can access the client data); authentication required (the client starts the pairing procedure); authorization required (any implementation concerning the authorization is left to developers). the pairing procedure allows a device to authenticate a different, remote one, before connecting to it, and to share with it long-term keys that will encrypt the communication. during the authentication, the user is asked to verify the identity of the remote device. this process might go through an interaction with the remote device display or through the input of a value into the remote device, according to its features. alternatively, the pairing procedure might also happen through the just works mode, in case the remote device has neither a display nor an input. when the pairing is completed, the two devices go through the bonding procedure, during which they exchange the long-term keys. unless a device is manually reset or unpaired, it will keep the key stored and use it to encrypt the communication at the link layer. use of bluetooth in android. in android, applications willing to use the bluetooth communication channel have to rely on the android.bluetooth api package, that enables the interaction with the system process under /packages/apps/bluetooth. moreover, since bluetooth is classified as a protected resource, applications have to declare specific permissions, according to their purposes: • bluetooth, to support bluetooth communications, such as requesting and accepting connections. • bluetooth_admin, to discover nearby bluetooth devices and to change smartphone bluetooth setting. currently, this permission requires also location related permissions, such as android.permission.access_fine_location or access_coarse_location. • bluetooth_privileged, to avoid any user interaction with the device during the pairing procedure (it can be granted only to system applications with a signaturelevel permission). when an application is granted the required permissions, it can start communicating with external bluetooth devices. if an external device requires a pairing procedure, this is usually handled by a single application. however, the bonding information of the external device is saved in a shared location, that can be accessed by any application with bluetooth related permissions. as soon as it has the permissions, any application can communicate with an external bluetooth device, if this has been already paired with the smartphone. bluetrace [ ] , robert [ ] , dp- t [ ] and gaen [ ] are among the most well-known proximity tracing protocols worldwide. proximity tracing consists in registering any physical contact between individuals, and it is a core part of the contact tracing process. despite sharing the same technology (i.e., ble on mobile devices), such protocols differ in the role assigned to the centralized server, which is controlled by the national health authority. according to this, the protocols can be classified into two different approaches, both shown in fig. : centralized and decentralized. the general workflow of proximity tracing protocols can be summarized as follows: • set up and configuration: in the centralized approach, the app registers to the health authority server and it receives multiple ephemeral bluetooth identifier (ebid). on the contrary, the decentralized approach assumes that the app locally generates its own ebid. • contact between two individuals: when two app users meet each other, the respective apps share their own ebid. • announcement of a new positive: when an app user is found infected, the centralized approach requires the user to share with the health authority server all his own ebid and the ones of the people met. on the contrary, the decentralized approach requires the infected user to share only his own ebid remotely. • risk score calculation: in the centralized approach, the health authority server calculates the risk for all the ebid that have been in contact with the infected one and it sends the notifications accordingly. in the decentralized approach, the apps periodically download the ebid of the new infected people and locally calculate the risk score. we will now provide more details about each proximity tracing protocol. bluetrace. bluetrace [ ] is a proximity tracing protocol designed by the government digital services team at the government technology agency of singapore and it adheres to the centralized approach. it was implemented on the first national act app deployed in the world, i.e., tracetogether [ ] . in this case, the centralized approach was chosen for two reasons: ) it allows for a human-in-the-loop design during epidemiological surveillance, and ) it allows to better monitor the adoption and usage of the app, by logging daily requests to the server. during app registration, the centralized server assigns the user a unique user identifier (userid), linked to his phone number. then, the app generates multiple temporary identifier (tempid) from the userid, thus allowing the centralized server to trace back the userid from the tempid. the messages exchanged between apps as advertisement packets contain the following data: • tempid -they frequently rotate out to prevent third-party tracking; • device model -it improves the calculation of distance estimates; • organization code -it indicates the country and health authority; • bluetrace protocol version. when a user is found positive to sars-cov- , bluetrace allows him to share his contacts history with the centralized server, together with additional data (e.g., rssi, device model, timestamp). the server decrypts the tempid in order to retrieve the userid and the validity period. then, it finds the closest contacts according to time of exposure and distance. during the medical interview of infected users, proximity and duration filtering thresholds are fine-tuned, before any in-app exposure warning is sent. robert. pepp-pt [ ] is a non-profit organization, based in switzerland, with members across eight european countries. the objective of pepp-pt is to develop a privacypreserving digital proximity-tracing architecture. part of the pepp-pt proposal involved the design of a centralized contact tracing protocol, named robert [ ] . during app registration, the centralized server assigns to the user a permanent identifier (id), saved in its database, and several ebid, which are valid within a specific time window. apps implementing the robert protocol regularly broadcast packets containing their ebid, and store ebid received from other users. whenever a user is found positive to sars-cov- , robert requires him to upload the list of ebid collected from other users in the past weeks. the centralized server is able to trace back the id from uploaded ebid, flags them as exposed, calculates the risk score and warns app users accordingly. dp- t. dp- t [ ] is a european consortium of technologists, legal experts, engineers and epidemiologists with the primary objective of developing a proximity tracing protocol able to prevent mass surveillance. the dp- t protocol adheres to the decentralized approach and provides three slightly different designs: (i) a low-cost decentralized proximity tracing protocol (good privacy and very small bandwidth required); (ii) an unlinkable decentralized proximity tracing protocol (far better privacy than the first solution, but with an increased bandwidth requirement); (iii) a hybrid decentralized proximity tracing protocol (a combination of the two previous designs). from now on, we will implicitly refer to the hybrid decentralized proximity tracing design, as it is more wellrounded: it features a high level of protection against linking ebid to the real identity of individuals, while still retaining a low-cost philosophy. an app implementing dp- t does not communicate with the centralized server during the setup process. on the contrary, it locally generates a secret key (sk), used to generate several ebid, which are valid within a specific time window. apps implementing dp- t regularly broadcast packets containing their ebid, and store the ebid received from other users. whenever a user is found positive to sars-cov- , dp- t requires him to upload past sk, along with the associated time windows. the centralized server simply acts as a database, allowing other app users to connect and download the newly uploaded data. by knowing a sk and its time windows, apps can generate ebid and compare them to the ones they collected during their proximity tracing activity. the health risk calculation is performed locally by the app, and the health status is updated immediately. gaen. the result of the collaboration between google and apple is the gaen [ ] protocol, which aims at enabling ble interoperability between android and ios devices. gaen provides a set of application programming interface (api) that are restricted only to the developers authorized by their own government to release a national act app. gaen shares several design features with the dp- t hybrid decentralized proximity tracing and it adheres to the decentralized approach (for more details on the gaen protocol, please, refer to section iii). overview of act apps. as already mentioned, bluetrace, robert, dp- t and gaen are proximity tracing protocols, which can be implemented by any national act app. table i shows an overview of the act apps that we analyzed. during our selection, we aimed to include entries for each of the most popular proximity tracing protocols, to represent a wide variety of countries and to showcase the various combination of technologies and privacy approaches with respect to the following criteria: the adopted proximity tracing protocol, the technology used for performing the contact tracing, the set of personal data required to be shared by the app user and, finally, the resilience to replay and relay attacks. as shown in the table, almost all apps are vulnerable to relay attacks. this is the main motivation that encouraged us to provide a design for a defence mechanisms, that prevents such attacks, while being compliant to existing protocols (for more details on the act apps, please, refer to section vi). in this section, we illustrate the system model (i.e., section iii-a) and the threat model (i.e., section iii-b) we considered for designing a relay attack against act apps and for our defence mechanism, i.e., actguard. in particular, among the different proximity tracing protocols, we chose to focus on gaen-based contact tracing apps, as gaen is the most widely adopted solution in europe. most gaen-based apps merely act as an interface, that strongly relies on the underlying protocol. thus, our security analysis concerning the resilience to relay attack, as well as the design of actguard, can be applied to any gaen-based contact tracing app. here, we provide the details of the gaen protocol in terms of pseudonym generation and exchange, as well as announcement of a new positive person. pseudonym generation and exchange. once a gaenbased app is installed on a device, and the onboarding process is completed, it generates a -byte temporary exposure key (tek), which is valid for a single day and then replaced by a new one. as shown in figure , a gaen-based app generates a -byte rolling proximity identifier key (rpik) and an associated encrypted metadata key (aemk) from its current tek. finally, the app derives the rotating proximity identifier (rpi) from the rpik. since rpi are not linked to a specific person or device, they can be exchanged as pseudonyms during contact tracing. every time the smartphone ble mac randomized address changes, the current rpi needs to be updated accordingly, and a new one is derived again from the rpik, with a two-hour validity windows. advertisement packets are broadcasted by the app via ble, and stored locally by other nearby apps. announcement of a new positive person. when a new user is found positive to sars-cov- , he can choose to upload his recent tek (also referred as diagnosis key) to the centralized server. other apps periodically download the malicious ✗ in our threat model, as well as in the scenarios described in section iv, we assume to have the components shown in fig. : a set of three honest users, among which one is positive to sars-cov- ; two malicious attackers; a set of servers, which could be either the health authority one or the actguard one; a gaen-based app; the actguard app. our relay attack against gaen-based apps utilizes the threat model shown in fig. . it involves the actors detailed in table ii : three honest users (i.e., a, b and c), with an active gaen-based app on their smartphone; two adversaries (i.e., adv and adv ) without any gaen-based app on their smartphones. in this scenario, we assume that a and adv are positioned in place x, while b, c and adv are positioned in place y. in our threat model, we assume that adv stands near b and relays his advertising data to adv . in particular, adv aims at stealing the signal transmitted by b, who is going to be found positive in a near future. once acquired, adv shares such signal with adv , which starts advertising it in a different location, exposing the victims to an infected relay attack impact. relay attacks make act apps significantly less reliable, as they can inflate health risk scores by creating contacts that have never occurred. adversaries performing relay attacks can choose sensible locations: place y could be a location with a high exposure risk (i.e. a hospital, a quarantined city), whereas place x could be a location with many individuals coming and going (i.e. a train station, an airport). apart from inflating health risk scores, the attackers might also push false positive warnings to specific individuals, instead of targeting whole groups. possible attack scenarios could be the following ones: a student enforcing a professor to the self-quarantine, thus leading to the cancellation of an important exam; the trading of pseudonyms derived from infected people in the dark net; the real-time monitoring of a person movements by locating several devices in places often visited by the targeted victim. iv. actguard design act apps using ble are vulnerable to relay attacks by design: proximity tracing protocols depend on the close distance between two app users, but malicious attackers can intercept and relay ble signals. solutions that implement authentication mechanisms, or rely on location or timing data have limitations: integrating authentication mechanisms in the communication protocol introduces privacy issues; location data can be used to validate the distance between users, but it can not be shared with anyone else, since it would still cause privacy issues; timing data could be useful, but rpi have a two-hour validity window, which could be long enough for an attacker to execute a relay attack. we designed actguard, considering the following two objectives: ) enabling gaen-based apps to defend against relay attacks. ) guaranteeing the same privacy level as the current gaenbased apps. our intuition behind the design of actguard relies on the collection of location data related to a contact between two app users. more specifically, by saving the gps coordinates of a contact between two users, we can detect when rpi are transmitted at the same time in two different places. this method prevents the delivery of false positive alerts to victims that received relayed rpi. the design of actguard requires the app user to save the following data for each contact occurred between user a and user b: to avoid any inconsistency, rpi are ordered alphabetically. actguard saves the set of hashes associated to contacts involving its owner. if any user is found infected, he uploads his past hashes to a central server. other actguard users periodically download the hashes of the infected people and check them against the ones they saved locally. if all the information concerning a contact between two users (i.e., both rpi, location, time) is equal, then hash values will perfectly match and actguard will confirm that the contact occurred for real. if this does not happen, there are two possible reasons: ) the infected user app did not upload the hashes to the centralized server; ) the infected user uploaded the hashes to the centralized server, and a relay attack was performed. besides preventing relay attacks, actguard provides also a privacy-preserving solution, since each user only shares hashes. thus, the centralized server is able to infer none of the contact information as hash functions are not reversible by design. to better explain the actguard resilience to relay attacks, we consider our threat model, which is illustrated in fig. , and analyze several scenarios that differ according to which actors are equipped with actguard. in particular, we assume that: • a, b and c always use a gaen-based app. • adv and adv use neither a gaen-based app nor actguard. • a always uses actguard, since he is the victim and he has to defend against relay attacks. • b and c might not use actguard. with the above-mentioned assumptions, we identified the following four scenarios: • scenario -a, b and c all using actguard. • scenario -a and c use actguard, but b does not. • scenario -a and b use actguard, but c does not. • scenario -a uses actguard, but b and c do not. in the rest of the section, we will discuss scenario (i.e, section iv-a) and scenario (i.e, section iv-b). since c is not infected, thus not going to uploaded his contact information, his adoption of actguard does not affect the defence against the relay attack. thus, we do not discuss scenario . similarly, the focus of the scenario would be b not having actguard, which is already analyzed in scenario . in scenario , we assume that all honest users (i.e., a, b and c) are equipped with a gaen-based app and actguard, while the attackers (i.e., adv and adv ) have none of them. scenario is illustrated in fig. and the details concerning the different actors are summarized in table iii . positive diagnosis. we assume that b gets diagnosed as positive to sars-cov- and through his gaen-based app, he uploads his past tek to the centralized server. b and c had a real contact. thus, gaen-app(c) downloads tek(b), derives all possible rpi and finds a match with rpi(b,validitytime) saved in its local storage. consequently, c is notified about a potential health risk. in this case, gaen-app(c) is working as intended. at the same time, gaen-app(a) finds a match with the rpi(b,validitytime) saved in the local storage, thus leading to a potential health risk warning, as well. however, in reality, a only met adv , while he was impersonating b during the relay attack. in this case, gaen-app(a) is not able to discriminate between a real contact and a forged one. meanwhile, actguard(c) finds a correct match and it confirms the health warning from gaen-app(c). on the contrary, actguard(a) does not find a match in the hashes due to the different location of the contacts. consequently, actguard(a) does not confirm the health risk warning from gaen-app(a). thanks to actguard, the relay attack against a is detected and a false positive health risk warning is prevented. in scenario , we assume that the honest users a and c are equipped with a gaen-based app and actguard, that b is only a gaen-based app user and that the attackers (i.e., adv and adv ) have none of them. scenario is illustrated in fig. positive diagnosis. we assume that b gets diagnosed as positive to sars-cov- and through his gaen-based app he uploads his past tek to the centralized server. b and c had a real contact. thus, gaen-app(c) downloads tek(b), derives all possible rpi and finds a match with rpi(b,validitytime) saved in its local storage. consequently, c is notified about a potential health risk. in this case, gaen-app(c) is working as intended. at the same time, gaen-app(a) finds a match with the rpi(b,validitytime) saved in the local storage, thus leading to a potential health risk warning, as well. however, in reality, a only met adv , while he was impersonating b during the relay attack. in this case, gaen-app(a) is not able to discriminate between a real contact and a forged one. both actguard(a) and actguard(c) do not find a match in the hashes shared by b, since b does not use actguard and did not upload any hash. since actguard(a) is not able to confirm the health risk warning from gaen-app(a), it can not confirm if a relay attack actually happened. the scenario opens up two further situations: in this section, we describe the act app we chose as a case study to develop a proof of concept of the relay attack and of the actguard design. in particular, in section v-a, we provide an overview of immuni [ ] , the italian official act app. in section v-b, we describe the relay attack we developed against immuni and, finally, in section v-c, we illustrate the implementation details of immuniguard, which is an instance of actguard referring to the immuni use case. immuni a mobile contact tracing app based on the gaen protocol [ ] , thus adhering to the decentralized approach and relying on the ble technology for proximity tracing. onboarding. once installed, immuni detects the user language, checks for required updates and provides a basic explanation of how immuni works. then, immuni shows the privacy note and the user agreement, asking if the user is more than fourteen years old. at last, the user submits his province of residence and enables the required app permissions. immuni interacts with a back-end architecture involving the following services: • exposure ingestion service. • exposure reporting service. • backend one time password (otp) service. • analytics service. • app configuration service. exposure ingestion service. the exposure ingestion service provides a set of api for immuni apps to enable the upload of tek generated over the past days, when an app user is found infected and he is willing to share his keys. contextually, the infected user uploads the epidemiological information from the previous days. if any epidemiological information is indeed uploaded, the user province of domicile is uploaded, too. this process can only take place with an authorised otp from the backend otp service. the exposure ingestion service is also responsible for the periodical generation of tek chunks, to be published by the exposure reporting service. the tek chunks are assigned a unique incremental index and are immutable. the province of domicile and the epidemiological information are forwarded to the analytics service. exposure reporting service. the exposure reporting service makes the tek chunks created by the exposure ingestion service available to other apps. only tek chunks from the past days are made available. backend otp service. the backend otp service provides a set of api to the national healthcare service for authorising otp that can be used to upload data from immuni apps via the exposure ingestion service. immuni generates the otp, that the app user personally communicates to a healthcare operator. then, the healthcare operator inserts the otp into the italian health information system, registering it on the backend otp service. the otp automatically expires after a defined time period. analytics service. the analytics service provides a set of api for immuni apps to upload data without identifying users, both during regular operations and especially when a match is found between tek chunks and rpi. to ensure the proper functioning of the system, maximize the effectiveness of the exposure notifications and provide an optimal healthcare assistance to users, the following information is sent to the server: tek are required to allow other immuni users to calculate their risk of being positive to sars-cov- . whenever tek are uploaded, the following epidemiological information is uploaded, as well: • the day the exposure occurred. • the duration of the exposure. • the signal attenuation information used for estimating the distance between the two users' devices during the exposure. whenever an exposure detection has been completed, the following operational information is uploaded: • whether the device runs ios or android. • whether permission to leverage the gaen protocol is granted. • whether the device ble is enabled. • whether permission to send local notifications is granted. • whether the user was notified of a risky exposure after the last exposure detection (i.e., after the app has downloaded new tek from the server and detected if the user has been exposed to users positive to sars-cov- ). • the date on which the last risky exposure took place, if any. along with genuine analytics uploads, after every exposure detection event, immuni apps may perform dummy analytics uploads, indistinguishable from the genuine ones. app configuration service. the app configuration service updates the configuration settings every time the app starts a new background or foreground session. such settings can be used for tuning traffic-analysis mitigation measures and update weights used in the risk prediction model. here, we illustrate the workflow of the relay attack we designed against immuni. our scenario involves the following actors: • immuni app : this is the first victim of the relay attack, since its ble packet is sniffed and re-transmitted by the attacker. • malicious app : this is the first malicious actor, responsible for sniffing the ble packet, which will be retransmitted afterwards. • malicious database: this database is used by the two attackers to share the sniffed ble packet, even though they are located in different places. • malicious app : this is the second malicious actor, responsible for the re-transmission of the sniffed ble packet. • immuni app : this is the second victim of the relay attack, since it receives the sniffed ble packet. as shown in fig. , immuni app keeps advertising its ble packets, which contain its current rpi. malicious app is located nearby the immuni app , thus being able to intercept its advertising packets and send them to the malicious database. meanwhile, malicious app downloads the sniffed ble packets from the malicious database and starts advertising them, pretending to be immuni app . finally, immuni app becomes the victim of the relay attack by receiving the sniffed ble packets from malicious app . even though our implementation refers to the immuni use case, the above-mentioned relay attack can be applied on any gaen-based app. since such apps do not track the user location, they are not able to detect relay attacks. attackers can broadcast sniffed packets from any location, pretending to be the original owner of the ble packet, as soon as it is re-transmitted within the time window of about two hours. our implementation of the actguard design is called immuniguard and it was developed considering the italian act app, i.e., immuni. we implemented immuniguard as an android app, requiring an android sdk version equal to . the experiments were performed on a xiaomi redmi plus and a complete demo video of the defence applied by immuniguard against a relay attack is available online . immuniguard requires the following permissions: • internet and internet_network_state, to interact with an online database. • bluetooth and bluetooth_admin, for bluetooth scanning and advertising features. • access_fine_location, for gps tracking (can be downgraded to access_coarse_location). • foreground_service, for regular monitoring of gps locations. below, we will go through the main steps involved in the immuniguard workflow. scanning for immuni advertisement packets. immuni-guard uses the standard ble library, provided by the android platform, to regularly perform scans. during the scans, immu-niguard specifically looks for immuni advertisement packets, identified by a value equal to " xfd f" in the service uuid, as any other gaen-based app. storing rpis. for each contact with another immuni app user, immuniguard saves the advertised rpi, the current location and the current time. this data is stored in two databases: • mycontactstable, containing information related to contacts with other immuni app users. • positivetable, storing information uploaded by positive users to immuniguard database. new entries are added to mycontactstable every time a new contact occurs, including their hash value. positivetable is updated whenever immuniguard tries to download new information from the immuniguard online database. immuniguard should also store the rpis broadcasted by the app user immuni app, but this information is currently restricted. in our current proof of concept, immuniguard randomly generates its own dummy rpi, unrelated to immuni. uploading data if infected. whenever immuniguard users are diagnosed as sars-cov- positive, they can upload the content of the mycontactstable database, locally stored on the app. in a real scenario, the data upload should be validated by public health authorities. in our proof of concept, this functionality can be performed anytime. periodically downloading new infected data. immuni-guard needs to regularly download information about new users diagnosed as sars-cov- positive. in a real scenario, this process would be automated, but in our proof of concept, this functionality can be performed anytime. updating health risk status. the health risk status is calculated by comparing and matching the hashes in my-contactstable and in positivetable, to find if the owner of immuniguard was in proximity of an infected user. in our proof of concept, the health risk status is represented by the number of infected contacts. immuniguard is a standalone app, and it is not allowed to access to the list of rpi broadcasted by the active instance of immuni app installed on the user's device. immuniguard needs that list, in order to compute its own contact hashes, store them inside mycontactstable and compare them with the infected hashes downloaded from the immuniguard server. otherwise, the detection of relay attacks is not possible. this issue can be easily solved with the integration of our solution as a feature of the official app, thus gaining direct access to the list of rpi broadcasted by immuni. in this section, we describe in details the act apps we analyzed (i.e., section vi-a), specifically focusing on the privacy (i.e., section vi-b) and security (i.e., section vi-c) issues that affect them. tracetogether. tracetogether [ ] is the first contact tracing app based on ble technology and developed in singapore, one of the first countries affected by sars-cov- . it implements the bluetrace protocol, exchanging temporary identifiers through ble advertisement packets and allowing user to willingly upload them on a centralized health authority server. during interviews with patients positive to sars-cov- , medical operators ask if the patient installed tracetogether and are able to dynamically adjust proximity and duration filtering thresholds for a better contact tracing performance. safe paths. safe paths [ ] is a secure location logging technology, based on the "private automated contact tracing (pact)" protocol, developed by mit researchers. it currently supports only gps tracking, with plans for other location and proximity technologies. safe paths stores the gps movement trails of its owner, who can willingly share them with an authorized public health authority as part of the contact tracing process. stayhomesafe. stayhomesafe [ ] is a mobile app developed by the government of the hong kong special administrative region. the hong kong procedure for compulsory quarantine requires wearing a wristband connected to the stayhomesafe app. the app utilizes an innovative geofencing technology. the initial setup requires the user to slowly walk around the house so that the app can record data signals (e.g., wi-fi, bluetooth, gps) as a form of unique signature of that house, and detects whenever the user leaves this virtual fence. stayhomesafe does not collect sensitive data, nor discloses the location of its user. hamagen. hamagen [ ] is a mobile app developed by the israel ministry of health, which employs both gps and ble technology. hamagen regularly saves the device location coordinates and compares them with the gps trails of infected people. additionally, it can perform proximity tracing through ble. all data, which the ministry of health has access to, is explicitly shared by infected patients. aarongya setu. aarongya setu [ ] is a mobile app developed by the government of india. the app is mandatory for all indian citizens, and it is also used by employers to monitor the health risk status of their employees. aarongya setu uses both gps tracking and ble technology. during the initial setup, the user is required to submit personal data such as his/her name and phone number. stopcovid. stopcovid [ ] is a mobile app developed by the national institute for research in digital science and technology (inria), as requested by the government of france. the app uses ble technology, and implements the robert protocol, as the government considers a centralized approach more secure. swisscovid. swisscovid [ ] is a mobile app developed by the switzerland federal office of public health. the app uses ble technology, gaen in particular, to exchange anonymous temporary identifiers. whenever a user is infected, he can willingly reveal and submit his temporary identifiers in order to warn other app users met in the past days. covid alert ny. covid alert ny [ ] is a mobile app commissioned by the new york state department of health. akin to many other american contact tracing apps, it implements gaen. a special feature allows users to willingly disclose personal information, anonymously, such as the user county, gender, age-range, ethnicity, and symptoms. technical data can be also be anonymously shared with the centralized server, if the user chooses to do so. the primary purpose of contact tracing app is to monitor the movements of citizens, so they suffer from a range of privacy issues. mass surveillance. since the government is in charge of the app and of the centralized server, mass surveillance is a plausible threat. g. avitabile et al. [ ] enumerate several types of mass surveillance scenarios applied to the dp- t framework. infected patients can be easily tracked by an attacker possessing a pervasive infrastructure (e.g., a corporation). the only requirement is a sufficiently large set of devices able to collect ble signals. infected patients upload their sk to the centralized server, and other users are able to calculate ebid from them. an attacker can track the movements of infected patients, if they stay nearby the set of devices. another scenario involves the centralized server colluding with health authorities to map anonymous app users with real identities, since the health authorities themselves perform tests and allow data uploads. social graph. in a centralized approach, the centralized server stores all of the infected patients ebid. s. vaudenay [ ] argues that the centralized server would obtain various lists of pseudonyms unlinkable to real users. thus, a large monitoring infrastructure and heavy data mining would be required in order to extract meaningful information. the issue has a more limited impact over decentralized approaches, since the act apps store ebid locally. sharing unwanted data. act apps may ask for, and share, more data than necessary, and potentially disclose them to third parties. even though many act app developers released their app source code, most of them did not release the source code of the centralized server. thus, there is no guarantee that the disclosed code is the real one. the indian official act app, i.e., aarongya setu, asks for a lot of sensible information (i.e., name, phone number, age, gender, profession, workplace, recent travels) without proper motivations behind this design choice [ ] . it also collects plenty of data using gps and ble, and the centralized server code is yet to be disclosed. d. leith et al. [ ] analyze the data shared through the gaen api. on android devices, the gaen api relies on the google play services, which connect to google servers multiple times a day, sharing quite a lot of sensible data, such as: email address, phone number, ip address (used to retrieve location), phone imei, hardware serial number, sim serial number and wi-fi mac address. ble technology allows devices to communicate without consuming a lot of battery power. however, the anonymity constraint, embraced by act apps to protect the privacy of users, allows any attacker to send false advertisement packets, or tamper with existing ones. security countermeasures are scarce because of the traditionally low computational power of ble devices. replay and relay attacks. bluetrace researchers acknowledge the susceptibility of their protocol to replay and relay attacks [ ] , but do not propose any countermeasure, apart from a human-in-loop methodology. p. dehaye et al. [ ] highlight the same issue with replay and relay attacks, and illustrate the idea of tampering with the rpi and the aem. tampering with the transmission power level value would alter the victim's risk score, making it completely unreliable. denial-of-service. m. e. garbelini et al. [ ] uncover how a faulty software implementation of several ble system-onchip vendors exposes devices to two families of vulnerabilities. malicious attackers in radio range can manufacture and send specific ble packets to trigger deadlocks, crashes and buffer overflows. smartphones may incorporate one of the affected system-on-chip, as a part of their bluetooth module. contact tracing is used to monitor the spread of sars-cov- , but suffers from scalability issues, since it requires human intervention all the way through. mobile apps are extremely powerful assets in the scope of act: they benefit from the popularity of mobile devices among the population, allow for pervasive proximity tracing and gather data in complete autonomy. many european and asian countries already released their own national contact tracing app, since each government could choose among the several protocols proposed by different actors: government agencies and commissions (i.e., bluetrace), researchers (i.e., pact, robert, dp- t), or vendors of the leading mobile os (i.e., gaen). to be successful, act solutions require consensus from the citizens, since they have to install and use them on a voluntary basis. thus, adoption rate is the most valuable metric: a. galanopoulos et al. [ ] describe how the current adoption rate varies between % and % for many countries, while the scientific community debates that, to be effective, act solutions require at least an adoption rate of %. tracetogether (singapore) was downloaded by more than one million users, representing the % of singaporean citizens. aarongya setu (india) achieved a total of million downloads, amounting to % of the indian population. immuni (italy) reached an adoption rate of . % in october , with more than million downloads. covid tracker [ ] (ireland) is possibly the contact tracing app with the highest adoption rate in europe, standing at %. it identifies approximately close contacts per single positive case, while the south korean mass surveillance system identifies slightly more than close contacts per case [ ] . there are multiple causes for such low adoption rates. user reticence is related to act unreliability and to the fear of privacy violation. d. zeinalipour-yazti et al. [ ] tested swiss, german and italian contact tracing apps risk score calculation, and argue that ble signal strength has little correlation to physical proximity of app users. if that was the case, the foundations of the whole act process would fail. citizens are worried about the government spying on them, and exploiting the critical situation to start a mass surveillance program. bluetooth proximity tracing protocols offer better privacy than gps ones, since they do not share any location data. however, they are still vulnerable to collusion between the government and the public health authority, and to classic security threats, such as relay, replay and denial-of-service (dos) attacks. possible consequences would be the disclosure of sensitive information, surveillance, quarantine enforcement, identity theft and data tampering. in this paper, we focused on the resilience to relay attacks of all the act solutions and, in particular, on gaen-based apps. we found out that they are vulnerable to such attacks and we were able to develop an attack proof of concept against immuni, the italian gaen-based app. by designing two malicious apps, we managed to capture the official immuni advertising packets from a victim app with the first malicious app and to relay them through the second malicious app, to a designed victim. we were able to impersonate an official gaen-based app just by re-transmitting the original ble advertising packets. to tackle the vulnerability we found out, we designed actguard, a solution that is compliant with current gaen design, while providing a defence against relay attacks and retaining the same privacy features of existing gaen-based apps. actguard locally stores the location of each contact between two app users, together with rpi of those two users and the time of the contact. however, by saving such information as the result of a hash function, actguard fully guarantees the privacy of its users, even when that information is shared with a remote server, functioning as a database for infected user hashes. we implemented a proof of concept of actguard by considering immuni, thus releasing the immuniguard app. we demonstrated how immuni is vulnerable to relay attacks and how immuniguard can effectively detect a relay attempt by finding inconsistencies between data downloaded from immuni server and from immuniguard server. crepuscolo: a collusion resistant privacy preserving location verification system apps gone rogue: maintaining personal privacy in an epidemic robert protocol exposure notification system bluetrace: a privacy-preserving protocol for community-driven contact tracing across borders pepp-pt aarongya setu for research in digital science and t. (france) swisscovid of health privacy policy bending spoons kong special administrative region, ""stayhomesafe" mobile app user guide towards defeating mass surveillance and sars-cov- : the pronto-c fully decentralized automatic contact tracing system centralized or decentralized? the contact tracing dilemma privacy prescriptions for technology interventions on covid- in india contact tracing app privacy: what data is shared by europe's gaen contact tracing apps swisscovid: a critical analysis of risk assessment by swiss authorities sweyntooth: unleashing mayhem over bluetooth low energy measurementdriven analysis of an edge-assisted object recognition system covid tracker u.s. states are rolling out covid- contact tracing apps. months of evidence from europe shows they're no silver bullet covid- mobile contact tracing apps (mcta): a digital vaccine or a privacy demolition key: cord- -idj hd authors: li, jinfeng; guo, xinyi title: covid- contact-tracing apps: a survey on the global deployment and challenges date: - - journal: nan doi: nan sha: doc_id: cord_uid: idj hd to address the massive spike in uncertainties triggered by the coronavirus disease (covid- ), there is an ever-increasing number of national governments that are rolling out contact-tracing apps to aid the containment of the virus. the first hugely contentious issue facing the apps is the deployment framework, i.e. centralized or decentralized. based on this, the debate branches out to the corresponding technologies that underpin these architectures, i.e. gps, qr codes, and bluetooth. this work conducts a pioneering review of the above scenarios and contributes a geolocation mapping of the current deployment. the apps vulnerabilities and the directions of research are identified, with a special focus on the bluetooth-inspired decentralized paradigm. in the recent few months, contact-tracing apps have emerged and pushed the boundary of innovations in response to the outbreak of the coronavirus (covid- ) [ ] . a contact-tracing app [ ] is a mobile platform that assists the identification of people who may have come into contact with an infected person, and the subsequent collection of further information about these contacts for containing the virus' spreading. there is an ongoing debate on the deployment of the apps regarding their technology framework, i.e. centralised [ , ] versus decentralised [ , ] , and their corresponding sensor technologies, i.e. the global positioning system (gps) integrated with quick response (qr) codes scanning [ , ] and big data analysis [ , ], versus the wireless bluetooth devices [ ] enabled by microwave [ , ] and millimetre-wave [ ] [ ] [ ] [ ] communications. in the centralised architecture, personal data collected through the app is controlled by government authority. these apps mainly follow the pepp-pt (pan-european privacy-preserving proximity tracing) [ ] protocol, but the consensus amongst the technical community is that this framework is too academic for practical development. for the decentralized approach, the personal data is enclosed or controlled by individuals only on personal devices. these apps follow the dp- t (decentralised privacy-preserving proximity tracing) [ ] data protection solution recently developed by the european academics. however, this framework is only partially decentralised, i.e. there is an anonymous centralised database for only the infected people. google and apple in partnership [ ] will launch an exclusive decentralised framework in may which will be more compatible with the android and ios systems. regarding the technologies and infrastructures that underpin the two architectures, gps is based on crowd mapping for tracking the spread of the covid- , while the qr codes scanning approach is combined with physical temperature testing equipment or thermal imaging cameras to track the healthy or infected individuals' movement on public transport. the bluetooth method detects other devices retained for a certain amount of time within a certain range of distance, and notifies the devices which have had sufficient contacts with the infected individual's device, assuming that the infected individuals report their anonymous infection states to the app. researchers from oxford [ ] recently modelled and proposed a threshold on the active user rates (at least %) for the app to fully deliver its valuable insights for the government to contain the virus. there is arguably a growing trend globally and especially in europe that the decentralised architecture would be preferable. keeping personal data safe and secure is one of the greatest challenges posed by the rapid development of today's health informatics. the up-to-date regulations and frameworks are detailed in sections below, including the general data protection regulation (gdpr) [ , ] , as well as the key competing architectures that have been mentioned in section . [ ] released on april which was followed by the german and italian governments, and was involved with the development of the uk government's nhsx app (centralised). . decentralized privacy-preserving proximity tracing (dp-ppt)/ (dp- t) [ ] released on april -no pooled data is collected, which largely mitigates the privacy risk. the none-infected individuals' data are decentralised based, and the infected individuals' information will be collected anonymously to a central database. . apple and google partner on covid- contact tracing technology framework [ ] (yet to be released in may) -privacy-preserving contact tracing, bluetooth based, decentralised, free of gps. apple and google tech is currently trading (integrating) with some of the governments self-running apps. . government-run contact tracing technology [ ] framework that not going to deploy apple & google's framework, e.g. the uk, france, and several us states. we produce the first geolocation mapping for the global deployment of the covid- contact-tracing apps in fig. , with the format codes in an order of the country name, app name, the number of users (download times), and the underpinning technologies (gps, qr codes, bluetooth). the color of the country represents the employed framework, i.e. with red denoting the centralised architecture, while green representing the decentralised (or being migrating into the decentralised framework, e.g. austria, swiss, estonia, finland, germany, alberta of canada, and vietnam). following a software vulnerability-mapping analysis paradigm [ ] , the flaws of contact-tracing apps are analysed and summarised below. for instance, one of the key non-technical but important questions for the uk nhs covid- tracking app is the fault positives (i.e. what if people without concern for covid- maliciously report a positive using the app) and the fault negatives (i.e. what if infected people do not report their cases in the app). ( ) the health code on alipay and wechat (qr code and big data based, centralised) used in china has achieved a % of the population coverage and % on travelers. however, it is introducing a significant cost for the temperature testing equipment. many staff are involved in the checking house by house and helping people without using the digital app. is applied to only self-quarantined people staying at home, and hence not a rigorous contact-tracing app. it is susceptible to the risk if people take off the wristband and phone together and go out. as observed from fig. and the flaws analysis, bluetooth (either in centralised or decentralised framework) has accounted for % of all the tracking technologies, as compared with the gps ( %), which merits a further analysis into both the technical and geopolitical characteristics. firstly, there is a trade-off between the data privacy and the insights. arguably, the decentralised and no gps solution provides the highest level of data protection for individuals as no personal data is collected unless the individual is infected. without the gps tracking, apps cannot collect and trace the movement of the population geographically. with a decentralised framework, however, any data collected from individuals cannot be driven into a centralised database for future analysis, i.e. less information will be provided to the government for controlling the self-quarantine and the movement of the disease among the population. secondly, existing decentralised tracing apps such as the austria's stopp corona are issuing a static unique digital id to each user with rolling public and private keys (keeping the message encrypted and increasing the data protection standard). if the digital id is unique and static, it runs the risk that certain digital id could be hacked and paired with a mobile device, thus compromising the individual privacy. thereby, a rolling base digital id to mitigate this vulnerability would be a better practice. in the practical situation, this would be relatively easy to tailor and optimise compared with other related challenges. furthermore, different mobile devices exhibit a variety of bluetooth signal intensity at the ism band, i.e. the capability of each mobile device to determine the social distance precisely can vary. accordingly, it is of research and development interest regarding how this can be manipulated (converted) in a unified framework that regulates different generations of devices to communicate and share data with each other. other factors, such as the multipath interference and spatial blockage between devices are also urgent yet promising research areas that could tip the balance on the functional performance and fault tolerance of the bluetooth based contact tracking. coupled with the technical hurdles, the risk-level evaluating standard based on the distance and time contained should be updated accordingly. this work reviews the states-of-the-art contact-tracing apps for the covid- . a systematic mapping of the global deployment architectures and technologies is proposed, with a detailed analysis of the flaws for each scenario presented. specifically, the key challenges facing the bluetooth based solutions are identified to assist the health informatics decision-making concerning the uk's current status in covid- (see appendix a for an exponential fitting performed to model the cumulative cases up to date). available online: https://www.research.ox.ac.uk/article/ - - -digital-contact-tracing-can-slow-or-even-stop-coronavi rus-transmission-and-ease-us-out-of-lockdown (accessed on may ). . general data protection regulation (gdpr). available online: https://gdpr-info.eu/ (accessed on may ). . european commission. coronavirus: guidance to ensure full data protection standards of apps fighting the pandemic. available online: https://ec.europa.eu/commission/presscorner/detail/en/ip_ _ (accessed on may ). rolling updates on coronavirus disease (covid- ) nhs contact tracing app: how does it work and when can you download it? the telegraph the uk's coronavirus contacts-tracing app explained nhsx differs with apple and google over contact-tracing app without apple and google, the uk's contact-tracing app is in trouble google ban use of location tracking in contact tracing apps mobile applications to support contact tracing in the eu's fight against covid- guidelines / on the use of location data and contact tracing tools in the context of the covid- outbreak the coronavirus contact tracing privacy debate kicks up another gear available online eu privacy experts push a decentralized approach to covid- contacts tracing governments have to decide whether to scrap their own covid- contact tracing apps in favor of tech built by apple and google. here's what's at stake vulnerabilities mapping based on owasp-sans: a survey for static application security testing (sast) key: cord- - wu bao authors: kaptchuk, gabriel; goldstein, daniel g.; hargittai, eszter; hofman, jake; redmiles, elissa m. title: how good is good enough for covid apps? the influence of benefits, accuracy, and privacy on willingness to adopt date: - - journal: nan doi: nan sha: doc_id: cord_uid: wu bao a growing number of contact tracing apps are being developed to complement manual contact tracing. a key question is whether users will be willing to adopt these contact tracing apps. in this work, we survey over , americans to evaluate ( ) the effect of both accuracy and privacy concerns on reported willingness to install covid contact tracing apps and ( ) how different groups of users weight accuracy vs. privacy. drawing on our findings from these first two research questions, we ( ) quantitatively model how the amount of public health benefit (reduction in infection rate), amount of individual benefit (true-positive detection of exposures to covid), and degree of privacy risk in a hypothetical contact tracing app may influence american's willingness to install. our work takes a descriptive ethics approach toward offering implications for the development of policy and app designs related to covid . a growing number of coronavirus (covid ) contact tracing apps are being developed and released with the goal of tracking and reducing the spread of covid [ ] . these apps are designed to complement manual contact tracing efforts using location data or bluetooth communication to automatically detect if a user may have been exposed to the virus [ , ] . unlike manual contact tracing, which an investigator reaches out directly to affected parties, the benefits of these apps for public health scale quadratically with participation [ ] . this is because both data collection and data distribution are part of the apps operation. thus, it is critical to understand that factors that determine if people will be willing to adopt these apps. there are a large number of considerations that may influence users willingness to adopt [ ] . for example, a person may weigh the features the app's offer, the app's benefits to themselves and their community [ ] , the provider offering the app [ ] , how well the app will preserve the user's privacy [ , ] , and the app's accuracy. understanding the impact of each of these factors can help app developers make design decisions that can maximize their impact. drawing from the idea of descriptive ethics as a more fair approach to setting societal norms [ , ] , in this work we use surveys to evaluate how well covid apps need to function for users to be willing to adopt the apps. we present the results of a series of surveys of a total of , americans, sampled using both crowd-sourcing and online survey panels that can satisfy census-representative demographic quotas. there are many ways to measure how well a covid app works. from a public health perspective, reducing infection rate (i.e., basic reproduction number) is a key measurement of success. however, in order to understand individual's choices when it comes to adopting an app, we must also consider what it means for a covid app to work well for the individual who has the app installed. thus, we examine not only the societal-level, public health benefit of infection rate reduction, but also how app accuracy and app privacy, risk of the app exposing information collected by the app to others, influence reported willingness to adopt a covid app. within app accuracy, we consider both false negatives, the app failing to detect an exposure to covid , and false positives, the app falsely notifying the user that they were exposed when they were not. understanding how rates of app failure may influence adoption allow us to estimate user response to potential app designs. for example, saxena et. al. [ ] have shown that using bluetooth as a method for detecting proximity may be innately error prone, estimating an approximate error rate of - % (including both false positive and false negative rates). our results allow us to estimate user response to such error rates. in summary, we address the following three research questions: (rq ) do both accuracy (precision and recall) & privacy influence whether people want to install a covid app? (rq ) do different types of people weight accuracy or privacy more heavily? (rq ) how much public health benefit, accuracy, and/or privacy is necessary for people to want to adopt covid apps? we find that: • between - % of americans report being willing to install an app that is "perfectly" private and/or accurate, a significant increase from the - % who are willing to install an app with unspecified privacy or accuracy [ , ] . • false negatives have a significantly stronger influence on reported willingness than false positives or privacy risks. • reported willingness to install correlates with the public health benefit and/or personal health benefit of a contact tracing app. specifically, the majority of americans report being willing to install an app that offers at least a % improvement in public health or in personal safety over the baseline rate offered when not using the app. we conducted a series of surveys to answer our research questions. in this section we discuss our questionnaires, questionnaire validation, sampling approaches, analysis approaches, and the limitations of our work. all studies were irb approved by the microsoft research irb, a federally recognized ethics review board. in this first survey we sought to understand how accuracy and/or privacy considerations might influence willingness to adopt (rq ) and how respondent demographics and experiences might affect the relative weight of these considerations (rq ). we used a vignette survey [ ] to examine these questions as vignette surveys are known to maximize external validity. questionnaire. our questions were framed around a contact-tracing app scenario. half of the respondents were placed in the proximity contact tracing scenario while the other half were placed in the location scenario. the proximity scenario was phrased as: imagine that there is a mobile phone app intended to help combat the coronavirus. this app will collect information about who you have been near (within feet), without revealing their identities. the app will use this information to alert you if you have been near someone who was diagnosed with coronavirus. if you decide to inform the app that you have been diagnosed with coronavirus, the app will inform those you've been near that they are at risk, without revealing your identity. the location scenario was phrased as: imagine that there is a mobile phone app intended to help combat the coronavirus. this app will collect information about your location. the app will use this information to notify you, without revealing anyone's identity: • if you have been near someone who tested positive for coronavirus • about locations near you that were recently visited by people who tested positive for coronavirus if you decide to report to the app that you have been diagnosed with coronavirus, the app will inform those you've been near that they are at risk without revealing your identity. participants were then routed to a set of control questions or a set of experimental questions regarding accruacy and privacy (in randomized order). all participants were asked "would you install this app?" after a given question, with answer choices "yes", "no", and "it depends on the [risk, chance of information being revealed, etc.]". control. we had three control conditions (respondents saw only one of these three questions). perfect accuracy: imagine that this app will work perfectly. it will never fail to notify you when you are at risk nor will it ever incorrectly notify you when you are not at risk. perfect privacy: imagine that this app perfectly protects your privacy. it will never reveal any information about you to the us government, to a tech company, to your employer, or to anyone else. perfect accuracy and privacy: imagine that this app works perfectly and protects your privacy perfectly. it will never fail to notify you when you are at risk nor will it ever incorrectly notify you when you are not at risk. it will also never reveal any information about you to the us government, to a tech company, to your employer, or to anyone else. experimental. these participants were asked about accuracy and privacy, in randomized order. imagine that this app occasionally fails to notify you when you have been near someone who was diagnosed with coronavirus. accuracy (false positives): imagine that this app occasionally notifies you that you have been near someone who has coronavirus when you actually have not been exposed. privacy: imagine that this app might reveal information about [who you have been near/your location] to [entity] . this information may be used for a purpose unrelated to the fight against coronavirus. we asked about four entities, drawn from the list of examined by redmiles and hargittai [ ] : "non-profit institutions verified by the government", "technology companies", "the us government", and "your employer". validation. the questionnaire design was validated through expert reviews with multiple external researchers. additionally, three attention check questions were included, one general attention check and two scenario-specific attention checks that ensured respondents understood the scenarios described. sample. americans answered our survey. the sample was quota sampled by cint to be representative of the us population demographics on age, gender, income, and race. analysis. we answered rq using x proportion tests to compare responses to our different sets of questions. we answered rq by constructing two mixed effects binomial logistic regression models. in both models, our dependent variable was willingness to install the app, with "yes" and "it depends on the risk" grouped together as a positive outcome and "no" was treated as a negative outcome. we model responses to the accuracy and privacy questions separately, controlling for data type and entity, in the privacy model, and both data and accuracy type in the accuracy model. we included as dependent variables the respondents' age, gender, race, internet skill (as measured using the web use skill index [ ] ), level of educational attainment, party affiliation, and if they know someone who died due to complications from covid . finally, we include a mixed effects term to account for our within subjects design. in this survey we sought to evaluate how people's reported willingness to install coronavirus apps correlates with the amount of public health (infection rate reduction) and individual health (notification of at risk status -e.g., accuracy) benefit of a hypothetical coronavirus tracking app. questionnaire. all questions, except one control condition (fn app control, addressed below), were asked in the context of the following scenario. as the type of information compromised, as well as the entity that could compromise the information had relatively little effect on willingness to install in our first survey (see section ??), we consider only proximity-based data in this scenario. future work may wish to replicate these results for location information. please consider the following scenario. imagine that public health workers will notify you if they are able to determine that you have recently been near (within feet) someone who was diagnosed with coronavirus. • you do not have to do anything in order for the public health workers to monitor whether you have recently been near someone diagnosed with coronavirus. • however, the public health workers are not aware of every time you are near someone diagnosed with coronavirus. imagine that there is also a mobile phone app available that will alert you if you have been within feet of someone diagnosed with coronavirus. • the app will do this by collecting information about who you have been within feet of (who you have been "near"). • the app will not reveal the identity of the people you have been near. participants were then assigned to one of the branches in table no information in this survey was expressed in terms of percentages, due to a plethora of research in health risk and numeracy showing that people interpret rates far more accurately than percentages [ , ] . below we describe exactly how each of the questions referenced in table . was asked in our survey. implicit privacy. pilot tests of our survey revealed that people had implicit privacy perceptions of the app described, which were influencing their willingness to adopt the apps. we used a modified version of the paling perspective scale [ ] -a well validated tool for eliciting health risk perception -to assess respondents' perception of the likelihood that information collected by this app would be compromised . this measurement allows us to (a) report on people's perceptions of the likelihood that information from a coronavirus app will be compromised, (b) control for the effect of differing implicit privacy perceptions on willingness to install, and (c) validate the influence of these privacy perceptions by comparing willingness to install given an implicit privacy perception vs. an explicit one that we set by telling the participant the risk their privacy will be compromised (described in the next section). the question we used to assess implicit privacy belief was: studies show that despite best attempts to protect the data of those who use this app, some people may have information about who they have been near compromised and used for purposes other than the fight against coronavirus. please indicate on the chart below how many app users you think will have this information compromised over the next year. explicit privacy. in order to understand how different privacy risks impacted respondents' reported willingness to install coronavirus apps we asked some participants about their willingness to install in the context of explicitly known (as opposed to implicitly perceived, as aforementioned) privacy risks. we asked about explicit risk using the following question: studies show that despite best attempts to protect the data of those who use this app, some people may have information about who they have been near compromised and used for purposes other than the fight against coronavirus. x out of people who use this app will have this information compromised. we also asked all participants in this branch the false negative question (below) in order to be able to cross-validate the impact of explicit declaration of privacy risk vs. the effect of implicit perception of that risk on willingness to install. as it would not make much sense to assess implicit risk and then ask respondents whether they would install given their implicit perception of risk, privacy questions need to be paired with a benefit question. we chose to make our comparison using false negative questions since the results of survey showed that false negatives were equally as important in users' consideration of whether to install as was privacy. false negative. we asked respondents whether they would be willing to install an app that could detect n out of exposures to coronavirus compared to manual contact tracing, which could detect exposures a baseline number of times: out of . the question was phrased as follows: imagine that you are exposed to someone who has coronavirus times over the next year. if you do not use the app, out of times public health workers will be able to detect and notify you that you were exposed. if you use the app, fn out of times the app will be able to detect and notify that you that you were exposed. to compare willingness to install a % effective app as a baseline, we also had a fn control condition. this condition consisted of a scenario that did not describe manual contact tracing, but just described the app (in the same way as above), which respondents were told could detect in exposures (the same as the manual contact tracing option offered in the other conditions). false positive. we asked respondents whether they would be willing to install an app that detected all exposures to coronavirus , but had n out of additional false negatives. the question was phrased as follows: imagine that you are exposed to someone who has coronavirus times over the next year. if you do not use the app, out of times public health workers will be able to detect and notify you that you were exposed. the app is not perfect. if you use the app, the app will correctly notify you every time that you were exposed ( out of times). the app will also incorrectly notify you an additional fp times, when you were not actually exposed. public health benefit. finally, some respondents were assigned to a branch that evaluated how reduction in infection rate among app users would influence people's willingness to install an app. we chose % as the baseline infection rate without app use as this is the currently estimated u.s. infection rate by the ihme [ ] . studies show that out of people who do not use the app will be infected with coronavirus in the next year. h out of people who use the app will be infected with coronavirus in the next year. validation. the questionnaire design was validated through expert reviews with multiple external researchers. additionally, three attention check questions were included, one general and two scenario-specific, as in survey . sample. , amazon mechanical turk workers responded to our survey. these workers were split into different survey branches, as aforementioned, so all results sections note the number of respondents used in a particular analysis. analysis. we analyze the data obtained in this survey descriptively, through data visualization, and using binomial logistic regression analysis: with willingness to install as the dependent variable and the dependent variables of the varied factor (e.g., chance of fn) and perceived implicit privacy risk. to evaluate the impact of privacy on decision making we use a x proportion test to compare the proportion of respondents willing to install given some fn rate in the implicit and explicit privacy conditions. as with all surveys, the answers represented in these results are people's self-reported intentions regarding how to behave. as shown in prior literature on security, these intentions are likely to align directionally with actual behavior, but are likely to over-estimate actual behavior [ ] . the goal of this work is to show how willingness to adopt may be influeced by privacy/accuracy considerations, and thus the precise numeric estimates should not be interpreted as precise adoption estimates. additionally, regarding the rq survey, there are always concerns about the generalizability of crowdsourced results. to address these concerns, we also conducted the rq ,rq survey on amazon mechanical turk. we found only one significant difference (with small effect size) in the amt results as compared to the online survey panel results. due to the quantitative nature of the rq survey and the sample size required, and our comfort in the relatively representative nature of amt results on this particular topic verified by our rq , rq comparison as well as prior work on the generalizability of amt results in security and privacy surveys [ ] , we chose to proceed with amt for rq . in this section, we detail our findings. for those who prefer a swifter visual summary, please see http://www.cs.umd. edu/~eredmiles/how-good-good-enough.pdf. the results of our first survey, shown in figure illustrate that both accuracy and privacy do indeed significantly (x tests in comparison to the control conditions, p < . , bonferroni-holm multiple testing correction (bh correction)). we find that respondents did not significantly differentiate between perfect privacy vs. perfect accuracy (x prop. test, p= . , bh correction), perfect accuracy vs. both perfect accuracy and privacy (x prop. test, p= . , bh correction), of perfect privacy vs. both perfect accuracy and privacy (x prop. test, p= . , bh correction). on the other hand, respondents were % less likely to install an app with false negatives, regardless of the number of the fn rate, than one with false positives, regardless of the fp rate (x prop. test, p= . , bh correction). respondents were similarly less likely to install an app with false negatives, regardless of the fn rate, than one with privacy leaks to any of the entities examined (x prop. test, p= . , bh correction). respondents were equally as likely to install an app with false positives as one with privacy leaks (x prop. test, p= . , bh correction). respondents were more likely to say that their decision to install would depend on the risk of false positives ( %) or false negatives ( %) than the risk of a privacy leak ( % across entities). finally, respondents' reported willingness to install did not significantly differ (x prop. tests, p> . , bh correction) based on what data the app might leak to a particular entity, except for hypothetical leaks to the respondents employer ( figure ). only % of respondents were willing to install an app that might leak their locations to their employer while % were willing to install an app that might leak information about who they have been near (their proximity data) to their employer. the next section provides regression comparisons of willingness to install based on the entity to which the information was leaked, and also control for data type differences (finding no significant differences). in order to examine whether some american's weighed accuracy or privacy considerations more highly than others, we constructed two mixed effects logistic regression models as described in section . table : mixed effects logistic regression model of willingness to install apps with accuracy errors. question baseline is fn, data baseline is location, political leaning baseline is republican, mixed effects term controls for within-subjects design. we find that those who know someone who died from covid are over × as likely as those who do not to be willing to install an app that has errors in accuracy. additionally, we validate that even when controlling for demographic variance, respondents are more comfortable with false positives than false negatives: respondents are % more likely to report that they would install an app with false positives than one with false negatives. respondents were more comfortable installing an app with potential privacy leaks to a non-profit organization verified by the government an app with potential leaks to any other entity (their employer, a technology company, or the u.s. government). those who identify as democrats are nearly × as likely as those who identify as republican to be willing to install an app with privacy risks. finally, those who are younger and women are less likely to report that they would install an app with privacy errors. this gender finding aligns with past work showing that women may be more privacy sensitive than men [ , ] . finally, those who have higher internet skill are more willing to install an app that has either errors in accuracy or privacy leaks, likely because those with higher skill are more likely to install covid apps in general [ ] . in the findings above, we validate that the individual considerations of accuracy and privacy both impact reported willingness to install. in our second survey we examine whether we can model how the quantitative amount of public health (i.e., infection rate reduction) and individual benefit (i.e., fn and fp rates) influences willingness to install. figure provides an overview of these findings. to examine the relationship between amount of benefit and willingness to install beyond visual inspection, we construct logistic regression models. we find that, for every % reduction to infection rate offered by the app, respondents are % more likely to report that they would install . implicitly perceived risk of privacy leak in covid apps influences willingness to adopt; risk of covid app privacy leak perceived by respondents as between . % - . % in our second survey, we not only measured willingness to install based on amount of benefit but we also measured implicit privacy risk perception. in this section we briefly summarize how likely respondents thought it was that information from a covid contact tracing app would be leaked and we confirm the results of survey one: that privacy risk, even when unmentioned, influences willingness to adopt a covid app. figure summarizes respondents implicit perceptions of the risk of a privacy leak of covid app information in the next year. the median respondent (n= , ) reported perceived the risk of a privacy leak of app information in the next year as between . and . %, equivalent to the annual risk of an american having unattended property stolen. % of respondents reported perceiving the risk as between . % and . %. finally, we compare the proportion of respondents who were willing to install a covid app given an explicit statement of privacy risk (privacy risks were drawn from the portion of the implicit risk distribution reported by the majority of respondents) vs. their own implicit perception. we find no significant different between the proportion of respondents who were willing to install an app with a given false negative rate when relying on their own implicit privacy assumption and the proportion who were willing to install given an explicit statement of the risk of privacy leak. a regression model of willingness to install in the explicit condition finds a significant relationship between risk perception and willingness to install (o.r.: . , % ci: [ . , . ], p < . ). this lends support for our implicit privacy risk measurements and suggests that these implicit risk perceptions affect willingness to install similarly to explicit risk statements. finally, further confirming the relevance of all three components studied in this paper -benefits, accuracy, and privacyin users' consideration of whether to install, when we add implicit privacy risk as a factor to the regression models for willingness to install dependent on public health and individual benefit, we find that it is significant in all three models. experimental vignette studies in survey research privacy sensitive protocols and mechanisms for mobile contact tracing infection fatality rate -a critical missing piece for managing covid- the definition of morality human perceptions of fairness in algorithmic decision making: a case study of criminal risk prediction an update on survey measures of web-oriented digital literacy will americans be willing to install covid- tracking apps? -scientific american blog network gender differences in privacy-related measures for young adult facebook users effect of risk communication formats on risk perception depending on numeracy strategies to help patients understand risks net benefits: digital inequities in social capital, privacy preservation, and digital parenting practices of us social media users user concerns & tradeoffs in technology-facilitated contact tracing how well do my results generalize? comparing security and privacy survey results from mturk, web, and telephone samples asking for a friend: evaluating response biases in security user studies to put that in perspective: generating analogies that make numbers easier to understand smartphone-based automated contact tracing: is it possible to balance privacy covid- contact tracing and privacy: studying opinion and preferences decentralized privacy-preserving proximity tracing a scramble for virus apps that do no harm -the new york times with thanks to eric horvitz for the idea of investigating quantiative tradeoffs in public benefit, accuracy, and privacy. with thanks to cormac herley and carmela troncoso for survey feedback and general contact tracing conversations that contributed to this paper. this work was funded by microsoft research. key: cord- - x ro p authors: jiménez, luisa fernanda mancipe; nieto, gloria ramírez; alfonso, victor vera; correa, jairo jaime title: association of swine influenza h n pandemic virus (siv-h n p) with porcine respiratory disease complex in sows from commercial pig farms in colombia date: - - journal: virol sin doi: . /s - - - sha: doc_id: cord_uid: x ro p porcine respiratory disease complex (prdc) is a serious health problem that mainly affects growing and finishing pigs. prdc is caused by a combination of viral and bacterial agents, such as porcine reproductive and respiratory syndrome virus (prrsv), swine influenza virus (siv), mycoplasma hyopneumoniae (myh), actinobacillus pleuropneumoniae (app), pasteurella multocida and porcine circovirus (pcv ). to characterize the specific role of swine influenza virus in prdc presentation in colombia, farms from three major production regions in colombia were examined in this study. nasal swabs, bronchial lavage and lung tissue samples were obtained from animals displaying symptoms compatible with siv. isolation of siv was performed in -day embryonated chicken eggs or madin-darby canine kidney (mdck) cells. positive isolates, identified via the hemagglutination inhibition test, were further analyzed using pcr. overall, of the farms were positive for siv. notably, sequencing of the gene encoding the hemagglutinin (ha) protein led to grouping of strains into circulating viruses identified during the human outbreak of , classified as pandemic h n - . serum samples from gilts and multiparous sows between and were obtained to determine antibody presence of app, myh, pcv and prrsv in both siv-h n p-negative and -positive farms, but higher levels were recorded for siv-h n p-positive farms. odds ratio (or) and p values revealed statistically significant differences (p< . ) in prdc presentation in gilts and multiparous sows of farms positive for siv-h n p. our findings indicate that positive farms have increased risk of prdc presentation, in particular, pcv , app and myh. swine infl uenza is an acute, highly contagious respiratory disease resulting from infection with type a infl uenza virus, a member of the orthomyxoviridae family. influenza a viruses have been isolated from different species, including humans, pigs, dogs, horses, sea mammals and birds (kuntz-simon g, et al., ; webster r g, et al., ) . the viruses are classified into subtypes based on antigenic properties of the external glycoproteins, hemagglutinin (ha) and neuraminidase (na). influenza a viruses have been further classified into ha and na subtypes (webster r g, et al., ; fouchier r, et al., ) . the proteins are highly variable and critical for the induction of antibody response in the host (gramer m, ) . although other subtypes have been identifi ed, infl uenza a virus h n , h n and h n subtypes are the most prevalent in pig populations worldwide (kuntz-simon g, et al., ) . in pigs, the disease is characterized by sudden onset, coughing, dyspnea, fever and prostration, followed by rapid recovery. lesions generally develop rapidly in the respiratory tract and regress quickly. the course and severity of disease are likely to vary with the strain of the virus, age and immune status of the host (easterday b c, et al., ) . morbidity is high (near %), but mortality is low (usually less than %). recovery begins to days post-infection. subclinical infections are common, and most pigs can be reinfected with other strains without showing clinical signs (reeth v k, et al., ) . porcine respiratory disease complex (prdc), a multifactorial condition, is a serious health problem in growing and finishing pigs, and poses a threat to the swine industry worldwide. prdc results from a combination of viral and bacterial agents, including porcine reproductive and respiratory syndrome virus (prrsv), swine influenza virus (siv), mycoplasma hyopneumoniae (myh), actinobacillus pleuropneumoniae (app), pasteurella multocida and porcine circovirus type (pcv ) (kim j, et al., ) . although the etiology of prdc involves multiple pathogens and varies among farms, several researchers maintain that etiology is mainly associated with myh and prrsv (dee s, ; thacker e l, et al., ) . pneumonia caused by prdc is characterized by slow growth, decreased feed effi ciency, lethargy, anorexia, fever and cough. while farms and slaughter houses in colombia have tested positive for siv over the years, the role of siv in prdc is yet to be established. there is evidence of classic h n and infl uenza a h n pandemic circulation in swine populations, but surveillance reports on prdc are scarce to date. the main goal of the current research was to generate surveillance, epidemiological, antigenic as well as phylogenetic data to ascertain the presence of swine influenza (h n ) pandemic virus and determine its association with prdc (prrsv, myh, app and pcv ) in sows from production farms in colombia. a total of gilts and multiparous sows distributed in farms from three major swine-producing areas in colombia (antioquia, valle del cauca and cundinamarca) were included in the present study. blood samples were obtained via venipuncture of the jugular vein using red top vacutainer ™ tubes. serum samples were identifi ed and stored at - °c until analysis. two hundred and forty-two nasal swabs, bronchial lavage (bl) and lung tissue samples of animals displaying symptoms compatible with siv (easterday b c, et al., ) were collected from the farms. samples were collected in brain-heart infusion medium (bhi) medium (bd®) supplemented with % antibiotic and anti-mycotic solution (sigma®), filtered through a . μm filter, and stored at - c until processing for siv isolation, either in chicken embryo eggs or the madin-darby canine kidney (mdck) cell line. briefl y, day-old spf embryonated chicken eggs were inoculated via the allantoic cavity with μl filtered sample, incubated at °c for h, and monitored daily. allantoic fluid was collected and evaluated for hemagglutination activity following standard procedures (who, ) . twenty four-well plates were seeded with × cells/well of mdck and grown in dulbecco's modifi ed eagle medium (dmem) (gibco®) supplemented with % fetal bovine serum (fbs, gibco®), % antibiotic (sigma®) and % l-glutamine (sigma®). complete growth medium was removed from confl uent monolayers and washed three times with pbs supplemented with μg/ml tpck trypsin (sigma®). each well was infected with μl of the original fi ltered sample and incubated for h at °c, % co , followed by the addition of ml/well complete dmem. cells were incubated at °c, % co , for - days and monitored daily for cytopathic effect (cpe). following the incubation period, cell culture supernatant (cs) and allantoic fl uid (af) were collected and tested with the hemagglutination assay (ha) using chicken erythrocytes according to the standard office international des epizooties (oie) protocol (swine infl uenza, ) . all ha-positive samples from egg inoculation or cell culture isolation were further analyzed for effi cient subtyping of virus. viral rna was extracted using a commercial rna extraction kit (qiamp viral rna®, qiagen, ca, usa) following the manufacturer's protocol. amplification of ha and na gene segments was performed with a duplex reverse transcriptionpolymerase chain reaction (rt-pcr) assay targeting the respective genes of h n swine infl uenza virus (choi y, et al., ) . initial reverse transcription was performed with m-mvl reverse transcriptase (invitrogen) using μl viral rna as template and the uni- primer (invitrogen, maryland, usa) to generate cdna. the conditions for reverse transcription (rt) were as follows: °c for min, °c for min, °c for min, and a fi nal step of °c. amplifi cation of bp ha (primers h f and h r) and bp na (primers n f and n r) genes of siv was performed (table ) under the following reaction conditions: °c for min, cycles of °c for min, . °c for min, °c for min, followed by °c for min, and a fi nal step of °c to terminate amplifi cation. pcr products were run on an agarose gel, and isolated and gel-purifi ed using the qiaquick gel extraction kit (qiagen ® ). gel-purified products were sequenced by macrogen ® , usa, using big-dye ® terminator cycle sequencing. dna sequences were combined and edited using the lasergene sequencing analysis software package (dnastar ® , madison, wisconsin). multiple sequence alignments were made using clustal w to identify related reference infl uenza genes. sequence comparisons and phylogenetic relationships through a bootstrap trial of were determined with the mega . program using the clustal w alignment algorithm, and evolutionary history inferred using the neighbor-joining method (saitou n, et al., ) for tree construction. gene sequences of colombian strains were compared with those of swine, avian and human infl uenza viruses, which were retrieved from the ncbi infl uenza virus resource. serum samples were subjected to the hemagglutination inhibition test to identify antibodies against siv using strain a/puerto rico / (h n ) as antigen, according to the who protocol ( ). prdc was diagnosed based on detection of antibodies against prrsv, myh, app and pcv in serum. specifi c antibodies were measured with elisa kits (table ) . data were analyzed using the statistix . program. the chi-square test was initially performed to determine the probability of presentation of respiratory disease complex with and without the presence of siv, and logistic regression analysis subsequently utilized to determine the risk factors that increase the presentation of disease when prdc farms and animals are positive or negative for siv. data were considered signifi cant at p< . , and graphs plotted using graphpad software. among the farms surveyed, were positive for swine flu pandemic virus, either from embryo chicken egg or mdck cell culture isolates (table ) . positive the hemagglutination inhibition test revealed positivity to siv-h n p in . % gilts, . % sows with - births and . % sows with multiparous (> ) births. the data suggest a trend in reactivity to siv-h n p, as the antibody response appears to decrease through time. prdc presentation was lower in siv-h n p-negative than siv-h n p-positive farms ( table ). the odds ratio value and p value revealed significant differences (p< . ) in prdc risk presentation in gilts and multiparous sows of siv-h h p-positive farms (figure ) . analysis was completed using logistic regression to de-termine the likelihood of increase in prdc disease presentation in siv-h n p-positive farms. according to our results, siv-h n p-positive farms had . , . and . times greater risk of introduction of app, myh and pcv , respectively, compared with siv-h n p-negative farms. in contrast, for prrsv, reduced risk of disease presentation (or, . ) was observed in farms positive for siv-h n p. the presence of siv-h n p in positive farms increases the risk of introduction of prcd especially pcv , app and myh. when analyzed by age group, the gilts are more susceptible to pcv , myh and app, respectively, while multiparous sows are more susceptible to pcv , app and myh. logistic regression analysis was performed to determine the risk of prdc presentation in farms positive and negative for siv-h n p. the odds ratio and p value revealed significant differences (p< . ) in risk prdc presentation per individual. in animals positive for siv-h n p, major risk of mycoplasma hyopneumoniae, pcv and app presentation was . , . and . times that of siv-negative animals, respectively. in contrast, for prrsv, the risk of presentation of disease (or, . ) was decreased in animals positive for siv-h n p. many groups have defined prdc as a multifactorial respiratory disease involving several pathogens (harms p a, et al., ; kim j, et al., ; opriessing t, et al., ; fachinger v, et al., ) , while other investigators (thacker e ,l, ) argue that prdc is enzootic pneumonia produced by mycoplasma spp. and other opportunistic bacteria, aggravated by the respiratory virus. the main goal of the present study was to conduct a systematic analysis of swine influenza virus infection and determine its role in prdc presentation in the major swine-producing areas of colombia. serological reactivity determined using hi disclosed that swine influenza virus h n is prevalent in gilts and multiparous sows from farms in colombia, although the antibody response decreased over time. we expected to fi nd isolates from infl uenza of classical h n virus origin in the fi eld. surprisingly, however, sequencing results led to grouping of viruses isolated from the farms within the original h n pandemic of , four of which shared % homology. therefore, it is important to consider swine infl uenza virus of pandemic origin as a pathogen playing an infl uential role in the presentation of prdc in commercial pig farms in the country. the pandemic h n / virus (a/ca/ / ) is a swine infl uenza a virus subtype h n strain responsible for the fl u pandemic. the virus originates from swine, but was never established as circulating in pig populations before its detection in humans (reeth v k, et al., ) . siv-h n p presentation in colombia may have been attributed to human transmission from pigs, as reported in various countries (weingartl h m, et al., ; sreta d, et al., ; pereda a, et al. ) . this hypothesis is based on the presentation of disease outbreaks in pigs and humans, as well as reports by the world organization for animal health (oie), which documented the presence of the virus in various countries, including argentina, canada, australia, ireland and united states. our results indicate that siv h n p positivity in animals increases the risk of presenting prdc, particularly pcv , app and myh. primiparous and multiparous sows are more susceptible to pvc , followed by myh and app in gilts, and app and myh in multiparous sows. a study by hansen m s, et al. ( ) in denmark revealed the presence of pcv in most lung samples from animals presenting prdc and indicated associations of this virus with the majority of viral and bacterial pathogens. the most frequently detected pathogens were in the order pcv , myc hyopneumoniae, pasteurella multocida and myc hyohinis, identified in different proportions in pigs with pneumonia in germany (palzer a, et al., ) , taiwan (chiou m t, et al., ) and the usa (choi y, et al., ) . this diversity in the presentation of pathogens involved in prdc may reflect the different diagnostic methods, health status of animals, farm management factors or simply the complex nature of pneumonia in pigs. siv-h n p-positive farms were more susceptible to prdc presentation for both gilts and multiparous sows, and interactions were mainly observed among pcv and siv. research in animals between and weeks of age showed that siv acts together with pcv in pigs under fi eld conditions. prdc induces severe clinical disease lesions presenting both infl uenza and wasting syndrome post-weaning (dorr p, et al., ; pallares f j, et al., ) . wei h, and co-workers ( ) concluded that subclinical pcv infection results in increased severity of subsequent siv-h n infection and siv can trigger the severe form of pcv , although the percentage of animals affected is comparable to that reported for other co-infections known to trigger severe pcv . myh is associated with presentation of swine enzootic pneumonia, characterized by high morbidity and low mortality in affected farms (opriessing t, et al., ) . myh acts as an immunosuppressant by increasing its pathogenicity as well as that of siv at the time of replication of the two agents (opriessing t, et al., , yazawa s, et al., . thacker e and colleagues ( ) reported that the percentage of lung lesions and clinical presentation of disease in pigs inoculated with siv days after inoculation with myh is higher than that in pigs inoculated with myh only. in our study, higher risk ( . %) of introduction of myh in siv-h n p-positive farms was observed, compared to siv-h n p-negative farms. app, a highly contagious infectious agent, causes pleuropneumonia in pigs, characterized by increased susceptibility to secondary infections (gutierrez c, et al., ) . app is considered an obligate parasite of the respiratory tract of pigs (taylor d j, ) . this report is consistent with the present study, where siv-h n ppositive farms displayed . % positivity for app. app has two different biotypes, specifically, biotype containing serotypes and biotype with six serotypes. all biotypes are capable of causing disease, with some serotypes being more aggressive than others (bossé j t, et al., ) . app is reported to increase the incidence of stress associated mainly with pleuropneumonia and interactions with viral and bacterial agents involved in the introduction of prdc (bossé j t, et al., ; kim j, et al., ) . regarding prrsv presentation in siv-h n ppositive and -negative farms, the incidence of prrsv was lower in positive ( . %) than negative farms ( . %). co-infection studies of prrsv and siv have yielded conflicting results in terms of clinical disease. in animals experiencing swine influenza infection and prrsv infection as a secondary pathogen, the disease is exacerbated, leading to prolonged fever, increased cough and weight loss ( van reeth k, et al. ; kitikoon p, et al., ). on the other hand, prior infection with prrsv does not aggravate acute siv presentation but can generate subsequent chronic infection (pol j m, et al., ) . prrsv is transmitted transplacentally through direct contact after birth, and piglets from infected mothers are infectious during subsequent stages of production, particularly the fattening and finishing stages (prieto c, et al., ) . similarly, siv is transmitted by direct contact between infected and uninfected animals via aerosol and can affect animals at any stage, particularly during fattening and fi nishing, as the pigs are raised in very close proximity to each other ( van reeth k, et al., ) . for this reason, we expected to find significant interactions between the two viruses in multiparous sows. however, this was not observed in our study. regarding tropism, prrsv has high affinity for differentiated macrophages. destruction of these cells in the lung by prrsv is suggested as the key event causing lung susceptibility to secondary pathogens (gucht s v, et al., ; van reeth k, et al., ) . prrsv and siv are the primary etiologic agents associated with respiratory disease of pigs in the united states (choi y, et al., association of siv-h n p with prdc in sows from commercial pig farms in colombia ) and co-infection could explain this role in porcine respiratory disease complex presentation. primary siv infection of epithelial cells of the airways results in cellular necrosis, production of infl ammatory mediators and rapid infiltration of phagocytic cells, including lung alveolar macrophages (pam) susceptible to prrsv infection. siv infection-infl ammation may increase the target cells for initial infection of prrsv, resulting in increased and prolonged pneumonia (kitikoon p, et al., ) . based on results from the current study, we propose that siv-h n p blocks prrsv replication or vice versa, and presentation of the disease is dependent on virulence and the time interval between infections (kitikoon p, et al., ) . prdc results from a combination of infectious agents as well as environmental stressors and challenges that affect the health of pigs, resulting in reduced performance, increased medication costs and high rates of mortality. clearly, the different forms of presentation of respiratory disease in pigs are attributed to the interactions of viral and bacterial agents, and although the etiology of prdc involves multiple pathogens with variations among farms, limited information is available on the spread of the virus in our pig population. the current study showed that pcv and app are the main viral-bacterial agents interacting with siv-h n p in positive animals. although it is difficult to establish whether swine flu virus is a pandemic primary, secondary or opportunistic agent, our data clearly indicate that siv-h n p plays an important role in the presentation of prdc, signifying the need to increase agricultural surveillance to prevent future outbreaks. this study was supported by colombia's agriculture ministry, colombian association of swine producers, cercafe and national university of colombia. we are grateful to mirela norho for statistical analysis and farms of antioquia y valle del cauca that participated in the study. the authors declare no conflicts of interest in terms of financial relationships with the industry or directly with the academy. all experiments were performed in accordance with the institutional and national guidelines for the care and use of laboratory animals. jjc, gcr, vjv designed experiments. lfm performed experiments. lfm analyzed data. lfm and jjc wrote paper. all authors read and approved the final manuscript. actinobacillus pleuropneumoniae: pathobiology and pathogenesis of infection etiological and epidemiological survey of prdc associated pathogens in taiwan detection and subtyping of swine influenza h n , h n and h n viruses in clinical samples using two multiplex rt-pcr assays retrospective analysis of etiologic agents associated with respiratory diseases in pigs the porcine respiratory disease complex: are subpopulations important? epidemio-logic assessment of porcine circovirus type coinfection with other pathogens in swine swine influenza the effect of vaccination against porcine circovirus type in pigs suffering from porcine respiratory disease complex characterization of a novel influenza a virus hemagglutinin subtype (h ) obtained from black-headed gull an update on swine infl uenza ecology and diagnostics. american association of swine veterinarians the combination of prrsv virus and bacterial endotoxin as a model for multifactorial respiratory disease in pigs simultaneous serological evidence of actinobacillus pleuropneumoniae, prrsv, aujeszky's disease and infl uenza viruses in spanish fi nishing pigs an investigation of the pathology and pathogens associated with porcine respiratory disease complex in denmark three cases of porcine respiratory disease complex associated with porcine circovirus type infection association of porcine circovirus with porcine respiratory disease complex vaccine effi cacy and immune response to swine influenza virus challenge in pigs infected with porcine reproductive and respiratory syndrome virus at the time of siv vaccination genetic and antigenic evolution of swine infl uenza viruses in europe and evaluation of their zoonotic potential porcine circovirus type associated disease: update on current terminology, clinical manifestations, pathogenesis, diagnosis, and intervention strategies concurrent infections are important for expression of porcine circovirus associated disease porcine circovirus type (pcv- ) coinfections in us fi eld cases of postweaning multisystemic wasting syndrome (pmws) associations between pathogens in healthy pigs and pigs with pneumonia pandemic (h n ) outbreak on pig farm dual infections of prrsvv/influenza or prrsvv/ actinobacillus pleuropneumoniae in the respiratory tract transplacental infection following exposure of gilts to porcine reproductive and respiratory syndrome virus at the onset of gestation first isolation and identifi cation of h n swine infl uenza viruses in colombian pig farms influenza in birds, pigs and humans: old theories versus current viewpoints the neighbor-joining method: a new method for reconstructing phylogenetic trees pandemic (h n ) actinobacillus pleuropneumoniae mycoplasma hyopneumoniae potentiation of porcine reproductive and respiratory syndrome virus-induced pneumonia immunology of the porcine respiratory disease complex mycoplasmal diseases dual infections of feeder pigs with porcine reproductive and respiratory syndrome virus followed by porcine respiratory coronavirus or swine infl uenza virus: a clinical and virological study influenza virus a pathogenicity: the pivotal role of hemagglutinin infection of cesarean-derived colostrum-deprived pigs with porcine circovirus type and swine infl uenza virus genetic and pathobiologic characterization of pandemic h n infl uenza viruses from a naturally infected swine herd who, . manual on animal influenza diagnosis and surveillance experimental dual infection of pigs with an h n swine infl uenza virus (a/sw/hok/ / ) and mycoplasma hyopneumoniae key: cord- -c sbqiyy authors: ivers, louise c; weitzner, daniel j title: can digital contact tracing make up for lost time? date: - - journal: lancet public health doi: . /s - ( ) - sha: doc_id: cord_uid: c sbqiyy nan can digital contact tracing make up for lost time? contact tracing is a fundamental public health intervention, and a mainstay in efforts to control and contain severe acute respiratory syndrome coronavirus (sars-cov- ), the virus responsible for the covid- pandemic. at the time of writing, the pandemic has caused more than million cases and more than deaths. regions with the most successful containment to date have approached the pandemic with integrated measures that include cohesive leadership, effective communication, physical distancing, wearing of face coverings, improvements in the built environment, promotion of hand hygiene, and support for the staff, supplies, and systems needed to care for patients-with testing and contact tracing as cornerstones of the approach. despite the emergence of some promising therapies and work towards a future vaccine, basic public health approaches remain the best available prevention and control interventions at this time. along with efforts to expand conventional contact tracing programmes, there has been an ongoing debate about the value of digital contact tracing, ranging from issues of privacy, questions about efficacy, lower user adoption rates, and concern from some public health experts that mobile apps might distract resources from the core work of conventional contact tracing. yet, in the face of ongoing challenges in disease control, the question of whether digital technologies can supplement existing efforts is one that we cannot afford to ignore. in the lancet public health, mirjam kretzschmar and colleagues model key steps in sars-cov- testing and contact tracing across a spectrum of scenarios and identify opportunities to maximise the effectiveness of the process in reducing the effective reproductive number of covid- . the study is important as initial large-scale physical distancing policies are relaxed and movement of people increases. research into how and where to best invest in improving systems of contact tracing is essential, as even those areas with low case burdens will face ongoing transmission events and must be prepared to quell outbreaks as they occur. not surprisingly, the authors conclude that speed is of the essence in testing and isolating: the study finds that keeping the time between symptom onset and testing and isolation of an index case at days or less is imperative for success in reducing the reproductive number, and that rapid testing of symptomatic people is at least as important as the efficiency of contact tracing. this study adds to the literature on the role of contact tracing in covid- and highlights the need for adequate testing capacity. the authors also suggest a meaningful contribution to contact tracing from mobile apps, which might minimise notification and tracing delays, although they do not consider a hybrid approach combining conventional and mobile app-based contact tracing. the authors make several assumptions that might blunt the impact of their findings: they assume that index cases are isolated with no further transmission, yet household transmission has been reported as important even when contact tracing was in place; that all traced contacts, regardless of symptoms, are offered testing, yet capacity to test remains an important challenge in many areas; and that those testing negative (once) do not spread infection, yet this might be an over-simplification of the sensitivity of tests and the dynamics of infectiousness. the importance of these assumptions could be tested in future research and modelling efforts, as could an analysis of a hybrid approach where exposure notification is used to support conventional contact tracing rather than replace it, which seems more likely in practice. a limitation of the study is the lack of detail on the mobile app technology in the model. while the researchers focus on uptake and speed of notificationstwo important parameters-there is a lack of discussion of the efficacy of an app in terms of its detector function (ie, the sensitivity and specificity of an app to determine if a contact event has occurred between two users ) and its effector function (ie, the effectiveness in contributing to the desired public health actions by the user, such as entering self-quarantine). the conclusion of the researchers, therefore, that "app-based tracing on its own remains more effective than conventional tracing alone, even with % coverage, due to its inherent speed" seems premature without a more nuanced discussion of efficacy and of the potential challenges and harms of digital approaches. this is not to claim that mobile apps are lacking in promise, but they do remain unproven as a public health intervention. therefore, as jurisdictions around the world roll out exposure notification apps, there are crucial questions that must be investigated to understand the efficacy of these apps and to make adjustments necessary to build user trust and adoption, if they are to make a contribution to pandemic response. first, how well do smartphones measure proximity? in other words, what is the effectiveness of the detector function and how many false alarms might be expected for each true contact detected? second, how will mobile apps integrate with overall contact tracing programmes? good contact tracing offers not just an epidemiological interventionquarantining enough individuals to reduce the reproductive number-but also a human one, that investigates outbreaks and understands linkages, and that recognises and addresses the challenges inherent in quarantine and isolation by providing a variety of supports. in our experience, success at this challenging endeavour requires public health workers as human beings to connect with a person, to build trust on a human level. these vital dynamics are not captured in epidemiological models, nor can we expect that notifications provided by a mobile app will fill the place of the detective work and supportive human interventions at the core of contact tracing. third, what factors will encourage users to trust the privacy and security properties of mobile apps? current adoption rates are low in every jurisdiction where apps have been deployed, with most peaking at download rates of about % of the population, and little data available about actual usage levels, which are likely to be lower. mobile app user behaviour depends on a subtle trust-benefit ratio calculation by users that is challenging to predict in advance. what is behind the public's decision to use or avoid these apps? do they have privacy or security concerns or question the benefit of the service? do they trust public health authorities with their data and do they trust the authorities' pandemic response? fourth, how will mobile apps affect health equity? to be successful in addressing the pandemic, any contact tracing system-conventional or digital-should be evaluated within a health equity framework to avoid perpetuating the deep disparities that the global pandemic has so glaringly exposed. as contact tracing remains a crucial component of the covid- response, mobile apps offer promise, especially when considering the speed and scale required for tracing to be effective-as highlighted in kretzschmar and colleagues' study. however, understanding the potential impact of apps as part of a comprehensive integrated approach requires more evaluation of their use in real life and multidisciplinary engagement of technologists, epidemiologists, public health experts, and the public. djw reports grants from the mit-ibm watson artificial intelligence laboratory and the us centers for disease control and prevention. lci reports grants from the bill and melinda gates foundation and the national institutes of allergy and infectious diseases, outside of the submitted work. djw is co-principal investigator and lci is senior medical advisor of the private automated contact tracing initiative, which is a collaboration led by the mit computer science and artificial intelligence laboratory, the mit internet policy research initiative, massachusetts general hospital center for global health, and mit lincoln pharmacologic treatments for coronavirus disease (covid- ): a review developing covid- vaccines at pandemic speed impact of delays on effectiveness of contact tracing strategies for covid- : a modelling study epidemiology and transmission of covid- in cases and of their close contacts in shenzhen, china: a retrospective cohort study months into virus crisis, us cities still lack testing capacity variation in falsenegative rate of reverse transcriptase polymerase chain reaction-based sars-cov- tests by time since exposure using bluetooth low energy (ble) signal strength estimation to facilitate contact tracing for virus-tracing apps are rife with problems. governments are rushing to fix them privacy tipping points in smartphones privacy preferences key: cord- - t incb authors: pförringer, dominik; ansorg, jörg; osterhoff, georg; dittrich, florian; scherer, julian; de jager, uwe; back, david a. title: digitalisierung in orthopädie und unfallchirurgie: stand in klinik und praxis date: - - journal: unfallchirurg doi: . /s - - - sha: doc_id: cord_uid: t incb this article deals with the current state mid- in the clinical and practical aspects from the perspective of orthopedics and trauma surgery. the risks, difficulties, potentials and options are discussed in detail. the following topics are specifically debated: infrastructure of telematics, apps and mobile health, online video consultation, electronic medical records and data protection. the advantages and disadvantages and the current state of each topic in the special case of orthopedics and trauma surgery are discussed. additionally, seven meaningful examples from the field of digital applications are named. a survey of members of the professional association of orthopedic and trauma surgeons (bvou) is described and analyzed. in a concluding perspective the current hurdles and future topics that need clarification are addressed. für eine rasche digitalisierung des deutschen gesundheitssystems machen sich seit der jahrtausendwende die gesundheitsminister stark. seit dem "gesetz zur modernisierung der gesetzlichen krankenversicherung" aus dem jahr ist auch die ärztliche selbstverwaltung aus kassenärztlicher bundesvereinigung (kbv) und bundesärztekammer in dieses umfangreiche projekt durch ihre beteiligung an der gesellschaft für telematikanwendungen der gesundheitskarte (gematik) eingebunden. neben der optimierung von behandlungsprozessen und patientenversorgung ist mit der digitalisierung des gesundheitssystems v. a. die hoffnung verbunden, die versorgungssektoren miteinander zu vernetzen und die in den letzten jahrzehnten betonierte sektorengrenze für informationen und optimierungsprozesse durchlässiger zu gestalten. nach jahren projektarbeit und milliardeninvestitionen sollen in diesem beitrag wesentliche aktuelle projekte und deren status reflektiert sowie konkrete anwendungsszenarien und erste umsetzungsbeispiele für orthopädie und unfallchirurgie (u und o) aufgezeigt werden. rückgrat für die digitale kommunikation zwischen Ärzten, praxen und kliniken sowie apotheken und weiteren gesundheitsberufen ist eine sichere digitale vernetzung. damit sollen medizinische informationen für eine effizientere patientenbehandlung schneller und vollständiger als im analogen zeitalter verfügbar und zugänglich gemacht werden. die telematikinfrastruktur (ti) wurde gemeinsam mit der ersten version der elektronischen gesundheitskarte (egk) seit von der gematik entwickelt und wird seit eingeführt. für den anschluss an die ti sind ein konnektor sowie ein kartenterminal für jede praxis erforderlich. vertragsärzte und praxen, die sich nicht an die ti anschließen, werden per gesetz mit sanktionen belegt. neben lieferengpässen war die einführungsphase von verunsicherung und sicherheitsbedenken aufseiten der Ärzte geprägt [ ]. beim anschluss an die ti [ ] kam auf, dass die it-infrastruktur vieler praxen erhebliche sicherheitsmängel aufweist [ ] . hinzu kamen mängel beim ausgabeverfahren der authentifizierungskarten [ , ] . das smartphone ist unbestreitbar eine der am kontroversesten diskutierten technologischen errungenschaften des . jh., die mit einer rasanten geschwindigkeit einfluss auf große bereiche unserer gesellschaft genommen hat. die liste der vor-und auch nachteile, die mit einer alltäglichen smartphone-nutzung einhergehen, ist lang und weitestgehend paritätisch besetzt. die vorzüge eines angemessenen einsatzes von apps im bereich der orthopädie und unfallchirurgie (ou) sind jedoch unbestreitbar und könnten ein enormes potenzial für die bewältigung zukünftiger herausforderungen im gesundheitswesen darstellen [ ] . auch politisch wurden mit dem kürzlich ratifizierten digitale-versorgung-gesetz (dvg) die weichen hin zu einer flächendeckenden implementierung von apps gestellt, u. a. durch die verschreibung von gesundheits-apps auf rezept. das dvg schafft im internationalen vergleich erstmalig einen leistungsanspruch auf digitale gesundheitsanwendung, beschränkt auf medizinprodukte niedriger risikoklassen. die aufnahme in das verzeichnis für digitale gesundheitsanwendungen erfolgt nach antragstellung durch den hersteller und prüfung, ob die gesundheits-app grundlegenden anforderungen an medizinprodukte und datensicherheit gerecht wird sowie positive versorgungseffekte aufweist [ ] . eine kritische auseinandersetzung der Ärz-teschaft mit dem neuen medium ist obligat. apps werden in deutschland aktuell von % der orthopäden und unfallchirurgen regelmäßig im klinischen alltag eingesetzt. aufgrund unscharf definierter juristischer, ethischer und medizinischer regularien bewegen sich viele apps jedoch in einer grauzone [ ] . vergangene datenskandale und eine intransparente datenverarbeitung haben zu einem grundmisstrauen gegenüber apps, die gesundheitsdaten erheben, geführt [ ] . eine der größten herausforderungen für deutsche Ärzte in o und u stellt bereits die suche nach der passenden app dar. die unübersichtlichkeit und dynamik der app stores lässt die suche nach der passenden app wie die suche nach der "stecknadel im heuhaufen" erscheinen [ ] . lösungsansätze können die stetig ansteigende zahl von app-store-basierten reviews kommerziell erhältlicher apps darstellen, die sich kritisch mit den derzeit verfügbaren apps auseinandersetzen sowie inhaltliche und methodische ansätze für die persönliche app-suche liefern [ ] . diese methoden könnten in zukunft um automatische algorithmische analysen oder künstliche intelligenz erweitert werden. form eines dgou-app-siegels ist aktuell in einem pilottest im einsatz neben der suche nach den passenden apps ist die objektive evaluation ihrer sicherheit und qualität problematisch (. infobox ). verschiedene testverfahren wurden entwickelt und werden ständig weiterentwickelt. etablieren konnten sich die auf die vertrauenswürdigkeit einer app fokussierende "app-synopsis" des peter l. reichertz instituts für medizinische informatik an der medizinischen hochschule hannover [ ] und die "mobile app rating scale" [ ] , die eine objektive bewertung des inhalts und der technischen spezifikationen einer app ermöglicht. eine qualitative verbesserung der evaluation könnte er-reicht werden, wenn zukünftige ratings adaptiver an die zweckbestimmung der app anpassbar würden. mehrere fachgesellschaften sowie private institutionen rezensieren apps nach einem standardisierten, transparenten und öffentlich zugänglichen fragenkatalog und veröffentlichen diese rezensionen in app-bibliotheken (z.b. https:// www.nhs.uk/apps-library/). ein entsprechendes verfahren eines app-siegels der deutschen gesellschaft für orthopädie und unfallchirurgie (dgou) ist auf grundlage definierter, evidenzbasierter gütekriterien aktuell in einem pilottest im einsatz [ ] . die abstract this article deals with the current state mid- in the clinical and practical aspects from the perspective of orthopedics and trauma surgery. the risks, difficulties, potentials and options are discussed in detail. the following topics are specifically debated: infrastructure of telematics, apps and mobile health, online video consultation, electronic medical records and data protection. the advantages and disadvantages and the current state of each topic in the special case of orthopedics and trauma surgery are discussed. additionally, seven meaningful examples from the field of digital applications are named. a survey of members of the professional association of orthopedic and trauma surgeons (bvou) is described and analyzed. in a concluding perspective the current hurdles and future topics that need clarification are addressed. mobile health · telecommunications · electronic medical records · data privacy · surveys ärztlich vertretbar ist, die ärztliche sorgfalt gewahrt bleibt und der patient über die besonderheiten der ovs aufgeklärt wurde. einen boom erlebt die ovs aktuell durch die coronapandemie. in zeiten, in denen physische arztbesuche eher vermieden oder wegen quarantäne nicht wahrgenommen werden können, sind telemedizin und insbesondere die ovs geeignete mittel zur aufrechterhaltung einer adäquaten arzt-patient-beziehung und für eine kontinuierliche betreuung des patienten. die integrierte videokomponente unterstützt, im gegensatz zum reinen telefonat, den persönlichen kontakt und kann auch zu diagnostischen zwecken eingesetzt werden. in der o und u können so aus der ferne z. b. wunden inspiziert, in rehabilitation befindliche patienten visitiert, bewegungsausmaße gemessen und dringende persönliche fragen beantwortet werden [ ] . durch einsatz der videosprechstunde können gesundheitskosten sowie (anfahrts-)kosten gesenkt und eine effizientere arzt-patient-interaktion etabliert werden [ ] . erste vergütungsgrundsätze für die videosprechstunde wurden im ebm [ ] und in der goÄ [ ] zentraler zukünftiger baustein der ti ist die epa. sie soll sektorenübergreifend die digitale dokumentation der krankheitsgeschichte gewährleisten. spätestens ab januar müssen die gesetzlichen krankenkassen ihren versicherten die epa anbieten [ ] . folgende informationen der versicherten sollen in der epa gespeichert werden: zunächst ist geplant, dass die zugriffsberechtigten leistungserbringer alle daten des patienten einsehen können, es sei denn, dass der versicherte sie löscht. ab sollen die versicherten ein abgestimmtes berechtigungsmanagement erhalten, in dem sie die in der akte enthaltenden daten gezielt für einzelne Ärzte und andere leistungserbringer freischalten können. der versicherte ist somit "herr" seiner daten. bei den Ärzten herrscht skepsis, da u. a. eine löschung von akteninhalten durch patienten Ärzten u. u. lebenswichtige informationen vorenthalten könnte [ ] . im rahmen einer nichtrepräsentative onlineumfrage des berufsverbands für orthopädie und unfallchirurgie (bvou) zur epa [ ] gaben im oktober nur % der befragten an, dass sie alle erhobenen befunde der patienten in die epa einstellen würden (. abb. ). zu fordern ist für die zukunft, dass für die valide nutzung der epa alle relevanten informationen, z. b. in einem unveränderlichem "arztbereich", zur verfügung stehen. alternativ wäre an eine kennzeichnung von befunden zu denken, die vom versicherten gesperrt/ gelöscht wurden. um schnittstellenprobleme zu vermeiden, ist eine gleiche semantik erforderlich. eine %ige kompatibilität zu allen gängigen praxis-/klinik-software-systemen mit einfacher bedienung sowie maximaler daten-und rechtssicherheit, besondere für den datentransfer zwischen den akteuren, muss gewährleistet werden. der aufwand für die einstellung und validierung der befunde kostet zeit und bedarf einer adäquaten vergütung, die durch die geplante einmalige -€-zahlung derzeit nicht gegeben ist. mit dem e-health-gesetz [ ] und dem dvg [ ] wurde vom gesetzgeber ein fahrplan für den weiteren ausbau der ti, die verbesserung der kommunikation verschiedener it-systeme im gesundheitswesen sowie für die einführung telemedizinischer leistungen (z. b. ovs) geschaffen. hinzu kommen regelungen für den zugang zu gesundheits-apps so- Ärzte und patienten stehen in einschlägigen umfragen digitalisierungsprojekten offen gegenüber. die eingesetzten lösungen müssen praktikabel sein und den anwendern einen klaren mehrwert bieten. dieser anforderung wird der aktuelle ausbaustand der ti nicht gerecht. sinnvolle anwendungen sind bislang nur angekündigt. viele niedergelassene kolleginnen und kollegen bedienen sich frei verfügbarer alternativangebote wie ovs, onlineterminvergabe und epa. sie stehen der staatlich verordneten digitalisierung skeptisch gegenüber. in der klinik ist die digitale prozessoptimierung bereits weit fortgeschritten. eine sektorübergreifende digitale kommunikation mit niedergelassenen Ärzten und von klinik zu klinik ist aber noch immer kaum möglich. in o und u wird der einstieg in die digitalisierung mit einer reihe von rahmenverträgen sowie durch integration in das patienteninformationsportal orthinform erleichtert. mit wenigen klicks können onlineterminvergabe, ovs, ressourcenmanagement usw. gebucht sowie in das persönliche arztprofil und die eigene webseite integriert werden [ ] . weiterhin wird der einstieg in die digitalisierung durch das aushandeln von selektivverträgen unterstützt, die eine bessere honorierung digitaler angebote wie beispielsweise der ovs oder der onlineterminvergabe garantieren. empfehlung zum anschluss an die telematikinfrastruktur (ti) informationsblatt der gematik warum eine komplette arztpraxis offen im netz stand. c't Ärztezeitung vom . telematikinfrastruktur: zugang mit identitätsdiebstahl Ärztetag beschließt liberalisierung der fernbehandlung. aerzteblatt.de vom . check my back" -selektivvertrag rückenschmerz gestartet. orthopädisch-unfallchirurgische mitteilungen und reputation und digitalisierung im internet mit orthinform synopsis -user-deutsche version einheitlicher kriterienkatalog zur selbstdeklaration der qualität von gesundheits gesetz für eine bessere versorgung durch digitalisierung und innovation (digitale-versorgung-gesetz) vom . dezember analysis of secure apps for daily clinical use as a german orthopaedic surgeon-searching for the "needle in a haystack apps in clinicaluseinorthopedicsandtraumasurgery:the status quo in germany ) iphone and ipad use in orthopedic surgery german mobile apps in rheumatology: review and analysis using the mobile application rating scale (mars) mobile app rating scale: a new tool for assessing the quality of health mobile apps cambridge analytica: the turning point in the crisis about. big data statistisches bundesamt (destatis) the opportunity awaits to lead orthopaedic telehealth innovation: aoa critical issues prospective randomized controlled trial using telemedicine for follow-ups in an orthopedic trauma population: a pilot study current low-cost video-based motion analysis options for clinical rehabilitation: a systematic review pkv: Ärzte erhalten extravergütungen in der coronakrise. aerzteblatt.de vom . . videosprechstunden in o&u: kann man oder muß man digitalisierungsreport der. dak, was Ärzte über die digitalisierung des gesundheitssystems denken interview mit ferdinand gerlach am online-umfrage zur epa. bvou, bd nutzen und akzeptanz von elektronischen gesundheitsakten. abschlussbericht zum forschungsvorhaben der bar-mer gek orthopäden und unfallchirurgen sollten sich weiter aktiv dem thema widmen und steuernd eingreifen, um die chancen und rahmenbedingungen der digitalisierung für ihre patienten und im interesse der heilkunst zum einsatz zu bringen. key: cord- -x yw authors: banskota, swechya; healy, margaret; goldberg, elizabeth m. title: smartphone apps for older adults to use while in isolation during the covid- pandemic date: - - journal: west j emerg med doi: . /westjem. . . sha: doc_id: cord_uid: x yw the maintenance of well-being, healthcare, and social connection is crucial for older adults (oa) and has become a topic of debate as much of the world faces lockdown during the coronavirus disease (covid- ) pandemic. oas have been advised to isolate themselves because they are at higher risk for developing serious complications from severe acute respiratory syndrome coronavirus. additionally, nursing homes and assisted-living facilities across the country have closed their doors to visitors to protect their residents. mobile technology such as applications (apps) could provide a valuable tool to help families stay connected, and to help oas maintain mobility and link them to resources that encourage physical and mental well-being. apps could address cognitive, visual, and hearing impairments. our objective was to narratively summarize apps that address physical and cognitive limitations and have the potential to improve oas’ quality of life, especially during social distancing or self-quarantine. in january , the first case of coronavirus disease was identified in the united states. shortly thereafter, visitation restrictions and guidance to reduce contact with older adults (oa), ≥ age , were put in place at many facilities caring for oas with the aim to protect them from infection. [ ] [ ] [ ] according to the world health organization, the case fatality rate for covid- in older adults in china years and older was . % compared to . % for people of all ages with no underlying health conditions. however, as many state and civic leaders are now debating lockdowns many oas may lack the assistance they need at home or in facilities to meet their daily needs. self-imposed and/or institution- in this narrative review of apps for oas, we aimed to find apps available to oas on the apple store that could potentially facilitate health during times of social distancing and/or selfquarantines. these apps were curated by a research team that included an emergency medicine attending and physician scientist in geriatrics and digital health, a medical student, a graduate student in biotechnology, and others. the apps are categorized by common healthcare needs within the oa population addressed by the following categories: ) social networking; ) medical, with subcategories a) telemedicine and b) prescription management; ) health and fitness; ) food and drink; and ) visual and hearing impairment. app categories were determined based on app categories already in place on the apple store, with the exception of a category to address the specific needs of oas with visual and hearing impairment, for which we did a custom search using the terms "blind" and "deaf" details about the app developer, cost (both to download and for services included in the app), function, ratings and reviews, and user experience (in the form of anecdotes) were searched and summarized. all app rating and review data was last updated to this article on march , . in the final list of apps, we aimed to include those that are either designed to target the oa population or have features that could benefit oas during pandemics and outbreaks when social isolation and/or self-quarantine is encouraged. apps with broad acceptability were given priority. hence, apps needed a rating of . or higher and at least reviews on the apple store. exceptions were given for apps with broad appeal and applicability to the objective, such as facetime, medisafe, and apps that assist people with vision and hearing impairment, as shown in figure . apps were further screened based on function and then ranking. users' experiences of the app were given consideration during the selection; hence, recent customer reviews that demonstrated that the app was a valuable product for an oa were selected and summarized as anecdotes. we conducted a literature review using pubmed and google scholar on the topic, but as many apps are not rigorously tested for usability and efficacy in the oa population, this selection was mainly based on expert review. we list several apps that assist oas with daily needs. these are summarized by cost and intended use in table . user ratings and reviews, in the form of anecdotes, are provided in table . many apps are available to help oas navigate isolation during the covid- pandemic. while not all of the apps on our list are marketed specifically to oas, we include apps with broad acceptability and positive user experience to ensure a list that helps access healthcare, maintain mental and physical the impact of social isolation on health could be as harmful as traditional risk factors such as high blood pressure, smoking, and obesity. even before covid- , % ( . million) of oas were living alone. social isolation has been linked to physical and cognitive conditions including heart disease, high blood pressure, anxiety, depression, alzheimer's disease, and a weakened immune system. fortunately, mt could provide a solution to isolation by enhancing the connection with loved ones in a safe and easy way, through apps such as facetime and skype. although mt cannot replace face-to-face interaction, it can still provide ease for those who feel a loss of connection. oas who use video chat apps, including facetime and skype, are estimated to decrease their symptoms of depression by half. in a survey of oas, those who use video chats were found to have lowered probability of depression symptoms, whereas depression rates among oas who use instant messaging and social media networks were similar to oas who do not use any communication technology. skype is the oldest video chat app that offers the widest device support, including for android, ios, windows phone, and blackberry. it can run on desktop software including windows pc and apple's macbook. nursing homes and oa living residences frequently use skype to connect oa residents to their loved ones, even though the app takes some explanation to learn the software so users can fully understand how to use it. additionally, per recent policy changes by the us department of health and human services (hhs) office for civil rights (ocr), medicare beneficiaries may have improved access to their medical providers through facetime and skype by approving reimbursement at the same rate for an in-person as a telemedicine visit. food and drink apps on the apple store can be a solution for vulnerable populations as users have access to same-day delivery services such as doordash and instacart, allowing them to remain in their homes and maintain social distance. doordash has implemented "no-contact delivery options" as a response to covid- . the app allows users to fill out personalized delivery instructions, requesting drivers to leave orders outside to avoid person-to-person contact. users have the ability to text pictures and/or descriptions to where drivers should place their order, which may be easier for some than typing due to the loss of dexterity with aging. due to the closure of many restaurants, individuals should verify that a restaurant is open before placing an order. instacart, a grocery delivery service, has seen a surge in demand for the month of march due to covid- , especially in states with an increased number of cases, and also promises dropoff delivery that minimizes contact. these apps can cater to the oa population by giving them the option to stay home or providing families with the option to order food for their older loved ones rather than deliver it on their own, if they themselves are in quarantine. appropriate apple store app categories: ) social networking; ) medical; apps with features that could benefit older adults during covid creation of a custom app category (apps for visual and hearing-impairment) due to relevance to vulnerable population and older adults apple store searched using terms "blind" and "deaf" app ratings ≥ . and ≥ , reviews on the apple store exceptions • facetime was included due to known popularity and use • medisafe was given an exemption from exclusion due to known beneficence from background literature search health, and meets oas' various social and functional needs during social distancing during the covid- outbreak. these apps could also provide oas fearing loss of independence a sense of purpose and control over their life and health. social isolation and self-quarantine, whether it is selfimposed, legally and/or institutionally mandated, can lead to negative impacts on an oa's mental and physical well-being. banskota et al. free to download app. app works with or without insurance and is available at reduced rates through many major health plans and large employers. the average cost of a video consultation copay with insurance is $ , and $ flat rate fee without insurance. per cms guidance, telehealth is covered at the same rate as in-person visits during the covid- crisis. provides face-to-face digital connection with a doctor, psychiatrist or psychologist through video on people's iphone or ipad; provides urgent care, behavioral health, preventive health, and chronic care management; provides services in many languages when appointment is scheduled. allows you to send "lightning-fast" video messages, enabling ondemand communication using sign language and visuals. medical apps: telemedicine apps (doctor on demand, teladoc, and k health: primary care) as a covid- response, hospitals and clinics across the country have started to defer elective appointments and surgeries. , oas may benefit from this restriction due to reduced exposure to the virus, but many have chronic health conditions that need to be addressed. telemedicine may provide a temporary solution for these needs. the centers for medicare & medicaid services' recent expansion of medicare coverage for telehealth services to its beneficiaries provides an alternative for in-person medical care, and the waiver of medicare's cost-sharing requirements for covid- will improve access to care. social networking "skype is easy and good to use in terms of functionality and interface. i use skype phone to call international phones because the rate is very reasonable" ("good and easy to use," ). medical apps: telemedicine teladoc . stars; k ratings; # in medical "this has become my go to for our family. we never have a long wait, the doctors are knowledgeable and we get our prescriptions right away. this service provides massive value" ("always reliable," ). k health: primary care . stars; k ratings; # medical "all of my kids were diagnosed with the flu. discovered this app and wow it was a lifesaver. spoke to the doctor and got my rx without having to leave the house" ("great for sick mom," ). doctor on demand . stars; k ratings; # medical user did not have to leave home to get an antibiotic prescription at a local pharmacy, and reported, "what a fantastic service!" ("amazing," ). goodrx-save on prescriptions . stars; k ratings; # in medical a patient was paying $ dollars for a prescription until they switched to goodrx. now they are only paying $ for the same medication ("saving $$$," ). medisafe medication management (medisafe) . stars; k ratings; # in medical "my wife just came home from hospital with medications from specialists and medication from a primary doctor. i struggled to keep up until i started this app" ("couldn't do without this app," ). health & fitness apps calm . stars; k ratings; # in health & fitness "i struggle with anxiety anyway, and with a pandemic upon us, i've enjoyed using calm as a tool. i've used it during the day to deepen my meditation and yoga" ("helpful," ). headspace: meditation & sleep . stars; k ratings; # in health & fitness "...headspace is always my go-to for high quality soothing meditations. it has helped me calm down in the covid- crisis, and headspace is none other" ("life-changing," ). yoga: down dog . stars; k ratings; # health & fitness user states "this app helped improve my physical and mental well-being. i was able to start to learn more about yoga, build core strength, and flexibility" ("great for beginners," ). glide-live video messenger . stars; k ratings; no ranking "i use this app fairly regularly to communicate via als. it works great and i love the many features" ("great for asl," ). will not conduct audits to ensure this), and that "providers must use an interactive audio and video telecommunications system that permits real-time communication." medical apps that provide telehealth could facilitate care "early during the course of an acute problem or chronic disease exacerbation," and provide healthcare access to those patients who have never had a prior correspondence with a provider. , these resources could be valuable to uninsured and undocumented oas in the us. , these platforms may also be viewed as an extra resource that provide patients, especially those living in medically underserved areas, where access to care is limited. these platforms can connect patients to remote physicians during emergency closures and during times of increased demand for medical services. for example, during hurricanes harvey and irma, doctor on demand offered visits for chronic conditions, advice, counseling, and refills, and back and joint concerns. doctor on demand, teladoc, and k health: primary care are options available on the apple store that provide access to licensed physicians for non-emergency medical problems and are health insurance portability and accountability act of (hipaa) compliant. [ ] [ ] [ ] doctor on demand and teladoc are considered leaders in telemedicine, and are covered by many insurances including unitedhealthcare, aetna, cigna, and some state medicaid programs, although coverage may be different, and different insurances have different preferred telehealth destinations. , [ ] [ ] [ ] [ ] it is also important to note that many states have made changes to their telemedicine license policies due to covid- . , the fact that our healthcare system was not equipped to provide telehealth on a mass scale for an outbreak is demonstrated by the waiver of penalties for hipaa violation for using "everyday communication technologies such as facetime and skype" to provide medical care during the covid- emergency. in contrast, smartphone apps we have listed that provide telehealth services ensure hipaa-compliant services, which may be preferred by some patients with privacy concerns. telemedicine has not always been embraced as a viable solution for patients. providers in these platforms do not have access to key information from physical examination and diagnostic testing; in addition, they lack access to care coordination and insight gained from longitudinal care. , however, telemedicine may be the only viable solution during covid- , and many experts predict oas could benefit long term from the improved access to care these platforms provide. telehealth clinicians have experience working with limited exam and diagnostics tools and should acknowledge when an actual visit is necessary due to the acuity of the condition or the need for an in-person exam or procedure. patients are generally satisfied with telehealth service use. , therefore, access to care during this time may contribute to reduction of anxiety and frustration, in addition to feelings of loneliness, in the oa population. it is important to note that racial disparity is known to exist in telemedicine access, as well as that the majority of current telemedicine users are younger adults. therefore, ensuring equity in telemedicine access is important during this crisis, along with special effort in introducing and orienting oas from underrepresented backgrounds. in adults years and older, more than % use two or more prescription drugs and % used five or more (called polypharmacy). furthermore, per the kaiser family foundation, "about one-fifth of older adults report[ed] not taking their prescribed medication as prescribed due to cost." goodrx provides discounts on medication, which could be particularly useful for oas with a limited budget or high out-of-pocket costs due to being on multiple medications. according to an aarp survey, % of midlife adults provided regular financial support for basic necessities to their parents regularly in , and more than a quarter of these adults reported that this caused them financial strain. hence, goodrx may be useful for adults financially supporting older parents, and for working americans laid off due to business shutdowns. this is also a time when family members and caregivers who typically visit oas and check on their medications are unable to do so because of social isolation and visitor restrictions at nursing homes and assisted living facilities. medisafe could help oas with trouble adhering to a medication regimen due to cognitive impairment or polypharmacy. self-reported medication nonadherence is common in community-dwelling older adults especially in those with cardiovascular disease. cardiovascular disease is a known risk factor for mortality among oas who contract covid- . medication nonadherence itself can be dangerous, as it contributes to more than % of hospital admissions in older adults, and is associated with increased incidence of heart failure. , hospital admissions may increase risk of exposure to covid- , and heart failure is associated with worse prognosis in oas with covid- . thus, oas should be especially careful about medication adherence during this pandemic to protect health. in one study, participants using medisafe had a small improvement in self-reported medication adherence. therefore, medisafe, along with its real-time missed medication alerts and frequent check-ins via phone calls by family members or healthcare providers, may help oas stay in the path of medication adherence. in , medisafe announced a partnership with goodrx to help lower medication costs. medisafe along with goodrx could help reduce barriers to medication adherence. oas are prone to worrying about their health. anxiety could be exacerbated during the covid- crisis. health anxiety has been found to be associated with more "distress, impairment, disability and health service utilization." this finding underscores the importance of curating apps targeting health applications for oas mental health. a study shows that oas are "motivated to use digital technologies to improve their mental health." in a study with participants aged - , frequent use of headspace for days was associated with improvement in mental health, specifically depressive symptoms and resilience. in another study among college students, students who used calm for eight weeks reported reduced stress. although there has been no published research looking at the effectiveness of using applications such ase calm and headspace in oas, these apps could be a useful tool to address anxiety. social isolation and quarantine can decrease physical activity and promote sedentary behavior, which is problematic in a population that already spends % of awake time engaged in sedentary activities. sedentary behavior is associated with disability in activities of daily living, development of metabolic syndrome, and an increased risk of all-cause mortality in the elderly. long duration of sitting is negatively associated with femoral bone mineral density (fbmd) in women, whereas duration of light intensity physical activity is positively associated with fbmd. physical activity intervention has been proven effective in improving physical activity behavior in healthy oas, and most sequences of yoga are classified as a light-intensity physical activity. , some small studies also suggest that, in oas, yoga may be superior to conventional physical-activity intervention. suggesting healthy oas to use an app such as yoga: down dog could reduce the ill-effects of sedentary behaviors. encouraging oa users to set a goal to pursue daily physical activity during social isolation and may serve as behavior intervention. yoga could protect psychological health in this difficult time, and help with sleep quality. , in a study in oas, chair yoga participants had more improvement in anger, anxiety, depression, well-being, general self-efficacy, and self-efficacy for daily living than control and chair exercise participants. chronic conditions common in oas, such as hypertension and diabetes, can be controlled with exercise and good diet. myfitnesspal, which provides a calorie counter and diet plan, could be a motivator for behavior change. myfitnesspal is a behavior intervention that could provide benefit of well-being, but it requires self-efficacy. , limitations of myfitnesspal include unreliable estimation of (micro-) nutrients ingestion and ineffectiveness in patients without goals and willingness to self-monitor calories. [ ] [ ] [ ] [ ] therefore, although myfitnesspal may be recommended to promote healthy behavior, oas should not use myfitnesspal by itself, and work in conjunction with a dietitian if possible. when asked about the vulnerable populations that have an increased risk of being affected by covid- , dr. lisa cooper of johns hopkins reported that individuals with vision and hearing impairments are also vulnerable. as of , an estimated four million oas had vision disability. vision impairments double the risk of falls, which one of four oas experience, and are associated with morbidity and mortality. oas with vision impairments who live alone and do not receive any caretaker service have to overcome greater challenges regarding activities of daily living and instrumental activities of daily living, which limits one's quality of life and independence. be my eyes, the largest online support for the visually impaired, may be a useful resource to these oas, especially at this time. , per be my eyes, over two million volunteers speaking over languages have signed up on the app to assist those with impairments, increasing acceptance, socialization, and independence for this population. , with the goal to help visually impaired individuals navigate through daily activities, volunteers have the ability to assist oas who do not have support at home by keeping them safe, enabling users to have a sense of independence and support. , an estimated one in three people between the ages - have difficulty hearing, with half of those older than having difficulty hearing. oas with hearing impairment have a greater chance of becoming depressed due to feeling frustrated and embarrassed about not understanding what is being said. howard a. rosenblum, chief executive officer of the national association of the deaf, stated that the us government must make information on covid- accessible in american sign language (asl), including information on how the virus affects education and employment access, among others. glide -live video messenger enables the ability to communicate to the hearing-impaired population through asl and/or just videos. this may negate feelings of loneliness and depression during times of social distancing for covid- . additionally, important information pertaining to disease characteristics, local and state business closures, financial updates, and other communications on covid- could be shared to those with hearing impairments effectively and promptly using glide. our summary of the apps, listed in figure , was based on the functionality of apps on the apple store primarily using the "top charts" list and expert opinion. rather than creating an exhaustive list, we focused on a brief list of apps that could be recommended to oas during the covid- pandemic. apple store is not accessible in all smartphones, and there is a far greater ownership rate of android devices compared to ios. however, except for facetime, the other apps on our list can also be found on google play store, the android app store. it is important to note that because app features may differ slightly on the two operating systems, user experience and ratings for the apps may vary between the two digital-distribution platforms. due to the limitations in our methodology, our apps list does not address the barriers faced by older adults with hearing impairments but without experience using sign language. for these older adults, live captioning apps such as ava, otter.ai, and microsoft translator may be suggested. these apps can be downloaded on both ios and android devices. while microsoft translator is a completely free, ava and otter.ai is free for occasional use, which limits users to hours/month and minutes/month, respectively. unlimited access can be purchased with a subscription to premium plans. it is also critical to acknowledge that while digital health and mt use by oas is increasing, few apps have been reviewed and tested for usability and efficacy in clinical trials among the oa population. in the future, additional research assessing the usability of these apps in the oa population using the mobile app rating scale, or other usability models such as the technology acceptance model, should be conducted. , however, many of the apps we have suggested fulfill an unmet need and could help oas maintain physical and mental health, independence, address disabilities, and some financial security. most importantly, they encourage and allow for a less imprisoning and isolating experience for oas during this crisis. apps are inexpensive and accessible, and research has shown that oas can use smartphones when provided the necessary training. there is an increase in the use of smartphones in the aging population. recommending these apps, along with providing some training and guidance, to an oa could help decrease loneliness and maintain and/ or improve the health and independence of oas during the covid- pandemic. while apps cannot substitute for all in-person care, they could supplement or substitute some inperson care. this publication was made possible by the national institute on aging (r ag ; k ag ), and the brown physicians, inc. academic assessment research award (pi: goldberg). the authors would like to thank armen deirmenjian and kunzhao li for support during the initial process of apps selection. nursing homes becoming islands of isolation amid 'shocking' mortality rate. the new york times families worried about loved ones in nursing homes amid coronavirus preparing for covid- :long-term care facilities, nursing homes covid- ) pandemic. world health organization mobile support for older adults and their caregivers: dyad usability study designing interpersonal communication software for the abilities of elderly users social isolation, loneliness and health in old age: a scoping review can digital technology enhance social connectedness among older adults? a feasibility study the effect of information communication technology interventions on reducing social isolation in the elderly: a systematic review getting connected: older americans embrace technology to enhance their lives sharing is caring: the potential of the sharing economy to support aging in place direct-to-consumer telemedicine loneliness, isolation and the health of older adults: do we need a new research agenda? loneliness, depression and sociability in old age social isolation, loneliness in older people pose health risks. national institute on aging using skype to beat the blues: longitudinal data from a national representative sample google? best app to call grandma medicare telemedicine health care provider fact sheet | cms food delivery companies offer 'social distance' with nocontact delivery options farmstead see demand surge amid covid- mayo clinic appointment guidance during covid- -mayo clinic key changes made to telehealth guidelines to boost covid- care | analysis of teladoc use seems to indicate expanded access to care for patients without prior connection to a provider direct-to-patient telehealth: opportunities and challenges the use of telehealth for serving the undocumented | cchp website. center for connected health policy hispanic migrant farm workers' attitudes toward mobile phone-based telehealth for management of chronic health conditions the promise of direct-to-consumer telehealth for disaster response and recovery virtual first responders: the role of direct-to-consumer telemedicine in caring for people impacted by natural disasters available at: https:// telemedicine.arizona.edu/servicedirectory/doctor-demand the evolving policy landscape of telehealth services delivered in the home and other nonclinical settings -issue brief. available at: www.milbank.org unitedhealthcare now covers doctor on demand, american well video visits too blue cross and blue shield of minnesota names doctor on demand as preferred telehealth destination. blue cross blue shield states waiving licensure requirements/renewals in response to covid- telemedicine spreads rapidly, but some physician groups worry about quality of care electronic visits for common acute conditions: evaluation of a recently established program patients' satisfaction with and preference for telehealth visits prescription drug use continues to increase: u.s. prescription drug data for data note: prescription drugs and older adults | the henry j. kaiser family foundation. kaiser family foundation a survey of midlife adults providing financial support to family members layoffs intensify leading to soaring unemployment claims as coronavirus business closures continue prevalence and correlates of self-reported medication non-adherence among older adults with coronary heart disease, diabetes mellitus, and/or hypertension are you at higher risk for severe illness? | cdc medication non-adherence among older adults: a review of strategies and interventions for improvement medication nonadherence: an unrecognized cardiovascular risk factor association of a smartphone application with medication adherence and blood pressure control: the medisafe-bp randomized clinical trial medisafe partners with goodrx to help app users lower medication costs -medisafe worry and use of coping strategies among older and younger adults health anxiety in australia: prevalence, comorbidity, disability and service use older adults' perspectives on using digital technology to maintain good mental health: interactive group study mobile mindfulness meditation: a randomised controlled trial of the effect of two popular apps on mental health efficacy of the mindfulness meditation mobile app "calm" to reduce stress among college students: randomized controlled trial smartphone applications for mindfulness interventions with suicidality in asian older adults: a literature review amount of time spent in sedentary behaviors in the united states sedentary behavior and health outcomes among older adults: a systematic review associations between objectively-measured sedentary behaviour and physical activity with bone mineral density in adults and older adults, the nhanes study interventions to increase physical activity among older adults: a meta-analysis a systematic review of the energy cost and metabolic intensity of yoga the effects of yoga on physical functioning and health related quality of life in older adults: a systematic review and meta-analysis effects of yoga on psychological health in older adults more than i expected": perceived benefits of yoga practice among older adults at risk for cardiovascular disease most common chronic diseases -healthy aging blog | ncoa. ncoa. available at the effects of choice and enhanced personal responsibility for the aged: a field experiment in an institutional setting the relative validity of a food record using the smartphone application myfitnesspal popular nutrition-related mobile apps: an agreement assessment against a uk reference method comparing self-monitoring strategies for weight loss in a smartphone app: randomized controlled trial effectiveness of a smartphone application for weight loss compared with usual care in overweight primary care patients the use of a food logging app in the naturalistic setting fails to provide accurate measurements of nutrients and poses usability challenges cnnnewsroom interview: anchor brooke baldwin interviews johns hopkins' dr vision health & age| risk | vhi | cdc. available at be my eyes" app lets volunteers help bring sight to blind and low-vision people. wtrk be my eyes app connects blind people with sighted volunteers. the washington post hearing loss: a common problem for older adults. national institute on aging national association of the deaf -nad. national association of the deaf mobile app rating scale: a new tool for assessing the quality of health mobile apps technology, gaming, and social networking older adults experiences of learning to use tablet computers: a mixed methods study technology use among seniors | pew research center key: cord- - k edc authors: jahnel, tina; gerhardus, ansgar; wienert, julian title: digitales contact tracing: dilemma zwischen datenschutz und public health nutzenbewertung date: - - journal: datenschutz datensich doi: . /s - - - sha: doc_id: cord_uid: k edc seit juni gibt es in deutschland die corona-warn-app als sog. contact-tracing-app, welche dank ihres hohen datenschutzniveaus eine breite akzeptanz in der bevölkerung genießen soll. der hohe grad des datenschutzes hat allerdings zur folge, dass möglichkeiten zur datenerhebung stark eingeschränkt sind und damit aussagen über die wirksamkeit und potentielle risiken einer solchen app nur in äußerst geringem maße möglich sind. zukünftig sollte sichergestellt werden, dass neben der einhaltung datenschutzrechtlicher prinzipien auch daten zur bewertung des nutzens von contact-tracing-apps erhoben werden können. dafür sollten die nutzer*innen von anfang an einbezogen werden, die kommunikation transparent und klar erfolgen und ergebnisparameter klar definiert werden. gleichzeitig sprach sich in der umfrage eine mehrheit der teilnehmer dafür aus, dass die durch die app erhobenen daten in anonymisierter form der forschung zur verfügung gestellt werden sollten. hohe datenschutzstandards können das vertrauen positiv beeinflussen. umgekehrt verhindert ein hohes datenschutzniveau aber das sammeln und auswerten von epidemiologisch relevanten daten und erschwert damit die wirksamkeitsbewertung aus public-health-perspektive. bei der entwicklung einer contact-tracing-app ist daher eine abwägung zwischen dem schutz der privatsphäre und dem potentiellen public-health-nutzen nötig. bei der deutschen corona-warn-app wurde ein sehr hohes datenschutzniveau angesetzt, um die akzeptanz und nutzungsbereitschaft zu erhöhen. die downloadzahlen zeichneten zunächst ein positives bild: in den ersten drei wochen nach der veröffentlichung der app am . juni wurde die app rund , millionen mal heruntergeladen. danach kamen allerdings nur noch wenige neue downloads dazu (zahl: , millionen, stand: . . ) ( ), auch wenn die ergebnisse einer wiederholten online-befragung von internet-nutzer*innen durch die tu münchen und die initiative d zeigen, dass die akzeptanz für die deutsche corona-warn-app seit juni generell gestiegen ist ( ) . insbesondere ist die sorge vor staatlicher Überwachung durch die app zurückgegangen: bei der befragung im august befürchteten weniger menschen, dass die app ungewollt private daten sammelt und weitergibt als im juni (juni: %; august: %). allerdings denkt nur eine minderheit, dass die corona-warn-app einen relevanten positiven effekt auf den verlauf der pandemie in deutschland haben wird: nur noch prozent erwarteten im august, dass sich mithilfe der app die zahl der neuinfektionen reduzieren lässt (juni: %). etwa die hälfte ( %) der befragten stimmte der aussage zu, dass die corona-warn-app nichts ändern wird, im juni gaben dies nur % an. es ist allerdings nicht klar, ob die niedrigere downloadrate durch die gesunkene nutzenerwartung verursacht wird. die eigentliche nutzungsrate mag in der realität auch viel geringer als die downloadrate sein, da mehrfaches runterladen für das gleiche smartphone nicht getrennt ausgewiesen wird. auch wird nicht erhoben, wie viele der heruntergeladenen apps auch tatsächlich dauerhaft in betrieb sind. eine geringere nutzung würde wiederum den potentiellen nutzen der app weiter senken. um den nutzen messen und bewerten zu können, müssen daten zur verfügung stehen, welche aufgrund der gewählten datenschutzstandards nur in äußerst eingeschränktem maße zur verfügung stehen. der vorliegende beitrag beschäftigt sich mit dem spannungsfeld zwischen der nutzenbewertung einer contact-tracing-app und dem datenschutzniveau. speziell wird auf kriterien der nutzenbewertung im sinne der evaluationsforschung und den hierfür notwendigen daten eingegangen. auf dieser basis diskutieren wir, ob es mittelwege geben könnte, bei denen der schutz der individuellen daten gewährleistet ist und gleichzeitig public-health-relevante daten zur nutzenbewertung der app gesammelt und ausgewertet werden können. grundsätzlich kann die datenverarbeitung von contact-tracing-apps zentral oder dezentral erfolgen, mit jeweils unterschiedli-chen implikationen für den schutz der persönlichen daten ( ) . in der funktionalität sind sich beide ansätze ähnlich: mittels bluetooth low energy (ble) sendet ein smartphone seine eigene geräte-id an nahe gelegene smartphones während es gleichzeitig die ids der anderen geräte empfängt und über einen definierten zeitraum speichert. zusammen mit der id wird der abstand zwischen den geräten und die dauer des kontakts zwischen den einzelnen smartphones erfasst. wurde eine person positiv auf das virus getestet, wird eine infektionswarnung an sämtliche geräte versendet, die über einen bestimmten zeitraum kontakt mit dem smartphone der infizierten person hatten. beide konzepte nutzen einen zentralen server, allerdings in unterschiedlichen intensitätsgraden. beim zentralen ansatz erstellt der server mit erfolgter registrierung durch die nutzer*innen eine permanente id für das entsprechende gerät. dabei handelt es sich um pseudonymisierte und damit personenbezogene daten. im infektionsfall werden sämtliche vom smartphone der infizierten person gesammelten daten auf den server geladen, die infektionsgefahr eingeschätzt und die mit der infizierten person in kontakt gekommenen app-nutzer*innen informiert. die dafür notwendigen geräte-ids der fremden nutzer*innen liegen bereits griffbereit auf dem server. Ändern sich erkenntnisse zum infektionsgeschehen, kann das auf einem zentralen server schnell umgesetzt werden. beim dezentralen ansatz hingegen erfolgen sämtliche schritte zur datenverarbeitung nur auf dem jeweiligen smartphone. im gegensatz zu den permanenten ids im zentralen ansatz werden hierbei flüchtige geräte-ids zur identifizierung verwendet, die direkt auf dem smartphone generiert werden und sich nach wenigen minuten aktualisieren. an den zentralen server wird ausschließlich die flüchtige geräte-id der infizierten person kommuniziert. die smartphones sämtlicher app-nutzer*innen gleichen die liste der ids der positiv getesteten personen mit den eigenen ids ab. die auswertung und warnung vor einem infektionsrisiko der nutzer*innen erfolgt dezentral auf dem jeweiligen smartphone. bei einer dezentralen lösung müssen die nutzer*innen, sobald es eine neue app-version gibt, selbst aktiv werden und die updates auf ihre geräte spielen. neue erkenntnisse zum infektionsgeschehen in form von app-aktualisierungen müssen dann erst auf allen geräten ankommen, sonst kann es zu unterschiedlichen risikoeinschätzungen und daher auch zu warnungen auf basis unterschiedlicher parameter kommen. der dezentrale ansatz bedeutet eine höhere datensouveränität und -sicherheit im vergleich zu einem zentralen datenverarbeitungsansatz, da alle persönlichen daten stets auf dem eigenen gerät verbleiben und die kontakt-ids ausschließlich lokal und anonym verknüpft werden. der nachteil des dezentralisierten ansatzes ist, dass die art der nutzung oder die auswirkungen der nutzung nicht bewertet werden können, da der zugang zu den daten limitiert ist. auch wenn modellierungsstudien die theoretische wirksamkeit von contact-tracing-apps in verschiedenen situationen gezeigt haben, bleibt eine nutzenbewertung im rahmen einer empirischen evaluation der entwickelten contact-tracing-app notwendig, um ) mehr über die ausbreitung von sars-cov- zu erfahren und ) zu bewerten, welchen anteil die app daran hat, infektionsketten zu unterbrechen. der mehrwert der app innerhalb des gesamten kontaktverfolgungssystems könnte u.a. anhand folgender parameter bewertet werden: (a) wie viele der durch die app benachrichtigen personen durch manuelle kontaktverfolgung nicht gefunden worden wären, (b) welcher anteil der durch die app benachrichtigen personen im vergleich zu den aktuellen durchschnittswerten tatsächlich positiv auf covid getestet wird, (c) wie viel schneller personen über die app benachrichtigt werden als durch manuelle kontaktverfolgung. bei einem dezentralen ansatz stehen dafür allerdings nur in begrenztem maße daten aus der app selbst zur verfügung. vor dem hintergrund der nutzenbewertung und nachzuweisenden wirksamkeit von contact tracing apps, wie der corona-warn-app, stellt sich die frage, was als ein positiver eff ekt definiert und gemessen werden kann. aufschluss hierzu bietet unter anderem die evaluationsforschung, welche im allgemeinen als die systematische anwendung empirischer forschungsmethoden zur bewertung des konzeptes, des untersuchungsplanes, der implementierung und der wirksamkeit sozialer interventionsprogramme verstanden wird ( ) . zentrale aspekte sind dabei, dass der evaluationsgegenstand, hier die corona-warn-app, hinsichtlich seiner wirksamkeit überprüft werden soll -es soll geprüft werden, ob die intendierten ziele der corona-warn-app auch erreicht wurden. entsprechend erfüllt eine evaluation hier primär eine kontrollfunktion, aber auch eine entscheidungs-(z.b. "soll die corona-warn-app weiter betrieben oder sogar ausgebaut werden?") bzw. legitimationsfunktion (z.b. "war das aufb ringen öff entlicher gelder im gegebenen umfang angemessen?") ( ) . um an diesen evaluationszielen arbeiten zu können, ist es erforderlich, dass die merkmale und ziele des evaluationsgegenstandes konkretisiert und objektivierbar gemacht werden, wofür prinzipiell daten benötigt werden. entsprechend der zielsetzung der corona-warn-app wäre eine verbesserte kontaktpersonennachverfolgung ein wichtiges ergebnis, was sich z. b. an einer wirksamen unterbrechung von infektionsketten widerspiegeln würde. weitere voraussetzungen und eine konkretisierung der ziele für eine evaluation der corona-warn-app können rahmenkonzepte wie das evidence standards framework for digital health technologies (dht) des national institute for health and care excellence (nice) ( ) bieten. hierbei werden digitale technologien und evaluationsanforderungen aufgrund ihres grades der interaktion klassifi ziert, was dazu führt, dass sich konkrete anforderungen je nach ausgestaltung der corona-warn-app unterscheiden können, wobei den anforderungen der niedrigeren stufe(n) ebenso entsprochen werden muss wie in ggf. höheren stufen, in denen sich die technologie verorten lässt (abbildung ). aufgrund eines aktuell nicht sehr hohen interaktionsniveaus der corona-warn-app mit den nutzer*innen der anwendung ließe sich diese aller voraussicht nach auf der ebene des ni- . auch kann man nicht ausschließen, dass ältere smartphones im umlauf sind, welche nicht mit der app kompatibel sind. ebenfalls ist nicht bekannt, wie oft die app wieder deinstalliert wurde. bezugnehmend auf eine bekannte simulationsstudie der oxford university, welche aufzeigt, dass contact-tracing-apps als ergänzende strategie bereits ab einer nutzer*innenquote von % in der gesamtbevölkerung als unterstützende maßnahme, zusätzlich zu hygienemaßnahmen, in der pandemiebekämpfung wirkung entfalten können ( , ) , kann die derzeitige nutzer*innenquote als positiv bewertet werden. allerdings würde dies zu den oben bereits genannten annahmen zusätzlich voraussetzen, dass die app korrekt und durchgängig genutzt wird und positive testergebnisse auf einen corona-test auch über die app gemeldet werden. es muss davon ausgegangen werden, dass unter realen bedingungen eine deutlich geringere nutzer*innenquote erreicht wurde, weshalb die hier vorgestellten schätzungen von einer wissenschaftlichen evaluationsperspektive als kritisch und aufgrund der hohen ungewissheit als wenig aussagekräftig anzusehen sind. wie viele menschen andere tatsächlich gewarnt haben und wie viele über die app darüber informiert wurden, dass sie möglicherweise einen oder mehrere risikokontakte hatten, kann nicht genau gesagt werden. dies ist auf den datensparsamen dezentralen ansatz der app zurückzuführen. nimmt man beispielsweise an, dass % der , millionen downloads auch aktive nutzer*innen sind, wären das , % der deutschen bevölkerung. geht man nun davon aus, dass sich app-nutzer*innen genauso oft infizieren wie nicht-nutzer*innen, müssten seit einführung der app also , % der infektionen über die app gemeldet worden sein, sofern jedes positive testergebnis auch in die app eingegeben worden wäre. seit veröffentlichung der app (stand . . ) ( ) wurden insgesamt . neuinfektionen gemeldet. von diesen , % sind . fälle, die über die app hätten gemeldet werden sollen. wie viele positive testergebnisse tatsächlich in die app übermittelt wurden, ist unbekannt. geht man allerdings von der anzahl der ausgegebenen teletans aus (die anzahl der ausgegebenen qr-codes ist nicht bekannt), wur-den bislang . positive testergebnisse an die app übermittelt (stand . . ) ( ) . das sind , % der fälle, die potentiell an die app hätten übermittelt werden können. das würde bedeuten, dass nur rund ein fünftel der positiv getesteten nutzer*innen ihr testergebnis auch an die app übermittelt hätten. da aufgrund des dezentralen ansatzes der app keine daten zu den benutzten teletans und qr codes zur verfügung stehen, ist nicht bekannt, wie viele alarme tatsächlich ausgelöst und mögliche infektionsketten dadurch unterbrochen worden sind. unter dem strich steht, dass nicht genau gesagt werden kann, wie wirksam die corona-warn-app ist. aufgrund der geschätzten zahl der nutzer*innen und der vorliegenden simulation der oxford university könnte diese aber einen hypothetischen mehrwert als unterstützende maßnahme in der pandemiebekämpfung haben. zwar wurden durch eine generelle neuausrichtung der corona-warn-app von einem zentralen ansatz hin zu einem dezentralen ansatz anfängliche bedenken von experten hinsichtlich des datenschutzes ausgeräumt und somit das vertrauen der bevölkerung in die app vermutlich erhöht, allerdings ging dies zulasten einer evaluation der corona-warn-app. relevante evaluationsziele können nicht messbar ausgewertet werden und evaluationsanforderungen, wie z. b. im nice framework definiert, können aufgrund fehlender daten und informationen nicht erfüllt werden. es lässt sich festhalten, dass die corona-warn-app ohne die erhöhten datenschutzmaßnahmen womöglich weniger vertrauen und akzeptanz in der bevölkerung und damit deutlich weniger downloads erzeugt hätte. damit zeigt die corona-app ein dilemma auf, das auch andere digitale gesundheitliche technologien aufweisen. ohne (berechtigtes) vertrauen der nutzer*innen in den datenschutz fehlt die akzeptanz, die für eine breite nutzung notwendig ist. umgekehrt kann ein hohes level im datenschutz dazu führen, dass die technologie weniger nutzenbringend ist und/oder ihre effektivität nicht bestimmt werden kann. dies ist ein besonderes problem bei technologien, die ein gewisses mindestmaß an nutzer*innenraten aufweisen müssen, um wirksam zu sein -eine typische voraussetzung bei maßnahmen, die auf die bevölkerung und nicht nur auf individuen abzielen. um in zukunft besser vorbereitet zu sein, sollten lehren aus der entwicklung der corona-warn-app dieser pandemie gezogen und der frage nachgegangen werden, ob es einen mittelweg für eine contact-tracing-app gibt, die den datenschutzanforderungen entspricht, es aber gleichzeitig erlaubt, epidemiologischrelevante daten auf freiwilliger basis zu sammeln. zum einen ist der einbezug der nutzer*innen von anfang an notwendig, um die präferenzen und bedürfnisse potentieller nutzer*innen und verschiedener nutzer*innengruppen zu ermitteln. darauf aufbauend ist eine klare und transparente aufklärung und kommunikation darüber notwendig, was mit den gesammelten daten geschieht und wie sie gesichert beziehungsweise vor missbrauch geschützt werden. damit kann man den datenschutzbedenken besser gerecht werden und einen kompromiss zwischen datenschutz und datenerhebung herstellen. weiterhin ist die entwicklung eines evaluationsrahmens und implementierungsplans zwingend notwendig. dafür müssen unter einbezug realistischer annahmen über downloads, nutzung etc. vorab klare definitionen der ergebnisparameter erstellt sowie analyserahmen und modellan-nahmen formuliert werden. ein abbruch der entwicklung einer app sollte erfolgen, wenn sich herausstellt, dass die notwendigen voraussetzungen (vertrauen, nutzungsbereitschaft und bereitschaft, daten für forschungszwecke bereitzustellen) für den erfolgreichen einsatz der app nicht erfüllt werden können. ob und in welchem ausmaß die corona-warn-app ihren zweck, infektionsketten zu unterbrechen, erfüllt hat, ist aufgrund ihrer datensparsamen konstruktion unbekannt. das bedeutet jedoch nicht, dass die app per se ein misserfolg sein muss. wie die modellierungsstudien gezeigt haben, hat digitales contact-tracing das potential, das infektionsgeschehen positiv zu beeinflussen, aber dieses potential muss auch gemessen werden können. wir bedanken uns für die unterstützung durch den leibniz wis-senschaftscampus bremen digital public health (lsc-diph.de), der gemeinsam von der leibniz gemeinschaft (w / ), der freien hansestadt bremen und dem leibniz-institut für präventionsforschung und epidemiologie -bips gefördert wird. quantifying sars-cov- transmission suggests epidemic control with digital contact tracing im spannungsfeld zwischen sicherheit und freiheit. hmd praxis der wirtschaftsinformatik akzeptanz app-basierter kontaktnachverfolgung von covid- infektionsketten digital unterbrechen mit der corona-warn-app egovernment monitor . corona-warn-app: einstellungen und akzeptanz der bevölkerung apps zum contact tracing bei covid- -fällen evaluation: a systematic approach evidence standards framework for digital health technologies configurations of a digital contact tracing app: a report to nhsx key: cord- -oyyj bl authors: parker, michael j; fraser, christophe; abeler-dörner, lucie; bonsall, david title: ethics of instantaneous contact tracing using mobile phone apps in the control of the covid- pandemic date: - - journal: j med ethics doi: . /medethics- - sha: doc_id: cord_uid: oyyj bl in this paper we discuss ethical implications of the use of mobile phone apps in the control of the covid- pandemic. contact tracing is a well-established feature of public health practice during infectious disease outbreaks and epidemics. however, the high proportion of pre-symptomatic transmission in covid- means that standard contact tracing methods are too slow to stop the progression of infection through the population. to address this problem, many countries around the world have deployed or are developing mobile phone apps capable of supporting instantaneous contact tracing. informed by the on-going mapping of ‘proximity events’ these apps are intended both to inform public health policy and to provide alerts to individuals who have been in contact with a person with the infection. the proposed use of mobile phone data for ‘intelligent physical distancing’ in such contexts raises a number of important ethical questions. in our paper, we outline some ethical considerations that need to be addressed in any deployment of this kind of approach as part of a multidimensional public health response. we also, briefly, explore the implications for its use in future infectious disease outbreaks. in this paper we discuss ethical implications of the use of mobile phone apps in the control of the covid- pandemic. contact tracing is a well-established feature of public health practice during infectious disease outbreaks and epidemics. however, the high proportion of pre-symptomatic transmission in covid- means that standard contact tracing methods are too slow to stop the progression of infection through the population. to address this problem, many countries around the world have deployed or are developing mobile phone apps capable of supporting instantaneous contact tracing. informed by the on-going mapping of 'proximity events' these apps are intended both to inform public health policy and to provide alerts to individuals who have been in contact with a person with the infection. the proposed use of mobile phone data for 'intelligent physical distancing' in such contexts raises a number of important ethical questions. in our paper, we outline some ethical considerations that need to be addressed in any deployment of this kind of approach as part of a multidimensional public health response. we also, briefly, explore the implications for its use in future infectious disease outbreaks. as we write this paper, europe is at the epicentre of the covid- pandemic. the pandemic has its origins in the emergence, late in , of a novel coronavirus in the chinese city of wuhan, which has a population of around million. it is estimated that between the official confirmation of the outbreak and the imposition of a lockdown, around million people left the city. the vast majority went to other parts of china. the epidemiological implication of this is that the chinese population outside wuhan came into contact with many more people infected with covid- than did the world outside china. despite this, as of april , around months later, china's total number of cases is , and its daily case rate is close to zero. by contrast, the global total of cases is now approaching million and doubling every few days in many places. compared with other countries, china has been very successful at controlling the spread of there are a number of features of china's response to covid- that would be unlikely to be effective or acceptable in other countries. this does not mean that there are not important lessons to learn from china's success. one element of the approach i there is debate about the accuracy of the figures coming out of china but broad agreement about the success of their intervention in reducing the number of infections. adopted by china and by several other countries in east and south east asia that has been highly successful in reducing cases is the use of mobile phone data combined with intensive testing programmes. there is evidence to suggest that the use of this kind of approach might be successfully transferable to other settings with different political and cultural systems. ii effective, rapid contact tracing is the cornerstone of effective public health response in the face of infectious disease outbreaks. its success depends on identifying cases (usually people with symptoms) quickly, gathering information from them about recent contacts and following up and quarantining those contacts to interrupt further transmission of the disease. covid- presents a problem for contact tracing as usually practiced because around % of transmissions happen early in infection, before symptoms start, and before test results can be acted on. this means that covid- moves too quickly through the population to be amenable to standard contact tracing methods. the use of a mobile phone app that captures 'proximity events'-events in which two mobile phones have been close enough for sufficient time for the risk of infection to be high-offers the potential for instantaneous contact tracing from the moment the infection is confirmed. iii this has the potential to stop the pandemic. the modelling for the use of a mobile phone app in covid- and a more detailed description of how this might work have been published elsewhere. a number of different approaches are currently under development by health systems in many countries around the world. in this paper, our aim is to set out a number of ethical considerations relevant to the use of mobile phone apps to enable rapid contact tracing. these issues will emerge in different ways in different settings. any consideration of the ethical questions arising in the context of the covid- pandemic has to place great importance on the moral significance of its international spread and the massive scale ii the effectiveness and reach of any implementation of the app in democratic societies will inevitably be affected by varying configurations of state-citizen relationships, as well as by the roles of civil society groups and non-governmental actors. iii the question of what constitutes adequate information about infection status for a population effect may be answered differently by different systems, ranging from self-reported symptoms through to clinically validated test results. current controversy of its impact. as of april , there have been confirmed cases and deaths globally. these figures are likely to be significant underestimates. it is important to highlight the fact that in addition to those who have died very much larger numbers of people will be suffering symptoms sufficiently serious to warrant hospitalisation and intensive care. in lowincome and middle-income countries in which health systems will often not have these facilities, the impact will be much greater. we are far from the end of the covid- pandemic: these numbers will continue to rise for quite some time. it hardly needs saying that the saving of lives and reduction of suffering are of immense moral importance and there are strong reasons to support efforts to achieve this. the ethical assessment of an innovation capable of making a contribution to addressing these harms needs to be understood and analysed against the dramatic scale of the deaths and suffering represented by these data. the policy decisions made by governments around the world in response to covid- have been inevitably varied. what is possible, what is required and what is socially and culturally appropriate will differ across the globe. such differences notwithstanding, many countries have introduced significant restrictions on freedom of movement with disruption to everyday life. one-third of the world's population is currently living under 'lockdown'. the terms and enforcement of this vary but all are causing serious economic and other harms to both individuals and institutions with long-term impact. their impact will be enduring. many people now and in the future will experience significant suffering as a consequence of these measures. in the context of public health emergencies, actions are often justified that would not be appropriate outside of such contexts. such actions do nonetheless require an explicit justification: the mere existence of an emergency does not in itself legitimise any intrusion on the autonomy or privacy of individuals or groups. the justification most commonly offered for the current imposition of lockdowns and other restrictions of movement has been that they are necessary to ensure sufficient 'physical distancing' to disrupt the transmission of the infection sufficiently to enable health systems to cope with predicted demand. it is estimated that the overwhelming of health systems, were it to happen, would be one of the main causes of death. current approaches to lockdown are, however, blunt tools applied at a national level. they apply to everyone, whether or not they are at risk, affected or immune. this is justified insofar as there is insufficient accurate, reliable information about the risk status of individuals or specific locations, which would enable more finely-tuned decisions to be made reliably. the justification of blanket lockdowns would be weaker were it possible to manage physical distancing in a more evidencebased, risk-adjusted way. were this so, it would remain the case that limiting the movements of those people who presented a high risk would be justified. it would not, however, be justified to restrict the movements of those individuals (and possibly populations) who were reliably known not to be contributing to this risk. rapid contact tracing enabled by the mobile phone app described above-combined with accurate testing-has the potential to be a tool of this kind. the evidence suggests the app has the potential to enable some (likely many) people to return more quickly to their lives. this evidence puts pressure on justifications for blanket lockdowns. the harms presented by such lockdowns also provide support for an argument that the development and implementation of the app as part of a broader package of public health interventions is not only ethically acceptable but also-where feasible-obligatory. iv the app is preferable to blanket lockdowns because intelligent physical distancing constitutes the minimum imposition compatible with addressing the epidemic safely. a fuller analysis would require the relative benefits and harms of other mooted options for non-pharmaceutical intervention to be compared and considered. controlled or delayed spread of sars-cov- with the primary intention of mitigating against overburdened healthcare resources, herd-immunity by controlled infection in the population, and cyclical lockdowns, have all been considered. mathematical modelling can be used to compare the likely reductions on the morbidity and mortality, alongside any societal costs of quarantine, mediated by each intervention, of note, of the options under consideration, however, only contact tracing aims to prevent transmission while explicitly minmising numbers of people in quarantine. v before the pandemic, questions about data protection, security and privacy were at or close to the top of lists of ethical concerns for many people. against that background, the use of a mobile phone app built on the gathering and sharing of proximity information, even if pseudonymised, may be seen as deeply concerning, particularly in combination with other socially restrictive measures. two important questions requiring clarification in this regard are: what is the nature of the infringement of privacy, if there is one, and, can this be justified in the context of the covid- pandemic? starting with the question of justification, it seems clear now that some privacy infringements are potentially justifiable where they have the potential to contribute to the saving of many lives and reducing enormous suffering. imagine a scale running from to . at the end of the scale would be someone (person a) for whom privacy is the concern that trumps all others. people at this end of the scale would place privacy above all other concerns and would be unwilling to give up any privacy to achieve another goal, no matter how important. a person at the other end of the scale would be someone (person z) who has no interest at all in privacy and would willingly give up % of their privacy for any reason. person a's view is likely to be a minority position with regard to this pandemic. the scale of the suffering caused by the covid- pandemic means that if a case can be made that some degree of privacy infringement will save significant numbers of lives and reduce suffering, the intervention may be justified. any such justification will depend on a clear case being made that the privacy infringement is either necessary or that it is significantly more effective than the alternatives. one aspect of a convincing attempt at justification might be the claim that the privacy infringement is less intrusive than blanket population level lockdowns for everyone. it would, however, also require a convincing case to be made that (i) any privacy impact would be minimised, (ii) that high standards of data security, protection and oversight would be in place, (iii) that there would be transparency about proposed and actual data uses, and (iv) that these would be complemented by other protections, for example, around non-discrimination. this is a iv subject to a number of caveats discussed later in the paper. v the true effectiveness and sustainability of these interventions remains to be seen; benefits and harms of interventions should be evaluated post-implementation and alternative strategies (including combinations of approaches) continuously reconsidered. useful reminder that person z's is also an ethically problematic position. of course, an important concern for many people will not only be about their privacy today during the epidemic, but also about their future privacy: will full privacy protections be reinstated after the epidemic? will data gathered now be used in unacceptable ways later? this final point highlights, importantly, the fact that any justification of infringements of privacy will need to include a convincing account of their scope and duration. the discussion above suggests not only that some constraints on liberty and on privacy may be justified in the context of a global health emergency. it also implies that there may be a tension or trade-off between them. would it, for example, be ethically justified to retain/impose a blanket lockdown on society as a whole -including those not at risk themselves or a risk to others -on privacy grounds alone? much would depend on the details of the scope of the privacy infringement. however, its potential use in enabling many people now, and ultimately all, to emerge safely from a damaging lockdown provides a strong autonomybased prima facie argument in favour of the introduction and use of the app even if it were considered to constitute a privacy infringement. it is worth noting that there are at least two ways in which the use of the app has the potential to be autonomy enhancing. the first is its potential to enable people to go about their lives freely without the constraints imposed by a lockdown. the second is that it would provide a tool to enable individual people to make informed choices about how to behave in a socially responsible way e.g. to self-isolate as necessary to reduce the risks to others. the ideal situation would be for the downloading of the app to be voluntary and for the scale of voluntary uptake to be significant. this is a possibility given that it is believed that an uptake of below % would still -in combination with other measures -be sufficient to make an important impact. there are a number of reasons why those who have smartphones will have a strong incentive to sign up. the first of these is that this would ultimately mean that they and everyone else will emerge from the lockdown more quickly and safely. a second is that, by so doing, they will then be enabled to contribute to saving the lives of others, particularly the vulnerable, and those in caring roles, both locally and globally. appeals to a sense of 'we are all in this together' of 'solidarity' may be effective. vi a third, is related to the impact on the user's own level of risk. although primarily aimed at population level impacts, if a person downloads the app -and so do their close contacts -their personal risk will be very significantly reduced. this is because the de facto effects of app uptake will mostly act very locally except in busy urban environments such as the london underground. what if this does not work? if actual or predicted uptake is insufficient, is there an argument for the use of incentives? against this background of the scale of the current lockdown, the use of incentives to minimise the length of lockdown while also saving lives might be justified if uptake was insufficient and that there was evidence that greater uptake would release large numbers of people from an avoidable lockdown. the nature of these incentives would need careful consideration on vi it is important to note that there are circumstances in which, and many people for whom, appeals to solidarity might also be exclusionary or deepening of existing social divisions. a case-by-case basis. some possible examples might include: a donation to a nomiated charity, or free mobile phone credit. the use of incentives inevitably raises a number of equity questions with regard to those who do not have access to suitable smartphones and would not have access to these benefits through this route. these would need to be acknowledged and addressed in any defensible policy. vii thus far, we have been considering ethical questions relating to the use of the app by individuals. there are, however, implications for institutions and professions such as those who manage care homes or places where large numbers of people congregate such as cafes and restaurants. as we emerge from the epidemic into a world in which infection rates are lower but in which there is not as yet a vaccine-a world in which the transmission of covid- needs to be minimised-such people might reasonably be expected to ensure that the level of risk in their establishment or workplace is minimised. in this transition period, people in these positions might reasonably be seen to have an obligation to allow entry only to people who are able to show they are low risk. it might reasonably be judged irresponsible of such an institution to subject residents or customers to avoidable levels of personal risk and to fail to contribute to the suppression of infection transmission in the public interest. viii this perspective suggests additional reasons for thinking that the uptake of the app might be high because there is good reason to assume that most people would want to be able to both emerge from the lockdown and also to know that when they went to work or to a café they would be safe to do so, and contributing to the safety of others. this might provide a way for professionals and institutions to meet their obligations and an additional incentive to individuals to act responsibly. if the app can be shown to offer the potential to provide information to enable individuals and those who manage institutions to ensure an intelligent and safe emergence from lockdown, there are good reasons for its use. the 'if ' here is important, however, because the app's success will depend not only on the effectiveness of the app itself but also upon the existence of complementary infrastructure such as easy access to reliable testing, support to make sustained self-isolation possible and employment protections to ensure that those who do self-isolate are protected. this suggests the need for an in-depth ethical analysis of the process of emerging from lockdown, potentially into a series of periodic lockdowns with significant impact on the lives and well-being of many people. should the data be deleted at the end of the epidemic? one way of increasing the chances that people will be willing to download the app and allow it to gather data of proximity events might be for clear legally enforceable commitments to be provided that when the epidemic is over (according to some agreed criteria) the app and its data will be deleted. if this is essential to create the conditions for sufficient uptake and hence for saving lives and reducing suffering, it should be considered. it is not an ethically unproblematic course of action, however. one of the most striking and disturbing aspects of the current pandemic has been the way it has revealed how poorly prepared vii these and other equity questions are expanded on below. viii this paragraph needs to be understood in the context of those made in the later section on 'equity, fairness, and justice'. the world and individual countries are for such an eventuality both in terms of health system resilience, availability of equipment and tests and in terms of reliable epidemiological modelling. against this background, it is clear that we have responsibilities not only to those who are currently suffering from covid- but also to future generations. if the app is adopted as an intervention, the data it produces could be an invaluable resource for the protection of future generations from serious harm i.e. through research, the development of modelling methods and evaluation of the range of current responses. if these data re to be retained for such uses, a number of important questions about security, oversight, and ownership will need clear and enforceable answers. the successful and appropriate use of mobile phone apps to facilitate instantaneous contact tracing in the context of covid- in democratic countries depends on the establishment of sustained and well-founded public trust and confidence. this applies to the use of the app itself and of the data. the use of 'well-founded' here is intended to emphasise that mere presence of trust is insufficient in itself: such trust must be genuinely warranted. the requirements for well-founded trust will vary from country to country and perhaps even from person to person. however, in democratic contexts, in addition to the provision of clearly articulated and justified answers to the questions set out above, requirements are likely to include: the establishment of effective, transparent, accountable and inclusive oversight-perhaps by an ethics oversight body including members of the public; the agreement and publication at the outset of ethical principles by which the use of the intervention will be guided; the use of a transparent, auditable and easily explained algorithm; the highest possible standards of data security; and effective protections around the ownership uses of data. all public health emergencies and the actions taken to deal with them raise important justice questions because they are situations in which infringements of justice, discrimination and stigma commonly occur. it is also well established that the development and introduction of new technologies are capable of creating new forms of discrimination and further enhancing those that pre-existed the innovation. these can take the form of bias within the technology itself (perhaps because of biased data), biases arising out of the uses to which the technology is put and bias out of the fact that it may be available to some but not all. the response to covid- has been no different to previous public health emergencies in this regard. against this background, an important requirement for the credibility of any attempt to justify the use of the mobile phone app as part of a wider set of public health interventions to address the threat of covid- will be recognition of the importance of engaging seriously with equity and justice issues. notwithstanding the impossibility of addressing all structural issues in the compressed timescale of a pandemic, evidence is needed of a clear, actionable and ambitious plan for addressing these issues. once the current pandemic is over, there will inevitably be reviews of scientific, epidemiological and medical evidence about which interventions were or were not effective. if it turns out to be the case that the use of instantaneous contact tracing combined with widespread testing is effective, questions will arise about the ethical implications for its use in other infectious disease outbreaks. would it, for example, be acceptable or even required for a specifically designed app to be used each year in the context of seasonal influenza? these are important ethical questions. although there are differences, there are also morally significant similarities between covid- and seasonal influenza. for example, while its transmission rate is generally lower than covid- , the numbers of deaths internationally from seasonal influenza are very large indeed. one important difference, at present, between the two diseases is that mechanisms capable of developing a vaccine each year with some degree of effectiveness against seasonal influenza are in place. this may suggest that, unless judged less harmful or more effective than vaccination, the use of the app in seasonal influenza may not be justified. however, it is possible that apps will be appropriate in other settings and, where likely to be effective, constitute an important and ethically justified part of the public health toolkit. in this paper, we have set out a number of pressing ethical questions raised by the proposed use of a mobile phone app, the collection of proximity data for the control of the covid- pandemic, and the safe emergence of populations from government-imposed lockdowns. scientific and epidemiological evidence suggest that an app of this kind has the potential to contribute to reducing the suffering caused by the pandemic and minimise the harms caused by long periods of lockdown. these benefits and the avoidance of harms are clearly of great moral significance. if they are to be realised, however, several other ethical requirements need to be met. we have highlighted a number of such requirements which deserve attention in any ethically justified use of this technological intervention. in the uk, there is early empirical evidence that a high proportion of the population would choose to download the app under current circumstances, given adequate protections. in an on-line survey of predicted user-acceptance conducted by our collaborators, % of respondents said they would definitely or probably install a contact-tracing app. before they are invited to do so, they need to be assured that adequate protections and oversight are in place. a profoundly important ethical question presented by this technology concerns the problem of how and whether societies can find ways to benefit from the potential of algorithmic approaches to improve public and individual health,while also ensuring that the legacy of the deployment of these technologies does not impact negatively on future generations. correction notice this paper has been corrected since it was first published online. there are two instances in the title and the main text where 'contact' was incorrectly spelt as 'contract'. global coalition to accelerate covid- clinical research in resource-limited settings invisible women: exposing data bias in a world designed for men urban social media demographics: an exploration of twitter use in major american cities racism and discrimination in covid- responses global mortality associated with seasonal influenza epidemics: new burden estimates and predictors from the glamor project acknowledgements our thanks to yasmin gunaratnam, jonathan montgomery, and mariam motamedi-fraser for their helpful comments on earlier versions of this paper.funding this study was funded by the wellcome trust ( ). patient consent for publication not required.provenance and peer review not commissioned; internally peer reviewed. key: cord- - wkes nk authors: goggin, gerard title: covid- apps in singapore and australia: reimagining healthy nations with digital technology date: - - journal: nan doi: . / x sha: doc_id: cord_uid: wkes nk widely and intensively used digital technologies have been an important feature of international responses to the covid- pandemic. one especially interesting class of such technologies are dedicated contact and tracing apps collecting proximity data via the bluetooth technology. in this article, i consider the development, deployment and imagined uses of apps in two countries: singapore, a pioneer in the field, with its tracetogether app, and australia, a country that adapted singapore’s app, devising its own covidsafe, as key to its national public health strategy early in the crisis. what is especially interesting about these cases is the privacy concerns the apps raised, and how these are dealt with in each country, also the ways in which each nation reimagines its immediate social future and health approach via such an app. a striking feature of the covid- pandemic has been the use of and appeal to digital technologies -fusing together what these technologies might offer in terms of efficacious communication and public health responses to help individuals and communities cope and contain the pandemic, on the one hand, as well as extending resources for social practices, expression, making sense, persisting with and reconfiguring rituals, and conjuring with the profound affective dimensions wrought by illness, death, loss, fear and isolation, on the other. in the pandemic, digital technologies have been used across societies, in a way that harked back to earlier ideas from the s of social life being nigh wholly dependent on life in 'cyberspace', and 'virtual communities'. with widespread access to and ownership and use of internet, mobile phones, social media, data, artificial intelligence (ai) and associated technologies already deeply, if very unequally, distributed globally, especially in middle-and high-income countries, the inception of the pandemic saw extended reliance on digital technologies -where terms of digital inclusion allowed for it. in a number of countries, governments also took the opportunity to issue calls to the acceleration of digitalisation, especially across groups and demographics where digital inclusion and take-up had been low, due to infrastructure, literacy and education, information and affordability. one stand-out area in this regard was apps. apps have been around since the s and s; however, it was the 'smartphone moment' of the launch of the apple iphone in , and subsequent development of the apps for apple mobile operating system devices (ios) and launch of its apps store, that kicked off the process by which apps became an integral part of everyday life for billions of users (goggin, ; miller and matyivenko, ; morris and murray, ) . from until the present day, technology companies around the world have offered their own apps and apps store, first with the 'app store' wars of - featuring many of the handset vendors that were household names in the worlds of g and g mobiles such as nokia and blackberry. competition was much more suggestion in china, evidenced by the many chinese app stores that dominate its huge market, and are significant distribution points for many users and communities internationally -especially given digital technologies being at the centre china's external trade, finance and soft power 'going out' (keane and wu, ) . thus, apps are key to what has been recently called 'infrastructural imaginaries' (nielson and pedersen, ; see also anand et al., ; athique and baulch, ; mansell, ) . so it is no surprise that apps formed a key part of the infrastructures woven into the pandemic, but also a specific, highly visible and 'normalized' response (hoffman, ) , in the form of dedicated apps -especially for tracing people and their 'contacts'. apps were used for many significant purposes during the pandemic. existing popular apps such as whatsapp were used in some countries to send official government messages and distribute crucial public health information. the data sets generated by smartphones, computers, apps and people's use of them, such as that data collected by apple and google, were used by public health officials, researchers and journalists to map population or district-level activity and movement, leading to the very interesting charts, graphs and visualisations in news and current affairs reports and features seeking to map and analyse the spread of covid and its impact on social and economic activity. apps allied with machine learning and ai were also used by medical researchers and clinicians to assist in the diagnosis of covid, by asking millions of users to track and enter their symptoms, diary-like, to offer a way of pinpointing when someone might have become positive. among the many varieties of covid-dedicated apps were apps devoted to the purpose of tracking people and their potential contacts, in case they contracted the virus. so many countries developed apps for tracking and contact tracing, with so many prototypes in development and implemented, that mit launched a contact tracing app database (https://www.scl.org/ news/ -the-mit-contact-tracing-app-database), based on key questions from american civil liberties union (aclu) white paper (aclu, ), to provide an authoritative reference point for those seeking to find their way through the claims and counter-claims of effectiveness. apple and google joined forces to amend their policies and create a joint protocol to make it easier for countries to use such data for contact tracing via apps (michael and abbas, ) . a full treatment of covid contact tracing apps is outside the scope of this article (see, for instance, cattuto et al., ; hoffman, ; vinuesa et al., ) . instead i focus on two especially interesting cases that offer us early insights into the socio-technical dynamics at play in such apps and the pandemic itself. these are singapore's tracetogether app and australia's covidsafe app. asian countries were often referred to for their decisive and often authoritative responses to the pandemic. however, it was singapore that attracted considerable early notice for its pioneering role in developing a particular kind of covid contact tracing app -that captured the imagination of many other countries. singapore was a pioneer in the development of covid bluetooth app in the form of its tracetogether app. what was less publicised was that, shortly after launch of tracetogether, singapore changed tack. this modification of the app deployment and promotion, and place in the overall public health strategy, was less evident outside the city-state. instead, singapore's tracetogether app became a stand-out model for other countries, rather than the various other apps being implemented around the world such as those developed by the united states, south korea, china, india or israel (babones, ) . australia comes into the picture because it is australia who first and most systematically sought to build on the tracetogether model, including its privacy safeguards, with its own covidsafe app. in the capstone analysis of their series of timely interventions into the privacy debates on the introduction of australia's covidsafe, leading privacy scholars graham greenleaf and katherine kemp ( ) note, 'australia's experiment is further advanced than most [countries] that are attempting to build a system based on voluntary uptake, protected by legislation (abstract, para ). the australian government sought to deploy covidsafe as a centrepiece of its effort to re-open australian society after the national and state lockdowns occasioned by the 'first wave' of infections from march to may . where public concern regarding and discussion of privacy issues was clearly presented but publicly muted in singapore, in australia there was furious debate. to explore the emergence, dynamics and implications of these two covid apps, i will proceed as follows. first, i introduce and discuss singapore's tracetogether, its development and first phase of take-up and deployment. second, i turn to australia's covidsafe and consider its fast journey from incubation and policy idea to the touchstone to warrant the country's re-opening, a veritable 'national service' (as prime minister morrison couched it). third, i return to singapore, to discuss the rebooting of tracetogether, after nearly months of tepid take-up, as that country's leadership sought to reassure its population that conditions were safe to re-open social life. finally, i offer concluding remarks about covid apps, social and technological imaginaries and digital media, as the nation state returns (flew et al., ) , and seek to gauge and exert its brittle powers, in a still deeply interconnected world. to great fanfare, a dedicated contact tracing app was unfurled as a breakthrough in monitoring outbreaks of covid- at the population level. while many teams around the world produced similar versions, the singapore government rolled out the first such app -called 'tracetogether'. tracetogether is an open source app based on bluetooth, using the 'bluetrace' protocol devised by a singapore government team led by the govtech agency -who have a track record of developing new kinds of open government apps, such as the parking.sg app. in a interview, for instance, janil puthucheary ( ), minister-in-charge of govtech, discussed how the 'govtech guys, as a result of having to do the code for the service . . . are having to . . . hack policy'. puthucheary explained, 'you have to be able to codify the policy', however that 'some of our governmental processes and regulations result in extremely inelegant code' (puthucheary, : ′ ″, ′ ″) . tracetogether was made available for adoption elsewhere via github. it is a combination of centralised contact tracing and follow-up (undertaken by government health authorities) and 'decentralised contact logging'. the user downloads the app and activates bluetooth on her device. the app can then detect another device in its vicinity, exchanging proximity information. to do so, the app uses information generated by the bluetooth relative signal strength indicator (rssi) readings that occur between devices over time to estimate proximity and duration of an encounter between users (team tracetogether, c). if a person fell ill with covid- , they could grant the ministry of health access to gather their tracetogether bluetooth proximity data -to assist in contacting people who had close contact with the infected app user. for their part, the developers emphasised their view that tracetogether would 'complement contact tracing, and is not a substitute for professional judgement and human involvement in contact tracing' (team tracetogether, d) . interestingly, they also underscored that the 'hybrid model' of decentralised and centralised approach is what they feel 'works for singapore' and that they 'built it specifically for singapore' (team tracetogether, b) . released on march by the ministry of health and govtech (baharudin and wong, ) , tracetogether received over half a million downloads in its first hours. a month later, the singapore government claimed the app had achieved a % adoption rate -some . million users, of an overall estimated population of . million users (team tracetogether, a). upon launch in singapore, there was relatively little public discussion of the privacy implications of tracetogether in mainstream media and fora -although there was considerable disquiet, criticism and debate evident in blogs, social media and elsewhere. for the most part, this is due to the structure and dynamics of singaporean society, and its political arrangements, public policy traditions and strong systems of social control and clear support for or alternatively discouragement and sanctioning of different kinds of expression and voicessomething well established in the scholarly literature (chua, ; george, george, , lee, lee, , , especially via various studies published in media international australia (most recently, lee and lee, ) . in recent years, the singapore government, following the dampened level of votes received by governing people's action party (pap), that has ruled since the , in the election, and a more sceptical populace (barr, ; zhang, ) , it has sought to extend consultation and formal 'listening' mechanisms to provide additional opportunities for citizens' voices. furthermore, while there has been increased discussion of privacy with the rise of digital technologies and unprecedented expansion of data generation, collection and use, the legal and regulatory framework is relatively weak in relation to privacy rights taken-for-granted in many jurisdictions (chesterman, (chesterman, , , even in the wake of the european general data protection directive (gdpr). however, as we shall see, such debate did build over some months, as tracetogether evolved. what is also important to note is that the singaporean government clearly acknowledged the strength of attitudes and importance of privacy and data protection concerns, and sought to anticipate debates by building in some level of privacy protection. the vision of tracetogether is that proximity data gathering is 'done in a peer-to-peer, decentralised fashion, to preserve privacy', and that it relies upon a 'trusted public health authority, committed to driving adoption' (team tracetogether, c). the developers and government emphasised that the privacy safeguards in the tracetogether app are in effect an effort of the longstanding ideal of 'privacy-by-design' (hustinx, ) . the government emphasised that the information was stored on a user's phone for days, and then deleted -and user's phone numbers are not exchanged, no geolocation data, personal identification data are not exchanged, so, as minister puthucheary noted, 'the engineering has preserved the privacy of the users from each other'-calling the app 'fairly elegant', in the way it 'preserves a fair degree of privacy' (ng, ; see also govtech, ). almost immediately the app did receive notice and discussion internationally, as one of a growing number of examples of covid- contact tracings apps raising privacy concerns (hu, ) . meanwhile in singapore, tracetogether downloads flatlined. this occasioned international deliberation, such as an article in the wall street journal entitled 'singapore built a coronavirus app, but it hasn't worked so far' (lin and chong, ) . the stalled downloads of tracetogether brought into view the conversations about privacy concerns, and whether this was a factor in user's lack of motivation to download the app. another reason advanced was that the app posed challenges for battery draining, due to the need to keep phones on. this was a view put by the ceo of singapore's investment or sovereign wealth company temasek holdings, madame ho ching who has a reputation as a prolific commentator on public affairs by dint of her regular controversial facebook posts (ho is also the wife of prime minister lee hsien loong) in a facebook post of may on the problems with tracetogether (ho, a) . whether by design or dawning acceptance, the government eased back its public communication and encouragement for citizens and other residents alike to download and use tracetogether. instead, it encouraged businesses, organisations, government offices and other entities to use a range of techniques to gather information about people's movements -especially when they visited or spend significant time in public places. check-in was principally done via scanning of a national identity card or employment or work permit id card, or via an app called safeentry. based on scanning of qr codes specific to each location, the safeentry app, and the policy it supported, was comprehensively promoted by government. this contrasted with tracetogether, which was only lightly promoted by the singaporean government, with the major campaign at the outset of its launch. presumably, on a small island -city-state, with strong civil service corps, existing id systems (singpass), and tightly managed immigrant and foreign worker id and records, and digital government and technology capabilities, this evolving contact tracing system did not need to premised on an app such as tracetogether, which presumably government was happy to allow to 'fail fast', given the bugs it faced. despite the effectively prototypical status of tracetogether, one of the first jurisdictions to adopt the technology was australia. prime minister scott morrison referred to such an app as a key requirement in australia's ability make its transition out of lockdown (prime minister et al., ) . in a radio interview with national talk show host alan jones, morrison's language is instructive, because it imagines technology, especially automated technology, as taking the vagaries and morally dubious qualities of human agency out of the picture: '[w]e need to get an automatic industrial level tracing of the coronavirus . . . now, we've been working on this automatic process through an app that can ensure that we can know where the contacts were over that infection period and we can move very quickly to lock that down' (morrison, a: para ) . this kicked off a heated debate about privacy implications, leading a high-profile member of his own coalition government, rural parliamentarian barnaby joyce to declare that he would not be downloading the app -countered by many other public figures who promised to do so. despite the widespread criticism and concern, there was also significant support with some million downloads in the first day of its release, topping the million mark in early may (koslowski, ) , then . million by june (meixner, ) . these figures raise various concerns, such as whether those who download the app used or continued to use it. also what the rate of downloads were in different parts of the nation (slonim, ) . let alone whether the covidsafe app was playing a role in helping to trace contacts and find positive cases of the virus (preiss and dexter, ) . for the prime minister, the covidsafe app was a rhetorical centrepiece of his policy initiative to vouchsafe a loosening of restrictions and begin to repair the economic damage the virus caused: the chief medical officer's advice is we need the covidsafe app as part of the plan to save lives and save livelihoods. the more people who download this important public health app, the safer they and their family will be, the safer their community will be and the sooner we can safely lift restrictions and get back to business and do the things we love. (prime minister et al., : para ) while he drew attention to the voluntary, consent-based nature of the app, morrison also sought to exert maximal symbolic pressure by framing adoption in patriotic terms, likening it to national service in wartime (and also not ruling out making it mandatory) (gredley, ) : i'll be calling on australians to do it as a matter of national service. in the same way people used to buy war bonds, back in the war times, you know, to come together to support the effort . . . if you download this app you'll be helping save someone's life. (morrison, b: sec : ) on may , morrison announced that australia had earned an 'early mark', with restrictions being lifted in a week. in doing so, he spoke of the download numbers being a 'critical element' in deciding to what extent the easing would occur: 'mr morrison said not installing the app was like going into the "blazing sun" without wearing sunscreen' (armstrong and minear, ) . various commentators and researchers expressed their views on how to promote downloading and take-up of the app. in the australian financial review, a piece by technology editor paul smith, entitled 'think like a founder', reported, 'entrepreneurs and health technology experts have urged the government to adopt all the tricks of the start-up trade to get more australians downloading the covidsafe contact tracing app' (smith, ) . the australian chief scientist through his rapid research information forum commissioned a brief on motivators for use of the covidsafe app, supported by the australian academy of humanities, with professor genevieve bell as lead author, and various leading media, communications and humanities researchers among the contributing authors (bell et al., (disclosure: i was a peer reviewer of this brief)). the brief suggested that 'illustrating that covidsafe works as intended may assist decision-making for those yet to download the app' (bell et al., : ) . it also concluded that the stories we will tell about australian responses to, and uses of, covidsafe will matter too. the voices of trusted figures, community leaders, healthcare workers and citizens will likewise inform the adoption, and continued use of, covidsafe. (bell et al., : ) many of these 'stories' clustered about the public perceptions and debate about the privacy, data and surveillances implications of the covidsafe app (bell et al., ) , driven by long-standing sensitivities and attitudes of australians concerning privacy. stretching back to the infamous and ill-fated australia card proposal of , citizens' privacy concerns had been more recently exacerbated by the federal government's poor handling of the ramp-up of its national e-health records registration system, myhealth, which switched from an 'opt-in' to 'opt-out' basis in (komesaroff and kerridge, ; goggin et al., ) . with much at stake in terms of public health concerns at a critical juncture of the covid pandemic, the australian government emphasised that it was keen to adopt a 'consent-based' model, hence its interest in adapting the singapore tracetogether app. the government sought a formal privacy impact statement from a leading law firm -which it published, with a detailed response from the department of health ( b; maddocks, ) . this privacy impact documentation put important details of the workings of the covidsafe application, and the production, storage and handling, of resulting user data on the public record. in response, the government emphasised that participation would be voluntary (department of health, a); however the privacy impact statement noted the potential for third-parties such as workplaces or businesses put pressure on or require people to use the app (maddocks, ) . deleting the app would also delete the data stored on a user's device, but not data in the national data store (however, the government guaranteed that all data held would be deleted at the end of the pandemic). the government was at pains to reassure the public on the secure hosting of the covidsafe data store, undertaken by amazon web services (aws). their guarantees related to the data privacy and security obligations applying to aws, but also to any prospect that such data might be requested and commandeered by the us government (given aws is headquartered in the united states, and subject to their laws). over some weeks a furious debate ensued, and the australian government proposed legislation to address the key concerns. this safeguard took the form of the privacy amendment (public health contact information) act . the bill quickly passed through the house of representatives and the senate and received assent on may . the act creates several serious offences dealing with covid app data, including 'non-permitted, use, or disclosure', 'uploading covid app data without consent', 'retaining or disclosing uploaded data outside australia', 'decrypting encrypted covid app data' and 'requiring participation in relation to covidsafe' (privacy amendment, : ). 'covid app data' is defined as 'data relating to a person . . . collected or generated . . . through the operation of covidsafe' and is either 'registration data' or 'is stored, or has been stored . . . on a communication device' (s. d ( ) (a-b), privacy amendment, : - ). while the bill was passed containing significant safeguards, it contained serious flaws. as summarised by greenleaf and kemp, these included key information upon which the law was based and would operate was not made available to the public, including advices to the minister upon which he relied to make the earlier determination, and, crucially, the agreements between the commonwealth and states and territories regarding the operation of the covidsafe app, and collection and sharing of app data; lack of public assessment of the law by the federal and states and territories privacy commissioners; and only the source code for the covidsafe app was released, not the code for the national covidsafe data store (i.e. the server-side of the system, where security and privacy issues often manifest) (greenleaf and kemp, ) . in addition, greenleaf and kemp critique the narrow focus of the privacy act amendment on 'covid app data', suggesting instead that what is being created is an information system they dub the 'covidsafe system' (greenleaf and kemp, ) . as well as the specific defects of the new law, then, the major issue it raises is precisely the one feared by many experts and members of the public alike: that the app-based contact tracing represented by covidsafe, and other apps around the world, represent a deepening of technologies of surveillance in social life. while such apps and measures in which they are embedded are justified as exigent public health measures crucial in the emergency conditions of a pandemic, there is wellfounded fears that this increase in surveillance will not be automatically or easily rolled back once countries feel the threat of covid is ended or at least contained. as australian debates over covidsafe privacy subsided, there was a slow return to tracetogether emerging in singapore as the country's leadership gingerly considered how to effect its re-opening from its -month circuit breaker. a task all the more urgent, given the ruling party's dwindling time to call a national election. singapore's was regarded an international model of wise and swift response with its handling of its 'first wave' of infections. however, in the second week of march , singapore tightened its measures, enacting a general shutdown and stay-at-home policy that it dubbed a 'circuit breaker'. initially the circuit breaker was announced to last for month, but with the rising tide of positive cases in the crowded migrant worker dormitories, the government quickly extended for a second month. a disturbing feature of singapore's data gathering and public reporting and communication during this period was the distinction clearly drawn and maintained in the daily bulletins between; cases in the migrant worker dormitories; and 'community cases' (these community cases were in turn divided between figures on singapore citizens, permanent residents (prs), migrant workers on work permits and workers on employment passes) (han, ; palma, ) . the migrant workers were quarantined in the dormitories, with many then moved to across other repurposed facilities. and the numbers of cases were similarly quarantined, in a communicative-epistemological manner, to emphasise that the 'real' community spread remained low (usually below cases in the 'community'). those numbered among the community included citizens and prs initially, but subsequently, foreign pass holders who do not reside in dormitories but lived among the regular population, became part of these statistics once the dormitory cases started subsiding. ahead of the planned end to the circuit breaker on june, the government made some mention of tracetogether at various times in its public communications. however, its main focus remained racking movement and individuals' location via check-in at the public places and business still open, such as convenience stores and shopping centres, or in taxis and ride-hailing services, especially via the safeentry app discussed above. as the re-opening loomed, there was increasing discussion in government, and in parliament, on measures that would need to be implemented to contain and reduce the number of infections via contacts with migrant workers, especially once they were allowed to more regularly leave the dormitories, where they had been quarantined during the circuit-breaker period, and so circulate in the 'community'. the government announced a new app, sgworkpass, to 'show which migrant workers can leave their dormitories for work' . workers will 'get a "green status" on the app to indicate that their employer has been granted approval to resume operations, and that the dorm they stay in has been cleared' . otherwise, the app will show red to indicate they 'cannot go out for work' . this is reminiscent of the chinese app, also adopted by india, which uses qrs, to show a user's status as green (when they may enter offices, restaurants, malls or parks), or yellow (at risk) or red (strict quarantine) (hu, ; india today, ) . at this stage, tracetogether returned -this time, as a central feature of the strategy. the government had been at pains to keep tracetogether opt-in, with foreign minister vivian balakrishnan, also minister-in-charge of the smart nation initiative, providing reassurance that the app would remain voluntary 'as long as possible' (balakrishnan, ) . in early june, balakrishnan noted the problems with tracetogether, including the technical issues with the app not running properly on apple. as a result, he let it be known that singaporean government was developing a 'portable wearable device' that will achieve the same end, that if it worked could be 'distributed to everyone in singapore': 'i believe this will be more inclusive, and it will ensure that all of us will be protected' (balakrishnan, ) . the government emphasised that there would no 'gps chip' on the device, nor any internet connectivity. even then, the tracetogether token would need to be physically handed to the health ministry for uploading of the data, if a user tested positive for covid- (yu, ) . the government's keenness to be seen to address privacy was doubtless fuelled by a public backlash against the token. as policy researcher and commentator carol soon, from the institute of policy studies, noted, 'within a short span of three days, a petition against the development of the device attracted about , signatories' (soon, ) . concerns of singaporeans regarding data privacy were addressed in a report authored by her colleagues, which found attitudes vary according to the technology involved, illustrated by the finding that nearly in respondents supported use of cctv to monitor people's movements during the covid 'circuit breaker period', but less than % were comfortable with having their mobile phone data tracked for contact tracing without their consent (tay, ) . to address such deep-seated concerns, soon suggested the need for singapore to urgently 'achieve a working compromise between personal data and public good', establishing principles and considering measures such as formation of a citizen's panel for public deliberation (soon, ) . regardless, the first batch of , tracetogether tokens were distributed to seniors shortly on the eve of the july general election -with officials from the smart nation and digital government group (sndgg) suggesting they were settling in for a long haul, saying the government will 'continue to generate more awareness about the token among our prioritised population' (sndgg officials quoted in yip, ). at the time of writing, the pandemic rages globally, and the career of covid- contact tracing apps is still unfolding -with little evidence as yet of their efficacy. however, there are already clear grounds for concerns. the strange thing about the australian embrace of bluetooth-based covid tracing apps is how strongly it figured, for a time at least, as instrumental to the country's public health response. various commentators noted the irony that at the point covidsafe was being pushed upon the public, australia was at a positive inflection point in terms of infections. as greenleaf and kemp note, this set the bar because other measures had already appeared to be successful in greatly abridging the spread of the virus (greenleaf and kemp, : ) . the other obvious thing is that where apps did play a role in diminishing infection rates, these were not: ( ) bluetooth-based tracing apps, ( ) and the apps used were integrated into a wider system of cross-referencing and marshalling personal identification and contact information and database systems (greenleaf and kemp, ). yet the australian government, for a short time at least, was very keen on the app as a symbolic game-changer in its public health approach to the pandemic -showing that it was taking charge. rather like british health secretary, matt hancock some weeks later, when he promoted the english app-based test-and-trace system, telling the public 'it is your civic duty': do it for the people you love. do it for the community. do it for the nhs and do it for all the frontline workers . . . you'll have the knowledge that when the call came you did your bit, at a time when it really mattered. (hancock quoted in bosley and stewart, ) in july , there was an outbreak of covid- cases that saw a lockdown re-imposed, and fuelled national concerns. at the time victorian chief medical officer brett sutton said the 'app has not added a close contact' that authorities had not already discovered via traditional contact tracing (borys, ) . federal health minister greg hunt advised that at least contacts nationally had been identified via the covidsafe app (borys, ) . for her part, nsw's chief health officer kerry chant, the state next in line for a potential resurgence of cases described the app as 'one of the tools', but not a 'major feature' in contact tracing (borys, ) . in his parsing of the app's effectiveness, australian deputy chief medical officer, dr nick coatsworth, suggested that because of movement restriction, people had not been circulating, so the 'app hasn't identified those cases', and that as 'numbers go up then the app can come into its own' (coatsworth, ) . with the groundswell for mask use in mind, coatsworth ( ) cleverly sought to link the two, suggesting 'if you are a supporter of mask use, you must also be based on the modelling, a supporter of downloading and activating the app' (here he refers to the study by sax institute, see currie et al., ) . for its part, singapore took a less dramatic, more considered approach, especially in the first phase as it developed and launched its tracetogether app. singaporean leaders and health officials were also preoccupied with promoting the app to gain the maximum take-up and adherence. yet, for reasons not entirely clear as yet, singapore was reluctant to push the adoption of the app to the extent that australia did -an interesting situation given earlier critiques of singapore technocratic approach to health care, in particular (barr, ) . as well as the privacy concerns that emerged in the second phase of the tracetogether token initiative, it may be that singaporean actors thought the app was promising but not the main game. this would be because of the already well entrenched systems of requiring and using personal data, through an extensive infrastructure of technologies (including the cctvs that featured in the ips report), without the kind of concomitant privacy rights and practices that would be expected in some other jurisdictions such as australia. the task of enlisting and normalising singaporeans participation in these aspects of its surveillance-extensive 'smart nation' policies, over cumulative implementation of technology is taken to be essential, but it is increasingly fraught (lee, ) . in the first months of pandemic response, then, the central element was singapore's established singpass and other systems of identification cards and passes, which could be used in coordination with video recordings, and the wealth of digital data available from urban transportation systems, stored valued and transit cards, ride-hailing accounts and so on. as well as also as the citizen and netizen sousveillance and activism that saw recordings of potential miscreants breaching the regulations circulated online. in addition in the early weeks of the pandemic, identifying details of people's residential locations, down to building numbers, were published in daily updates from ministry of health, and reprinted in media outlets. such measures point to the differences in privacy laws and protections in singapore, as compared to australia. whereas australian privacy act dates back to , singapore only enacted its first comprehensive law in , the personal data protection act. at the time, legal scholar simon chesterman suggested that singaporean had taken a 'pragmatic approach', potentially striking a balance between european and us approaches: in singapore, at least, reform is not being driven by the desire to defend the rights of data subjects; rather, it is based primarily on economic considerations, as well as the desire to position singapore as a leader in the region for data storage and processing. (chesterman, : ) . the singapore laws and approach to privacy and data protection have not substantially changed since (chesterman, ; ong, ). yet clearly citizens do have concerns -as the public response to the tracetogether token suggest. from a broader perspective, the return of tracetogether to the fore of the singaporean government's strategy, especially to assist with the re-opening process after its 'circuit breaker', is very interesting indeed in the context of the country's digitally underpinned governmentality (ho, ; lee, lee, , willems and graham, ) . this is worth being in mind in interpreting the election, in which the government received some strong criticism by opposition candidates for its poor handling of the pandemic, especially concerning the continuing high number of cases in migrant worker dormitories. the pap was returned to government, still with a 'super majority', of seats out of the available . however, it was chastened by its share of the vote being reduced to . % (from its . % share in the election) -and an unprecedented seats won by the opposition workers' party (loh, ) . in the aftermath, the government has signalled its willingness on listening to electorate concerns, especially those of young voters (yong, ) . all in all, in both these case studies, we see that the variations of the covid- contact tracing apps, and the technical, social, policy and design dynamics of these, offer rich food for thought when it comes to understanding apps. health information is an area of considerable sensitivity for most people. trust is key, and with the widespread diffusion of mobile communication there has been considerable work on how to design and implement systems that can support cooperative and sustainable sharing of information between people and authorities to map the spread of infectious diseases (lwin et al., ) . however, it is now evident that the task of assembling appropriate social and cultural understandings of people's lives and identities, their data selves (lupton, ) , the intricacies of technologies, the enmeshing of privacy expectations in design, and the construction of suitable legal, policy and governance arrangements, is challenging. in the covid- pandemic, many countries across the world have had recourse to apps, as flexible agents with capacity to encode, materialise, represent and integrate such requirements, including some contradictory ones, and imagine and forge majoritarian supported social action. it is difficult not to see the turn to tracing apps as a pivotal moment in the expansion and entrenchment of surveillance technology in digital societies, of which singapore in particular has been a leading example (lee, ) -but is also playing out in contests and debates in many countries especially in europe and asia. how this ultimately turns out, and with what benefits for health, as well as legacies for democratic freedoms and daily life, we must wait and see. american civil liberties union (aclu) ( ) aclu white paper -principles for technology-assisted contact trading the promise of infrastructure coronavirus: australia set to start easing covid- restrictions, donald trump's plan to punish china countries rolling out coronavirus tracking apps show why they can't work coronavirus: singapore develops smartphone app for efficient contact tracing. straits times skype interview with annelise nielson, sky news australia. transcript, may. available at singapore: the limits of a technocratic approach to health care ordinary singapore: the decline of singapore exceptionalism what motivates people to download and continue to use the covidsafe app? rapid research brief government says coronavirus app has identified contacts, as victorian authorities say it has not helped them hancock: it is public's 'civic duty' to follow test-and-trace instructions in england. guardian, may the institutionalisation of digital public health: lessons learned from the covid- app data protection law in singapore: privacy and sovereignty in an interconnected world after privacy: the rise of facebook, the fall of wikileaks, and singapore's personal data protection act liberalism disavowed: communitarianism and state capitalism in singapore deputy chief medical officer press conference about covid- . australian government department of health stemming the flow: how much can the australian smartphone app help to control covid- . public health research & practice ( ): e . department of health ( a) covidsafe app the covidsafe application privacy impact assessment: agency response global media and national policies: the return of the state. houndsmill: palgrave macmillan the air-conditioned nation: essays on the politics of comfort and control incomplete: reflections on a first world nation's arrested political development ubiquitous apps: politics of openness in global mobile cultures data and digital rights: recent australian developments help fight the spread of covid- with virus mobile tracking app may be mandatory. canberra times australia's covidsafe experiment, phase iii: legislation for trust in contact tracing singapore locks away migrants in pandemic fight one of the key tenets of agile development is to put aside face. facebook, may smart subjects for a smart nation? governing (smart)mentalities in singapore which workers can leave dorm for work? corona tracing apps-an infrastructural perspective. paper presented to technology policies and data governance in times of crisis, online session of communication policy & technology section beijing rolls-out color coded qr system for coronavirus tracking despite concerns over privacy, inaccurate ratings privacy by design: delivering the promises india follows china's lead to widen use of coronavirus tracing app lofty ambitions, new territories and turf battles: china's platforms 'go out the my health record debate: ethical and cultural issues covidsafe downloads reach million as experts question technical flaws. sydney morning herald, may from contempt of court to fake news: public legitimisation and governance in mediated singapore the media, cultural control and government in singapore. london media governmentality in singapore tracing surveillance and auto-regulation in singapore: 'smart' responses to covid- singapore built a coronavirus app, but it hasn't worked so far singapore election results give pap supermajority as rivals rise. nikkei asian review data selves: more-than-human perspectives a st century approach to tackling dengue: crowdsourced surveillance, predictive mapping and tailored communication the covidsafe application: privacy impact assessment imagining the internet: communication, innovation, and governance how many people have downloaded the covidsafe app and how central has it been to australia's coronavirus response? abc news getting behind covid- contact trace apps: the google-apple partnership the imaginary app appified: culture in the age of apps interview with alan jones interview with brian carlton coronavirus pandemic changes how your privacy is protected infrastructural imaginaries: collapsed futures in mozambique and mongolia singapore report: data protection in the internet surge in covid cases shows up singapore's blind spots over migrant workers all the right things': mcdonald's praised as covidsafe app yet to find a case in victoria. the age, may minister for health, minister for government services and chief medical officer ( ) covidsafe: new app to slow the spread of coronavirus. media release, april privacy amendment conversation with dr janil puthucheary. moderator: peter ho. stack conference, suntec singapore conference and exhibition centre, october here's why the government should share what it knows. the conversation, may think like a founder: how to get people to download the covid app getting buy-in for tracetogether device and future smart nation initiatives singaporeans accept some privacy loss in covid- battle but surveillance method methods: ips study. strait times, may team tracetogether ( b) can you make tracetogether available to other countries? media release how does tracetogether measure distance and duration of contact? what thresholds constitute close contact? media release team tracetogether ( d) what is blue trace? media release available at a socio-technical framework for digital contact tracing the imagination of singapore's smart nation as digital infrastructure: rendering (digital work) invisible. east asian science yip wy ( ) coronavirus: , seniors get first batch of tracetogether tokens. straits times ge : signs of young voters' crucial role in election outcome singapore looks to ease privacy fears with 'no internet' wearable device. zd net social media and elections in singapore: comparing my thanks to two reviewers for helpful suggestions. thanks also to rosemary curtis for her proofing of this article. the author(s) declared no potential conflicts of interest with respect to the research, authorship and/or publication of this article. the author(s) received no financial support for the research, authorship and/or publication of this article. https://orcid.org/ - - - key: cord- - u ptqjs authors: wells, philippa m.; doores, katie j.; couvreur, simon; nunez, rocio martinez; seow, jeffrey; graham, carl; acors, sam; kouphou, neophytos; neil, stuart j.d.; tedder, richard s.; matos, pedro m.; poulton, kate; lista, maria jose; dickenson, ruth e.; sertkaya, helin; maguire, thomas j.a.; scourfield, edward j.; bowyer, ruth c.e.; hart, deborah; o'bryne, aoife; steel, kathyrn j.a.; hemmings, oliver; rosadas, carolina; mcclure, myra o.; capedevilla-pujol, joan; wolf, jonathan; ourselin, sebastien; brown, matthew a.; malim, michael h.; spector, tim; steves, claire j. title: estimates of the rate of infection and asymptomatic covid- disease in a population sample from se england date: - - journal: j infect doi: . /j.jinf. . . sha: doc_id: cord_uid: u ptqjs background: understanding of the true asymptomatic rate of infection of sars-cov- is currently limited, as is understanding of the population-based seroprevalence after the first wave of covid- within the uk. the majority of data thus far come from hospitalised patients, with little focus on general population cases, or their symptoms. methods: we undertook enzyme linked immunosorbent assay characterisation of igm and igg responses against sars-cov- spike glycoprotein and nucleocapsid protein of unselected general-population participants of the twinsuk cohort from south-east england, aged - (median age ; % female). participants completed prospective logging of covid- related symptoms via the covid symptom study app, allowing consideration of serology alongside individual symptoms, and a predictive algorithm for estimated covid- previously modelled on pcr positive individuals from a dataset of over million. findings: we demonstrated a seroprevalence of % ( participants of ). of seropositive individuals with full symptom data, nine ( %) were fully asymptomatic, and ( %) were asymptomatic for core covid- symptoms: fever, cough or anosmia. specificity of anosmia for seropositivity was %, compared to % for fever cough and anosmia combined. individuals in the cohort were predicted to be covid- positive using the app algorithm, and of those, ( %) were seropositive. interpretation: seroprevalence amongst adults from london and south-east england was %, and % of seropositive individuals with prospective symptom logging were fully asymptomatic throughout the study. anosmia demonstrated the highest symptom specificity for sars-cov- antibody response. funding: nihr brc, cdrf, zoe global ltd, rst-ukri/mrc the covid- pandemic has constituted an international emergency, accounting world-wide for greater than , deaths thus far. in order to understand the transmission of the virus through the population, and to estimate protection afforded to those post-infection, we must first understand the seroprevalence to sars-cov- with regards to demographics and clinical presentation. data on the rates of infection in the united kingdom (uk) come mainly from the office for national statistics (government ons) surveys. as of th june , the ons estimate . % ( % ci to . ) of the uk population to be seropositive for igg or igm s glycoprotein detected using enzyme linked immunosorbant assay (elisa) testing, based on blood tests of individuals since april . , in the th july report from the ongoing ons household survey involving pcr swab testing of around , people, the authors reported an asymptomatic infection rate of %. however, these surveys did not study the full range of symptoms associated with covid- , and only assessed symptomatology at the time of swabbing, and therefore may have overestimated this rate. we undertook a population-based study of the humoral immune response to sars-cov- , with regards to longitudinal clinical symptoms collected through a mobile phone app in a population-based sample of twinsuk volunteers. participants were members of the twinsuk cohort, the largest uk registry of adult twins. participants were visited in their home to obtain saliva and serum samples to test for active infection and antibody response. the majority of participants had completed regular logging of symptoms, via the c- covid symptom study app since, enabling measurement of antibody response to covid- with regards to clinical symptoms. participants were members of the twinsuk cohort, the largest uk registry of adult twins. participants were visited in their home to obtain saliva and serum samples to test for active infection and antibody response. for three months prior to the visit, the majority of participants had completed regular logging of symptoms, via the c- covid symptom study app , enabling measurement of antibody response to covid- with regards to clinical symptoms. the inclusion criteria for the study were residence within an -mile radius of the cohort headquarters st thomas' hospital in central london, active use of the covid symptom study app, and availability for visit between th april and nd june . the exclusion criterion (for safety reasons) was report of recent symptoms indicating potential covid- at the time of the study or within days prior, which participants were required to confirm via telephone consultation. ethics approval for the study was granted by nhs north west -liverpool east research ethics committee (rec reference /nw/ ), iras id , and all participants gave written informed consent. twins met the geographical and logging based inclusion criteria. of these, could not be contacted, declined a visit as they were either not interested or did not want a home visit due to being in the high risk category for covid- , or because they or a member of their household were currently experiencing symptoms suggestive of covid- infection and were not able or willing to be seen later (n = ). a further could not be visited within the study period. visited participants comprised twinsuk volunteers, aged between and , of whom ( %) were female. they were visited in their home for antibody testing between th april and nd june , an average of days from the first peak of the pandemic in the uk on the rd april . the demographics of the study participants in the context of the wider geographical area in which they reside are summarized in table serum samples obtained on home visits were tested for igg and igm binding to sars-cov- s and n proteins using enzyme linked immunosorbent assay (elisa) using serum diluted at : . the elisa serology methods used in this study have been previously described. briefly, the n/s elisa demonstrated % specificity (as determined using pre-covid- serum samples), and sensitivity was secondary to time post infection, improving in performance with increasing days from initial infection: after days it was . %. after days it was . % and at greater than days it was . % sensitive. a participant was considered seropositive if an igg response (optical density (od) value) to both n and s was detected that was -fold above the background of the assay. this cut-off is based on the analysis of + pre-covid- serum samples.  the elisa was validated by a separate laboratory at imperial college london, who employed a hybrid double antigen bridging assay (daba) using immobilised s and hrplabelled s receptor binding domain (rbd) of known high specificity (> . %) to compare antibody reactivity in unselected samples. of these, were reactive for rbd using the hybrid daba whereas the n/s elisa method determined of these to be seropositive. all samples found reactive in the n/s elisa were also shown to be reactive using the hybrid daba. thus, comparing to hybrid daba, our elisa showed sensitivity % ( % ci - , and specificity % ( % ci - ), % cis were calculated using the clopper pearson method. testing for active the presence of sars-cov- buccal swabs obtained during the home visits were used to test for current sars-cov- rna (and by inference active replication) using rt-pcr as previously described by lista et al. briefly, nasopharyngeal swab samples were heat inactivated at °c for min and µl were used to extract rna with the beckman coulter rnadvance blood kit, ending with elution in µl of water. for qpcr, the us cdc designed primer/probe set were used for the n gene (n and n ) and rnasep with µl of rna sample and taqman fast virus - step master mix (thermofisher). longitudinal experience of potential covid- symptoms was prospectively logged by participants via the c- covid symptom study app, the primary app for self-report of symptoms during the pandemic in the uk. the c- symptom study app has been developed by the health science company zoe in collaboration with king's college london. thus far, , , participants have logged symptoms, providing invaluable epidemiologic information with regards to the pandemic. the symptoms which app participants were asked to record are listed in table . algorithm for predicting prior covid- from symptoms reported a predictive algorithm for identification of covid- using longitudinal symptoms reported via the c- app was used to identify participants who were predicted to have had prior symptomatic infection, using the method recently described by menni et al. briefly, the algorithm was developed from symptoms in people testing positive for sars-cov- using rt-pcr. the formula includes two core symptoms (anosmia and cough), two non-core symptoms (fatigue and skipped meals), in addition to participant age and sex. in addition, participants were asked whether they had experienced any symptoms suggestive of covid- prior to launch of the app on th march . consideration of serology in relation to longitudinal symptoms logs and statistical analysis participants had regularly logged their symptoms prospectively from app launch on th march, and also reported on symptoms retrospectively prior to this date. symptom profiling included core covid- symptoms and general symptoms, in addition to algorithm prediction of prior covid- . participants were delineated according to antibody status in relation to reported symptoms and predicted covid- prior infection. using the status regarding prior sars-cov- infection as predicted by the algorithm, we calculated seroprevalence amongst those who were likely to have had covid- . in these participants, we investigated association of demographic factors: age, sex, and bmi, with seroprevalence (student's t test for difference between groups). confidence intervals for proportions were calculated using the clopper pearson method. all analyses were performed using the r software environment for statistical computing. in our sample of adults aged - years living in london and south-east england, seroprevalence using our elisa assay was demonstrated to be % ( participants, % ci = . - . ). during the study period, beginning th march , a total of participants completed longitudinal logging of symptoms via the c- covid symptom study app, providing detailed insight into clinical presentation of this community sample. the median number of app entries to record symptoms, per individual, was days (iqr to ). ( %; % ci - ) were seropositive, of whom, / ( %; % ci - ) were completely asymptomatic, including no prior symptoms before the app launch. % reported only noncore symptoms, (i.e. neither fever, persistent cough, nor anosmia; table ), and would have been reported as asymptomatic in other surveys. the symptom which most strongly predicted seropositivity was anosmia. of participants reporting anosmia, . % had detectable igg to sars-cov- . the specificity of anosmia for seropositivity was %, whereas for fever, cough, and anosmia combined was less at %. both were not highly sensitive with (anosmia %; all core symptoms together %). there were individuals in the cohort who were predicted to have had a sars-cov- infection using the app based algorithm. the algorithm correlated better with serology than did core symptoms alone ( figure b; table ). of the individuals with predicted covid- , % ( ) were seropositive ( factors associated with seroprevalence amongst those with predicted covid- on comparison of the seropositive and seronegative participants with predicted covid- (n = , out of participants who prospectively logged their symptoms), seropositive participants were older (median age seropositive , median age seronegative ; p = . ). no difference in sex (% female of seropositive participants , and for seronegative) or bmi (median . seropositive; . seronegative) was evident between the groups. understanding seroprevalence is important in order to estimate epidemiological spread of covid- through the population, to inform on the potential efficacy of vaccination, and for antibody fold change above background antibody fold change above background the concept of herd immunity which may eventually be achieved via a combination of vaccination and convalescence. we estimate the seroprevalence rate within our sample to be %, higher than the ons estimate for the uk of % (studies summarized in supplementary table ), but could be compatible given the potentially higher prevalence in london and the south-east and the fact that, unlike the ons surveys, we only included adults. our results indicate that a substantial portion ( %) of those who have detectable antibodies to sars-cov- were entirely asymptomatic, confirming that even those who have clinically very mild disease with no perceivable symptoms may produce antibodies. this frequency is substantially lower than most other estimates of asymptomatic covid- , which we believe to be due to our more complete assessment of symptoms over time. indeed, this may still be an overestimate, as, at the start of the survey during the peak epidemic, we did not ask about rashes, which our data suggest are present in % of positive cases. our estimate relates to asymptomatic development of specific antibodies, rather than asymptomatic carriage of sars-cov- , as it has been reported that not all who have covid- develop detectable antibodies. , in a population study by solbach et al. of german participants who had had pcr confirmed prior covid- and who self-reported symptoms, there were ten asymptomatic individuals ( %). of these ten asymptomatic cases, four were seropositive, and the remainder had no detectable antibodies in two consecutive analyses. it is clear that asymptomatic disease holds important implications with regards to transmission, as asymptomatic individuals are unaware of their infected status, and may support a significantly longer period of viral shedding, thereby exacerbating the potency of transmission potential. , without parallel pcr testing alongside symptom tracking, we cannot be certain whether seronegative individuals reporting covid- symptoms either did not develop detectable sars-cov- antibodies, their antibody levels had declined to undetectable or whether their symptoms related to another infection or condition. within the people who were predicted to have had covid- using the app based algorithm, there was approximately a / likelihood of participants having a detectable antibody response ( % were seropositive, whilst % had no detectable antibodies). our understanding of the human immune response to sars-cov- continues to improve, and emerging evidence indicates that t cell mediated immunity plays an important role in the immune control of for example, a small community-based study of patients and their household families showed than an unexpected six out of eight family members who had been infected demonstrated a covid- -antigen specific t cell response, but had no detectable antibodies. speculatively, therefore, t cell mediated responses may afford efficacious immunity in the absence of a detectable antibody response. in the participants with symptoms predictive of covid- using the algorithm, age differed by antibody response: seropositive individuals being older than those who were seronegative (p = . ). this may be secondary to increased shielding of older participants, however it is consistent with previous reports of a positive correlation of igg covid- antibodies with age. , whether t cell mediated reponses are present or more prominent in younger individuals is part of an ongoing follow-on of this study. using longitudinal symptoms logged using the app, we demonstrated that of all the individual clinical symptoms, anosmia was the strongest indicator of seropositivity; the specificity of anosmia was % and sensitivity was %, whilst the specificity of all core symptoms combined was %. using data from the same covid symptom study app, our group was able to demonstrate previously that anosmia is a core symptom of covid- , in addition to the prior recognised symptoms of fever and persistent cough. , . these results further highlight anosmia as an important core covid- symptom which correlates strongly with both swab positivity and antibody response. this study, like many similar surveys has a number of limitations. an overarching consideration with regards to seroprevalence to sars-cov- is the longevity of seropositivity. in a longitudinal study of rt-pcr confirmed prior covid- cases by doores et al., individuals mounted a range of antibody responses, and a decline in levels and virus neutralisation was typically observable within three months of the onset of symptoms. thus, it is plausible that antibody responses would have fallen below the level of detection by the time of assessment in some of our study participants. estimation of prior covid- infection was via a symptom based algorithm as pcr confirmation had not been undertaken, nevertheless, we have previously published that this algorithm had a high ppv of close to percent and was trained on symptoms of swab positive individuals. no volunteer cohort is fully representative of the general population, but we sampled a range of age and ethnicities and included people from a wide area of deprived and affluent neighbourhoods and with a range of bmi. despite this, the cohort is detectably more affluent and white,than the general population which would serve to reduce our estimate of prevalence, give the extra burden of disease shouldered by less affluent groups and people of black, asian and minority ethnic background. additionally we use the index of multiple deprivation which is an area-level indicator rather than individual level socioeconomicposition. the excess of females in our data set reflects the twinsuk cohort and, so far, differences in seropositivity between genders have been minor in other datasets. it is possible that women report symptoms more readily than men which means that the true assymtomatic rate may be higher. the data are from london and the south east and seroprevalence data from this study will therefore not be generalisable to the whole of the country or to children. further work is planned to extend the study using larger numbers from the million app users who reported symptoms to address many of the underlying factors influencing swab positivity and antibody responses. the present study is underpowered in this regard with the asymptomatic seropositive group comprising individuals out of a total of seropositive participants. in conclusion, we estimated that % of the se england population were seropositive between th april and nd june, an average of days from the peak of the epidemic in the uk. we estimate the asymptomatic rate to be %. anosmia was the symptom with the highest specificity for seropositivity. these data should be useful for both epidemiology and public health planning and reinforces the need to collect good symptom data as neither pcr testing nor antibody tests adequately capture all disease. ons diagnostic value of skin manifestation of sars-cov- infection convergent antibody responses to sars-cov- in convalescent individuals antibody profiling of covid- patients in an urban low-incidence region in northern germany clinical and immunological assessment of asymptomatic sars-cov- infections prevalence of asymptomatic sars-cov- infection longitudinal evaluation and decline of antibody responses in sars-cov- infection robust t cell immunity in convalescent individuals with asymptomatic or mild covid- sars-cov- -specific t cell immunity in cases of covid- and sars, and uninfected controls intrafamilial exposure to sars-cov- induces cellular immune response without seroconversion. infectious diseases (except hiv/aids) this study was supported by a covid urgent response grant from the chronic disease research foundation (cdrf). zoe global limited developed the app with the guidance of clinicians. investigators received support from the wellcome trust, the mrc/bhf, alzheimer's society, eu, nihr, cdrf, rst-ukri/mrc and the nihr-funded bioresource, clinical research facility and brc based at gstt nhs foundation trust in partnership with kcl. we thank the volunteers of twinsuk without whom this work would not be possible, and all participants who entered data into the c- covid symptom study app. we thank the staff of zoe global, the department of twin research at king's college london and the clinical and translational epidemiology unit at massachusetts general hospital for tireless work in contributing to the running of the study and data collection. we thank e. segal and his laboratory for helpful input. thank you to philip brouwer, marit van gils and rogier sanders (university of amsterdam) for the s protein construct, and leo james, jakub luptak and leo kiss (lmb) for the provision of purified n protein. cjs, ts & mhm designed and implemented the study. kjd led methods for serology. js, cg, sa, ab, kjas, oh and nh undertook the elisas, cr and mom undertook the dabas, and pmm, kp, mjl, red, hs, tjam and ejs performed the pcr work. sc analysed the elisa and app data. cjs supervised the analysis. pmw contributed to data analysis and drafted the manuscript. rb curated the demographic data. all authors contributed to reviewing of the manuscript. tds is a consultant to zoe global. jcp and jw are employees of zoe global. protein microarray using serum samples. testing of individuals in the usa.antigens: s glycoprotein, receptor binding domain (rbd), s glycoprotein and sars-cov using serum samples. . ) and were higher in participants with more than days since onset of symptoms (p-value= . ), and iga levels were higher in symptomatic than asymptomatic subjects (p-value= . ).garcia-basteiro et al. supplementary table summary of prior studies of the general population antibody response to sars-cov- .supplementary figure . antibody response in relation to symptom pattern. core symptoms are defined as fever, cough and/or anosmia. colour depicts symptom pattern of core symptoms (red), non-core symptoms (green), and asymptomatic (blue). shape depicts antibody status of detectable (positive; circle) or undetectable (negative; triangle). antibody level denotes fold change above background, and a threshold of < indicates seropositivity. predicted covid- (predcovid) is the algorithm score predicting which participants are likely to have had covid- using symptoms reported via the c- app. the threshold for predicted covid- of . is marked.visualisation of longitudinal symptom reporting using heat maps demonstrated that anosmia clearly delineated seropositive from seronegative individuals (supplementary figure ) . key: cord- - tfvmwyi authors: hoplock, lisa b.; lobchuk, michelle m.; lemoine, jocelyne title: perceptions of an evidence-based empathy mobile app in post-secondary education date: - - journal: educ inf technol (dordr) doi: . /s - - - sha: doc_id: cord_uid: tfvmwyi cognitive empathy (also known as perspective-taking) is an important, teachable, skill. as part of a knowledge translation project, we identified a) interest in an evidence-based cognitive empathy mobile app and b) which faculties believe that cognitive empathy is important for their profession. students (n = ) and instructors/professors (n = ) completed a university-wide survey. participants in education, social work, and the health sciences were among those most interested in the app. the majority of participants said that they would prefer for the app to be free or less than $ for students. most participants preferred a one-time payment option. across faculties, all but one had % or more of its sampled members say that cognitive empathy is important for their profession. results illuminate perceptions of cognitive empathy instruction and technology. results also provide insight into issues to consider when developing and implementing an educational communication app. self-reflecting; hoplock and lobchuk ; vorauer ) , it has been linked to positive outcomes such as increased patient compliance and satisfaction within healthcare (kim et al. ) , engaging in prosocial behavior (davis ) , team effectiveness, and engaging in high quality communication (parker et al. ) . people vary in their ability to engage in cognitive empathy, but fortunately, it is a skill that can be taught (e.g., brunero et al. ; richardson et al. ; teding van berkhout and malouff ) . one way that cognitive empathy can be taught is by using technology (e.g., lobchuk et al. ) . the present research examines people's perceptions of a cognitive empathy-training mobile app as well as how perceptions of cognitive empathy's importance vary by profession. cognitive empathy is studied in a variety of fields, such as healthcare (e.g., blanch-hartigan and ruben ; lobchuk et al. lobchuk et al. , , psychology (e.g., marangoni et al. ; vorauer and sasaki ) , business (e.g., ku et al. ) , and law (e.g., bandes ). one frequently-used paradigm involves filming an interaction between two people, having the interaction members report what they were thinking and feeling throughout the interaction, having the interaction members guess what the other person was thinking and feeling throughout the interaction, and then providing an accuracy score (ickes ) . this accuracy score indicates how accurate a person is at inferring the thoughts and feelings of another. accuracy is associated with relationship satisfaction (sened et al. ; thomas and fletcher ) , skillfully providing social support (verhofstadt et al. ) , and accommodating during conflict (kilpatrick et al. ) . thus, accuracy is helpful for facilitating successful communication and relationships across contexts. people's perspective-taking ability improves when they are self-aware of personal values/emotions that can thwart empathy (lobchuk et al. ) ; receive instruction (lelorain et al. ) and feedback (noordman et al. ) , and self-evaluate with video-feedback (fukkink et al. ). the authors have been conducting iterative research on an intervention that takes these findings into consideration, incorporating ickes' ( ) paradigm and including perspective-taking instruction. however, currently, the intervention involves coming into a lab for the filming. to increase accessibility and cost-effectiveness, the intervention could be adapted as a mobile app for use within any setting, not just the classroom. most students world-wide own a smartphone (e.g., farley et al. ; nason et al. ; o'connor and andrews ; williamson and muckle ) . while mobile devices such as smartphones and tablets (crompton and burke ) are sometimes seen as a distraction in the classroom, they are increasingly being leveraged to facilitate learning (langmia and glass ; nguyen et al. ) . indeed, with classes being forced online due to covid- , it is expected that instructors will incorporate technology into their courses more than ever before (bates ) . it is important that devices are used in a way that facilitates taking an active role in learning (norris et al. ) . when they are used in that way, mobile devices may improve student motivation and productivity (cotter et al. ) , confidence (koohestani et al. ) , and performance (dunleavy et al. ; hsueh et al. ) . for example, a recent metaanalysis of healthcare research on using mobile technology within education (mlearning) found that participants who experienced mlearning tended to have superior knowledge and skills than those who received traditional education (dunleavy et al. ) . thus, growing evidence supports the use of mlearning. we are interested in people's perceptions of using an app for teaching and practicing cognitive empathy. research findings do not always get translated into practice in fields like education, and it can take a long time for them to be implemented if they do (burkhardt and schoenfeld ) . we follow the canadian institutes of health research knowledge to action process (kta; government of canada ) framework to expedite the integration of our evidence-based intervention into students' learning activities. according to this framework, knowledge is created (e.g., through research) and then translated into application through an iterative cycle. knowledge creation includes inquiry, synthesis, and product-creation (government of canada ). we aim to move our research towards the product-creation phase and ensure that it becomes incorporated more quickly into practice. we currently conduct the intervention with students from a variety of healthcare disciplines (e.g., nursing, occupational therapy, kinesiology; (e.g., lobchuk et al. lobchuk et al. , . one version of the intervention has also included videoconferencing (hoplock and lobchuk b ). yet, this intervention could be made more accessible if it were converted to mobile technology. many students and instructors/professors own mobile devices and use apps, making them the ideal population to assess the potential use of mobile learning for educational purposes in an academic setting. it is important to conduct market research with the population of interest (university students and instructors/tenure-track and tenured professors), so that we understand potential app users and the merit of creating the app before we start creating it. we want to ensure that what we create is meaningful to the people who will use it and that it will be used within communication skills curricula. to date, we have conducted intervention research with health professionals. however, it is possible that the potential intervention user-base is larger than just those in the health profession (e.g., law, education, or business). thus, we circulated a university-wide survey to students and instructors/professors to better understand who might be interested in the intervention and cognitive empathy. the purpose of the present research is to identify the target market, demand, and price point as well as to solicit student and instructor/professor perceptions of the empathy-training mobile app. we had the following research questions: this work may appeal to people who study empathy, education technology, marketing, and business. this research may also help people who want to create related apps. cognitive empathy is a valuable skill for successful interpersonal relationships (e.g., batson and ahmad ; davis ; davis ; galinsky et al. ) and so converting a successful intervention to be able to teach empathy accessibly is a worthy goal. after obtaining ethics approval, we conducted a survey to answer our research questions. in accordance with simmons et al. ( simmons et al. ( , , we report how we determined our sample size, all data exclusions, all experimental manipulations (there were none), and all study measures. our aim was to give all students and instructor/professors the opportunity to describe their opinions or attitudes toward our empathy application. a census sampling frame was determined to be most appropriate. this sampling frame aims to collect information from every eligible member of the population. our decision was not to exclude any student or instructor/professor at the university so as to boost our success in accruing a representative sample of students and faculty by taking a census sampling approach. because this work was exploratory, we did not conduct a formal power analysis. instead, we aimed to recruit at least students (hao et al. ) and instructors/professors (vrana ) or as many participants as we could before our stopping rule: complete data collection by the end of december, (approximately month after data collection started). we chose this stopping rule due to time constraints. participation was restricted to students and instructors/professors over the age of . approximately , students and instructors/professors were emailed a survey link; people accessed the survey. data from participants were retained ( students; instructors/professors) after exclusions (n = did not provide consent; n = did not fully complete the survey; one person provided nonsensical responses). most participants identified as white ( % students, % instructors/professors) and as women ( % students, % instructors/professors). the average ages of students and instructors/professors were . years (sd = . ) and . years (sd = . ) respectively. participants volunteered for an online study on "perceptions of an evidence-based empathy mobile app in post-secondary education." at the start of december , they were emailed a study description and link. they were told that they would have until the end of the month to complete the study. a reminder was sent mid-way through the month. participants did not receive compensation. materials can be found on the open science framework: https://osf.io/bh su/?view_only= ff fe e c f d ac f c . the majority of questionnaire items were author-created and inspired by the literature (e.g., alwraikat and tokhaim ; sevillano-garcia and vazquez-cano ; vrana ) . we obtained suitability and coverage feedback on our survey from experts in technology development at the university's technology transfer office. the overall structure of the questionnaire was the same for students and instructors/professors, however the content of the questions varied. students completed questions regarding using the app as part of their education. instructors/professors completed questions regarding using the app as a teaching resource as well as for personal use. after providing informed consent, participants were first presented with a description of the app's purpose and how it would work. they saw two wireframes to give them a sense of the app idea and what the app might look like. next, they were asked their perceptions of the app. students rated their agreement on four statements ( = strongly disagree, = strongly agree), providing their perceptions of the app as being applicable to their profession, their perceptions of the app helping them to meet their needs, their liking using mobile learning as part of their educational curriculum, and their confidence in using mobile technology to achieve their learning goals. instructors/professors were asked to rate their agreement with six similar statements using the same scale. four of the statements related to using the app for teaching or professional work, one related to their perceptions of using mobile technology as an education tool, and one related to their confidence in using mobile technology to achieve their teaching goals. participants then indicated their interest in the app, with faculty indicating both their interest in the app as a teaching tool ( item) and their personal interest in the app ( item; = very disinterested, = very interested). using an open-ended question format, participants provided their reaction to the app (adapted to an open-ended question format from a -pt likert scale from surveymonkey n.d.). they then indicated how likely they would be to consider buying the app ( = extremely unlikely, = extremely likely; adapted from a -pt scale; surveymonkey n.d.). those who selected or on the scale were asked an open-ended question regarding why they were unlikely to consider buying the app. faculty were also asked how likely they would be to recommend that student buy the app ( = extremely unlikely, = extremely likely; adapted from a -pt scale and to this context; surveymonkey n.d.). those who selected or on the scale were asked an open-ended question regarding why they were unlikely to consider recommending buying the app. to better understand price point and pricing strategies, we asked participants to check all that apply when considering what one-time price they would feel comfortable paying for the app ($ . ; $ . -$ . ; $ . -$ . ; $ . -$ . ; over $ . ), what payment type they would prefer (subscription (i.e., several smaller payments); one-time payment (i.e., one larger payment)), and what payment version they would prefer (consumers may choose between a basic version of the in your shoes mobile app that is free and a version that has extra features and a cost; consumers use a free trial of the in your shoes mobile app with extra features and payment is required later). students responded while imagining that the app was required in a course and then responded imagining that the app was not required in a course. instructors/professors were asked what they would feel comfortable asking students to pay if the app was required in a course, and then if it was not required. instructors/ professors were also asked what they themselves would pay if using the app for personal use. participants next completed demographics questions asking about gender, age, ethnicity, income (statistics canada ), smartphone or tablet ownership and use (chen and denoyelles ), device brand (adapted from chen and denoyelles to ask about what brand they primarily use instead of what device they own), hours spent on their phone or tablet for things related to work/school (chen and denoyelles ), whether they use an app for coursework (adapted from chen and denoyelles to ask whether they have used an app instead of how often), their faculty, college, department, and class format (in-person/in-class; online/distance; mix of in-class and online courses). students were asked about their program year, degree, and student status. instructors/ professors were asked their academic rank and teaching experience (the latter was adapted to include more options for those with fewer years of experience; alwraikat and tokhaim ). we also wanted to know if participants' respective programs had at least one course that focuses on interpersonal communication (yes, no, unsure) . those who answered "no" or "unsure" were asked if interpersonal communication was incorporated into their curriculum to some degree. finally, participants rated whether cognitive empathy was important for their profession ( = not at all, = extremely). we employed a descriptive, cross-sectional, online survey with participants. with all students and instructor/professors at the university having been invited to participate, we had the opportunity to 'drill down' and conduct exploratory analyses of linkages between student and instructor/professor characteristics and their opinions or attitudes toward the empathy application. descriptive statistics (medians, means, standard deviations, frequency counts, and percentages) were used to describe the sample of students and instructor/professors, as well as address research questions to . we ran the responses to the open-ended questions through a sentiment analyzer and a word cloud generator located on danielsoper.com to get an objective sense of participants' sentiment towards the product. sentiment analyzers use "computational linguistics and text mining to automatically" determine the overall degree of negativity (− ), neutrality ( ), or positivity ( ) in the text (soper n.d.) . word cloud generators analyze text to determine whether certain words are used more frequently. a strength of these two approaches is that it analyzes the results impartially and, thus, will help validate results found using our other methods. a limitation of these approaches is that they examine the overall text, without nuance. we also analyzed the responses to the open-ended questions using content analysis (see online supplemental materials for additional details; lincoln and guba ; patton ) . credibility was established by recruiting participants from the target population (elo et al. ) . dependability and confirmability were met with an audit trail documenting coding decisions and template development (e.g., saldana ). confirmability also occurred through independent coding and analysis and by an iterative feedback process until consensus was reached (graneheim and lundman ) . we also used participants' own words for codes and themes when possible to ensure that we stayed close to the data (levitt et al. ). there are faculties at the university. at least one student from every faculty participated. instructors/professors from faculties participated. therefore, we obtained diversity in the faculties that participated. over % of participants owned and used a smartphone or tablet, and most participants ( . % students; . % instructors/ professors) used apple products. for students, % have used a mobile app required or suggested by their instructor for course work. for instructors/professors, % have required or suggested to students to use a mobile app for course work. most participants preferred the app to be free ( % of students; % of instructors/professors) or less than $ for students ( % of students; . % of instructors/professors). participants also preferred a one-time payment option ( % of students; % of instructors/professors) as well as the ability to choose between a basic version of the app that is free and a version that has extra features and cost ( % of students; % of instructors/professors; see online supplemental materials for other participant details). addressing research question (who would want to use this app), % of students and % of instructors/professors said that their program features at least one course that focuses on interpersonal communication. of those who said that their program did not feature an interpersonal communication course or that they were not sure if it did, % of students and % of instructors/professors said that interpersonal communication is incorporated into the curriculum to some degree. faculties from where most professors and instructors indicated that they had an interpersonal communication course in their program included business, education, health sciences, law, and social work. instructors/professors from the arts, education, health sciences, and law were among those most interested in using the app as a teaching tool (table ) . instructor/professors from business; education; environment, earth, and resources; and law were interested in using the app for personal use (table ). students in agriculture; art (e.g., fine art); arts (e.g., sociology); education; environment, earth, and resources; health sciences; music; science; social work; extended education; and university (a faculty for those just starting at the university) were interested in the app (table ) . thus, these faculties may be a good target for uptake of the app. when analyzing participants' reactions to the app, one theme that emerged, target or use case, involved commenting on the target people or use case (i.e., when or in what contexts someone might use it) for the app. this theme's categories included that the participant thought the app was not relevant to them; the participant commented on who the target audience might be; and the participant commented on potential use cases for the app (table ) . for example, one participant (woman, instructor/professor, faculty of agricultural and food sciences) wrote, "looks great for areas where you need to communicate with a patient/client on more than one level. not so much in my area." another (man, student, faculty of health sciences) wrote, "the app sounds fun and with right pricing could be helpful to new international students." these responses help narrow the target market. addressing research question (what people from the target population think of the app idea), sentiment analysis and word cloud generator results of the open-ended questions indicated that responses were generally neutral to negative. students' responses to "what is your reaction to the in your shoes mobile app" were neutral ( . ; possible range = − to ). the word cloud indicated that responses were focused ratings were made on a -pt scale with higher numbers indicating greater interest in the app around empathy, apps, and the perception that the idea is interesting (see online supplemental materials for the word clouds). instructor/professor responses were somewhat negative (− . ). their word cloud was somewhat similar to that of the students, but was also focused on students, teaching, and learning. diving deeper, when analyzing participants' reactions to the app, two additional themes emerged: participants described practical issues with the app, which might affect uptake and participants commented on the concept idea (table ). the first theme's categories included barriers to empathy and barriers relating to the intervention procedure. for example, one participant (man, student, university faculty) wrote, "seems like a great idea but would be a little weird to find a partner to record conversations and get them to tag their thoughts and feelings." these responses help identify potential concerns and factors to watch out for when creating the app; for example, finding the right dialogue partner and drawing on a relevant context to engage in a meaningful dialogue. the second theme had four categories: ) positive reactions to the app idea (e.g., favourable evaluation of the app, willingness to try the app, and visualizing positive outcomes as a result of using the app); negative reactions to the app idea (e.g., unfavourable evaluation of the app; doubts in teaching empathy with an app; negative comments about empathy training in general; and comments about the app not practical issues with the app: participant describes barriers to using the app, which might affect uptake barriers to empathy: participant describes barriers that relate to empathy "it seems impractical -a lot of people would probably be very uncomfortable with the exercise described above and would not want to put in the time to tag a recording of their conversation." (woman, student, faculty of arts) procedure: participant describes barriers that relate to the intervention protocol "the instructions are a little complicated, which i believe will prevent some students from using it unless they really want to put in effort. it would be better if scenerios (sic) were supplied in addition to this feature so it didn't require two people to use." (man, student, university ) these responses indicate varying support for the app idea and help clarify the sentiment analysis and word cloud results. participants who indicated that they would be unlikely to buy the app (i.e., they selected or on the question "how likely are you to consider buying the app") were asked about their response. students were understandably somewhat negative given the nature of the question (− . ). their word cloud indicated that responses focused on money, apps, and the word "don't." (e.g., "i don't buy apps"). instructors/professors' responses were unexpectedly positive ( . ). the associated word cloud focused on empathy and technology needing to add value to the course. three themes emerged from qualitative analyses that help us to understand the disinterest in buying the app: ) responses indicating that participants did not think they fit the target market; ) responses about money or purchasing apps in general; and ) responses about the app itself (table ). like the target or use case theme for who would use the app, the first theme's categories included that the app is not relevant or that they lack interest in it. for example, one participant (woman, instructor/professor, faculty of social work) wrote, "i have taught interpersonal communication skills for approximately yearsnot necessary" the second theme's categories included rarely or never spending money on apps; preferring a free app or a free alternative to the app; and that the app would not be a priority purchase (not worth the money). for example, one participant (agender, student, faculty of arts) wrote, "i don't buy apps." another (woman, student, school of business) wrote, "i would only get the app if it had a free trial period to try it out." the third theme's categories related to perceptions of the app's effectiveness and the app's protocol (e.g., the app is impractical, privacy concerns, and confusion about the app's use case or procedure). for example, one participant (man, student, faculty of engineering) wrote, "low chance of repeated use, requires time commitment and analysis." together, these responses provide additional insight into potential concerns, factors to consider when creating the app, and the target market (e.g., offer a free trial, ensure reliable security, promote broad applicability). it is possible that negation within the responses affects scoring as all responses except one included the word "not." thus, responses like "not necessary" may have been interpreted positively. according to the site, "this tool produces an overall sentiment score. although various passages within a sample of text may be particularly positive or negative, the sentiment score produced by this tool considers all of the text in the sample… research shows that in about % of all cases human beings will disagree about the sentiment of written text." (https://www.danielsoper.com/sentimentanalysis/default.aspx) faculty who indicated that they would be unlikely to recommend buying the app (i.e., they selected or on the question "how likely are you to consider recommending buying the app") were asked about their response and (as would be expected given the nature of the question) responses were quite negative (− . ). the word cloud highlighted a focus on the classroom, skill development, empathy, and being unconvinced. three themes emerged from qualitative analyses: ) responses indicating that the app is not relevant to the participant or their field (mirroring responses to being unlikely to buy the app); ) responses indicating skepticism that empathy can be taught with an app (i.e., they are not convinced); and ) responses about money or purchasing apps (table ) . for example, one participant (did not report gender, instructor/ not relevant/no need: participant states that the app is not relevant to them or their field or states disbelief regarding the need for an app to teach empathy (e.g., they say that they are already empathetic) "not appropriate to my area of teaching" (man, instructor/professor, faculty of environment, earth, and resources) lack of interest: participant lacks interest "there would never be a time when i would want to use this. the only way i would ever use it is if were part of a leadership training thing." (man, student, faculty of science) about money: participant makes a comment related to money or purchasing apps rarely or never spend money on apps: participant says that they do not like purchasing apps or that they do so rarely "i never pay for apps" (woman, student, faculty of agriculture and food sciences) free app or free alternative: participant indicates that they would download the app if it were free, otherwise they would prefer a free alternative professor, faculty of arts) wrote, "i don't think they would use it and i'm not sure that i feel that an app is the best approach to learning empathy, so i'm not convinced on its utility." while another (did not report gender, instructor/professor, did not report faculty) wrote, "students have better things to spend their money on, however small the amount." together, these responses provide additional insight into the target market and instructor perceptions of using technology to teach empathy. addressing research question (which professions believe that cognitive empathy is important to their profession), % of students and % of instructors/professors said that cognitive empathy is important for their profession. most participants ( % or higher) from all faculties except one (school of agriculture) thought that cognitive empathy was important for their profession (table ). the present research helps us progress towards the product-creation phase within the kta framework to facilitate uptake of an evidence-based intervention more quickly into the classroom. like other research (farley et al. ; nason et al. ; o'connor and andrews ; williamson and muckle ) over % of participants owned and used a smartphone or tablet. some, but not the majority, of instructors/professors were asking students to use mobile apps for their course work (see also ariel and elishar-malka ) . this amount is likely to increase spurred by the covid- pandemic. we also found that when considering the empathy-based app for class use, qualitative results echoed those found in previous research: participants mentioned usefulness, money, the ability to use a trial version, and enjoyment (kim et al. ). these variables are important because they predict intention to purchase an app (kim et al. ). together, results indicate a potentially large market for apps within higher education and that apps might be successful if perceived as useful, enjoyable, and triable. these findings are timely because there is a worldwide movement in education toward remote learning and using technology to deliver content (e.g., bates ). our research adds to the literature by illuminating perceptions of app cost within education. the majority of participants said that they would prefer for the app to be free or less than $ for students. most participants preferred a one-time payment option. they also preferred the ability to choose between a basic version of the app that is free and a version that has extra features and a cost over a free trial and later payment. qualitative responses indicated that some participants just do not purchase apps. additionally, student participants described being stretched for resources and unable to spend money on apps. apps tend to be free and people often prefer free alternatives to having to pay (hsu and lin ) . thus, making the app free may increase uptake. app developers and marketers should consider making apps free for students and passing the cost on to institutions, when possible (see table for all recommendations). doing so would allow a greater number and diversity of people to learn about concepts such as empathy. apps provide opportunities to advance changes in how content is being taught in and outside of the classroom. currently, when instructors and researchers want to use paradigms like ickes' ( ) , they often require students to come into a lab (e.g., lobchuk et al. ) . this makes participating unfeasible and costly for many. indeed, the present research indicates that cost is top of mind for students. the ubiquity of (lobchuk et al. ) ; receive instruction (lelorain et al. ) and feedback (noordman et al. ) ; and self-evaluate with video-feedback (fukkink et al. ) so that their perspective-taking ability improves. the present research indicates that people recognize the value that a cognitive empathy app can hold and how it can be used. the present research also highlights concerns that people have about using apps for empathy instruction. we found that some people are wary of using their personal devices for recording conversations and being vulnerable. these people might benefit from coming into the lab and participating in a safe experience before using the app, as well as from receiving tips on finding a dialogue partner and drawing on relevant context to engage in a meaningful dialogue. testing to ensure app security and communicating this security to users will also be important as will be ensuring a seamless experience and reducing procedural barriers via usability testing. additionally, our research finds that some people might need to be convinced of the benefits of using technology to teach empathy skills. developing a compelling marketing strategy that demonstrates how mobile devices can improve confidence (koohestani et al. ) and performance (dunleavy et al. ) , and cites evidence of the intervention's effectiveness might be beneficial here. the present research indicates that many people recognize the importance of cognitive empathy for interpersonal relationships and for their profession. convincing them that it can be taught with technology would ensure that more people get the instruction that they need. consistent with previous research (keyworth et al. ) , students want communication skills training. while previous research has studied cognitive empathy within a variety of fields (e.g., teding van berkhout and malouff ), the present research indicates that more fields are interested in it than likely previously thought. participants were interested in the app idea, with those in education, social work, and health sciences among the most interested. even participants in the physical sciences were interested and believed that cognitive empathy is important to their profession despite other research finding that being low in cognitive empathy predicts enrollment in the table key recommendations to develop a mobile empathy app recommendations . make the app free or for a minimal cost to students and pass the cost to institutions, when possible . make the app a one-time payment for students . prioritize targeting people in the fields of business, education, health sciences, and law . empathize the benefits of dialoguing, self-reflection, and actively taking another person's perspective in marketing communications . provide tips on how to find a dialogue partner and draw on relevant context to engage in meaningful dialogue . engage in testing to ensure app security and communicate this security to users . conduct usability testing to ensure a seamless user experience and reduce procedural barriers . develop a compelling marketing strategy that highlights benefits gained across a variety of disciplines as indicated from previous research. developing concrete examples of when the app could be used and including testimonials would be helpful physical sciences (thomson et al. ) . perhaps they are aware of being low and want to improve their perspective-taking skills because they believe it is important for their profession. future research should investigate this further. the confusion and questions about the app protocol as well as the varying sentiment about the app indicates that more research is needed. future research will iterate app descriptions and wireframes to increase clarity and value of the app. scenarios where the app might be used will be guided by research and used in marketing so that the value of the app (e.g., trust-building; blatt et al. ) is evident to a wider range of people. one reason why increasing perceived value will be important is because it predicts app purchase intentions (hsu and lin ) . while the present research was a necessary first step into assessing perceptions of the concept idea, future research will also indicate how perceptions of the app and the idea of using technology to teach cognitive empathy changes as people start interacting with app prototypes. one limitation to this study is participant self-selection bias: the people who opted to participate in a study about empathy may have been more interested in this topic than others. this is evident, for example, in the number of instructors/professors who indicated that they taught courses having to do with interpersonal communication ( %), as well as the greater participation from people who participated in certain fields (i.e., health sciences, education). while we received student participation from every faculty area, this participation was not equal and there was less representation among instructors/professors. it is possible that we would have received a wider variety of responses had participants not known the app's topic (empathy) prior to starting the study. stating that the study focused on an instructional app or an app on communication skills (without mentioning empathy) might have increased the variability. generalizability is also limited because the sample was restricted to one canadian university. future pre-registered research examining perceptions of empathy's importance across disciplines should include a greater number of participants from diverse areas and could examine interest in an empathy app with a non-student population. for example, corporations may have more resources and be more interested in furthering the empathy of their employees than those within educational institutions. additionally, the survey was accessible during a busy month (december), thus people may have been more likely to participate had it occurred in a different month. moreover, the survey contained some adapted and investigator-developed questions that would benefit from additional validation. future research could conduct cognitive interviews to ensure that questions are being interpreted as they were written. despite these limitations, this research has a number of strengths, including the use of multiple methods and analysis techniques (morse ) . our methods allowed us to identify potential barriers to uptake, which is a step in the kta framework (graham et al. ) . potential user feedback spotlighted critical features to incorporate in an empathy app that is designed to foster targeted, meaningful, and efficient learning which can be differentiated based on discipline or profession. additionally, to our knowledge, this is the first study that has examined perceptions of the importance of cognitive empathy across a broad range of fields/professions. our finding that participants in a variety of fields/professions find cognitive empathy important will be of interest to people studying empathy, marketing, and education. taking the perspective of another and understanding where they are coming from is an essential skill for many professions. thus, being able to teach cognitive empathy effectively and accessibly is an important goal for instructors. as impacted by the covid- pandemic, efforts in restructuring how we teach in basic and continuing education are reflective of increasing reliance on technology-based learning. not everyone is enthusiastic about using technology in empathy education. some people are concerned that technology impedes empathy and learning by, for example, acting as a barrier between people. however, we believe that, done correctly, technology can instead act as a bridge. availability of data and material consent was not provided to share non-aggregated data. materials and word clouds are available on the open science framework: https://osf.io/bh su/?view_only= ff fe e c f d ac f c funding information this research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors, nor has the app development received any funding to date. conflicts of interest/competing interests the results of the present research are being used to guide development of an evidence-based empathy app that the first two authors are creating. code availability not applicable. exploring the potential of mobile learning use among faculty members learning in the smartphone era: viewpoints and perceptions on both sides of the lectern. education and information technologies empathetic judging and the rule of law online enrolments after covid- : some predictions for canada training clinicians to accurately perceive their patients: current state and future directions does perspective-taking increase patient satisfaction in medical encounters? a review of empathy education in nursing improving educational research: toward a more useful, more influential, and better-funded enterprise exploring students' mobile learning practices in higher education perceived benefits of mobile learning devices for doctoral students in a school of allied health professionals the use of mobile learning in higher education: a systematic review developing empathy in nursing students: a cohort longitudinal study a perspective on cultivating clinical empathy oxford library of psychology. the oxford handbook of prosocial behavior mobile digital education for health professions: systematic review and meta-analysis by the digital health education collaboration qualitative content analysis: a focus on trustworthiness how do students use their mobile devices to support learning? a case study from an australian regional university video feedback in education and training: putting learning in the picture perspective-taking and self-other overlap: fostering social bonds and facilitating social coordination lost in knowledge translation: time for a map? the journal of continuing education in the health professions qualitative content analysis in nursing research: concepts, procedures and measures to achieve trustworthiness influential factors for mobile learning acceptance among chinese users are perspective-taking outcomes always positive? challenges and mitigation strategies qualitative insights into adapting an in-lab empathy intervention for use via videoconference with groups of nursing students what drives purchase intention for paid mobile apps? an expectation confirmation model with perceived value an app to enhance resident education in otolaryngology measuring empathic accuracy nursing students' perceptions of obesity and behaviour change: implications for undergraduate nurse education empathic accuracy and accommodative behavior among newly married couples the effects of physician empathy on patient satisfaction and compliance. evaluation & the health professions investigating decision factors in mobile application purchase: a mixed-methods approach. information and management the educational effects of mobile learning on students of medical sciences: a systematic review in experimental studies the promise and perversity of perspective-taking in organizations coping with smart phone 'distractions' in a college classroom a systematic review of the associations between empathy measures and patient outcomes in cancer care journal article reporting standards for qualitative primary, qualitative, meta-analytic, and mixed methods research in psychology: the apa publications and communications board task force report naturalistic inquiry impact of patient smoking behaviour on empathic helping by family caregivers in lung cancer development of a novel empathy-related video-feedback intervention to improve empathic accuracy of nursing students: a pilot study heart health whispering: a randomized, controlled pilot study to promote nursing student perspective-taking on carers' unhealthy lifestyle choices empathic accuracy in a clinically relevant setting critical analysis of strategies for determining rigor in qualitative inquiry the use of smartphone applications by urology trainees. the surgeon ipads in higher education -hype and hope communication-related behaviour change techniques used in face-to-face lifestyle interventions in primary care: a systematic review of the literature using smartphones as essential tools for learning smartphones and mobile applications (apps) in clinical nursing education: a student perspective building better workplaces through individual perspective taking: a fresh look at a fundamental human process thousand oaks nursing therapeutics: teaching student nurses care, compassion and empathy the coding manual for qualitative researchers empathic accuracy and relationship satisfaction: a meta-analytic review the impact of digital mobile devices in higher education false-positive psychology: undisclosed flexibility in data collection and analysis allows presenting anything as significant a word solution tax filers and dependants with income by total income, sex and age product testing survey template the efficacy of empathy training: a meta-analysis of randomized controlled trials mind-reading accuracy in intimate relationships: assessing the roles of the relationship, the target, and the judge empathy or science? empathy explains physical science enrollment for men and women the role of cognitive and affective empathy in spouses' support interactions: an observational study the case for and against perspective-taking distinct effects of imagine-other versus imagine-self perspective-taking on prejudice reduction acceptance of mobile technologies and m-learning in higher education learning: an explorative study at the faculty of humanities and social science at the university of zagreb students' perception of technology use in nursing education publisher's note springer nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations key: cord- -l icku i authors: olaoye, omotayo; tuck, chloe; khor, wei ping; mcmenamin, roisin; hudson, luke; northall, mike; panford-quainoo, edwin; asima, derrick mawuena; ashiru-oredope, diane title: improving access to antimicrobial prescribing guidelines in african countries: development and pilot implementation of an app and cross-sectional assessment of attitudes and behaviour survey of healthcare workers and patients date: - - journal: antibiotics (basel) doi: . /antibiotics sha: doc_id: cord_uid: l icku i smartphone apps have proven to be an effective and acceptable resource for accessing information on antimicrobial prescribing. the purpose of the study is to highlight the development and implementation of a smartphone/mobile app (app) for antimicrobial prescribing guidelines (the commonwealth partnerships for antimicrobial stewardship—cwpams app) in ghana, tanzania, uganda and zambia and to evaluate patients’ and healthcare providers’ perspectives on the use of the app in one of the participating institutions. two structured cross-sectional questionnaires containing likert scale, multiple-choice, and open-ended questions were issued to patients and healthcare workers six months after the introduction of the app at one of the hospital sites. metrics of the use of the app for a one-year period were also obtained. download and use of the app peaked between september and november with pharmacists accounting for the profession that the most frequently accessed the app. more than half of the responding patients had a positive attitude to the use of the app by health professionals. results also revealed that more than % of health care workers who had used the cwpams app were comfortable using a smartphone/mobile device on a ward round, considered the app very useful, and found it to improve their awareness of antimicrobial stewardship, including documentation of the indication and duration for antimicrobials on the drug chart. it also encouraged pharmacists and nurses to challenge inappropriate antimicrobial prescribing. overall, our findings suggest that its use as a guide to antimicrobial prescribing sparked positive responses from patients and health professionals. further studies will be useful in identifying the long-term consequences of the use of the cwpams app and scope to implement in other settings, in order to guide future innovations and wider use. antimicrobial stewardship programs in hospitals are focused on optimising antimicrobial prescribing to improve individual patient care, decrease healthcare costs and combat antimicrobial resistance [ ] . the availability of accurate and up-to-date information is important to guide the right diagnosis and prescription of antimicrobials. healthcare providers' attempts to access this information are influenced by previous training, availability of the information, ability to access and leverage technology [ ] . there has been a recent rise in the use of smartphones generally across global population and it is predicted to be rising fastest in africa. there has been increased development of smartphone apps designed for use in healthcare, including in the area of antimicrobial stewardship [ ] [ ] [ ] [ ] [ ] . current research in medicine has shown that the use of mobile phones and devices in medical settings is more popular and is increasingly being brought to the fore of international research [ ] . for instance, recent studies have shown that % of smartphone users access medical information through their devices [ ] . a study by kamerow, chief scientist and associate editor for the british medical journal, revealed that there are approximately , health-oriented smartphone apps and, by the year , over million smartphone owners worldwide will use these apps [ ] . the study also highlighted that, although designed for health professionals, around % of health apps are now marketed to patients to help them monitor, evaluate, and transmit medical data such as blood pressure and body weight among other health checks [ ] . the author also stated that the use of these apps was higher amongst the younger population, females, and people who earned a higher income. similarly, results from a longitudinal study of medical doctors working at hannover medical school, germany in the summer of and spring of also revealed a rapid increase in the use of mobile devices in medical settings during patient interaction and professional collaboration [ ] . this significant increase was observed in both the frequency of use and the expansion of the areas of application of these devices. smartphones have specific features that support their increasing use in healthcare delivery and behavioural interventions. they are highly portable, more convenient, cost-effective and interconnected compared to reference books and computers, thus promoting improved communication and the sharing of knowledge, data and resources among health professionals and as well as facilitating regular updates as new data becomes available [ ] [ ] [ ] . furthermore, the ability of smartphones to use internal sensors to deduce context including emotions, location and activity has greatly increased their relevance in the consistent monitoring and tracking of health-related behaviours and healthcare delivery [ ] [ ] [ ] [ ] [ ] [ ] . in the early days of their use, there was a significant paucity of academic research on users' viewpoints and experiences with the use of these apps. the recent literature has provided positive feedback on the acceptability and workability of smartphone apps although it has also been recognized that this evolving technology may raise concerns regarding privacy and security [ , ] . in the past decade, there has been a rapid increase in the use of mobile phones in africa [ ] . there has also been a rapid integration of mobile health technologies and telecommunication into the healthcare system, especially in low and middle-income countries. in addition to this there has been an increased investment in mobile healthcare interventions including the use of these technologies for behavioural change communication [ ] . with the increasing burden of communicable and non-communicable diseases in africa, low-cost mobile health technology has the potential to make healthcare more accessible to disadvantaged communities [ ] . for example, in zambia and ghana adverse event reporting apps were developed by medical regulatory authorities in [ ] . the zambia medicines regulatory authority-zamra also launched adverse drug reaction application (adra), a new mobile application for android phone users for reporting adverse medicines reactions in [ ] . furthermore, apps have been used to identify falsified and substandard medicines in kenya [ ] . these technologies also offer great solutions aimed at improving the speed, safety and quality of healthcare provision in resource-constrained settings by providing easy access to local and international guidelines and resources. the purpose of the study is to highlight the development and implementation of an app to support prudent antimicrobial prescribing and improved antimicrobial stewardship practice; as part of the commonwealth partnerships for antimicrobial stewardship (cwpams) programme in ghana, tanzania, uganda and zambia and to conduct a pilot antibiotics , , of study assessing patients and healthcare providers' perspectives on the use of the app in one of the hospitals in ghana. the commonwealth partnerships for antimicrobial stewardship app was developed to improve antimicrobial prescribing and stewardship practices among health professionals in ghana, tanzania, uganda and zambia. the app provides, for the first time in the four countries, easy access to infection management resources to improve appropriate use of antimicrobials in line with national and international guidelines. following the launch of the app in four countries, there were downloads of the app and , guide opens within months. ghana had more page hits ( . %) than uganda ( %), tanzania ( %), zambia ( . %) and others ( . %) ( table ). the most visited section of the app was the national prescribing guidelines, accounting for . % of the total number of page hits while the section for updates on antimicrobial resistance (amr) (coming soon) was the least visited ( . %). pharmacists ( . %) and nurses ( . %) accounted for the highest number of registered users while pharmacists ( . %) and medical doctors ( . %) had the highest frequency of downloads and guide opens ( table ). a cross-sectional attitude and behaviour survey was carried out on patients and healthcare professionals to determine their attitudes/views on the use of antimicrobial prescribing guidelines by health professionals. a total of patients and health professionals participated in the survey; response rates were % and %, respectively. demographics presented in table shows that respondents comprise various age groups and educational qualifications and professions. patients' views on the use of the app by health professionals obtained using a likert scale of five options (strongly agree, agree, neutral, disagree and strongly disagree) are presented in table . more than % of patients had a positive attitude to the use of smartphone apps by health professionals and the fact that it increases the quality of healthcare offered by health professionals and quickens access to healthcare. patients' greatest concern was that the use of smart phone mobile apps in healthcare delivery could be a distraction to healthcare provision. this was followed by concerns that their data may not be protected/secure and that mobile devices may not be technically reliable enough. patients' least concern was that the health professional "may not be competent enough". the highest proportion of patients who had no concerns with their use of smartphone apps by health professionals were aged - ( . %). this was followed by patients aged and above ( . %), - ( . %), - ( . %) and - ( . %) in descending order. patients aged - had concerns with health professionals' use of smartphone apps. patients with the most concern with health professionals' use of smartphones were aged - . with respect to patients' highest level of education, patients with tertiary education ( . % had the least concern with health professionals' use of smartphone apps while patients with basic primary education ( %) had the most concern. patients' preferences for health professionals' use to access medicines information the highest proportion of patients wanted health professionals to use a computer or laptop ( . %). this was followed by smartphone mobile apps ( . %), reference books ( . %) and tablets ( . %) in descending order. a computer/laptop/reference book was preferred by . % of patients while . % preferred any of a smartphone, computer/laptop or tablet, a smartphone, computer/laptop or reference book, a smartphone or tablet, and a computer/laptop, reference book or tablet. additionally, . % of patients had no preference (n = ). thirty-eight healthcare workers (hcws) comprising of four doctors, eighteen nurses, six pharmacists and ten other healthcare workers participated in the survey. on a daily basis, mobile phones ( . %) and printed posters ( . %) were most predominantly used by the hcws, while tablets and computers ( . % each) were the least used devices (table ). mobile phones were used more than once a day by . % of healthcare workers. percentages of healthcare workers antibiotics , , of who had never used a tablet, pocketbook, printed posters and computers were . %, . %, . % and . %, respectively. healthcare workers' responses showed that many respondents had not consulted the cwpams app for antimicrobial prescribing information. the british national formulary (bnf)/national guidelines, a printed copy of standard treatment guidelines, senior colleagues and junior doctors were mostly consulted daily. in descending order, internet search engines, senior colleagues and pharmacists were consulted more than once a day. no additional source of information on antimicrobial prescribing was mentioned. an assessment of the various sources of information on antimicrobial prescribing used by healthcare workers showed that the cwpams app was mostly used by nurses and other health workers. bnf and national guidelines were mostly used by doctors ( %) and pharmacists ( . %) and least used by nurses ( . %). internet search engines were mostly used by pharmacists ( %) and least used by doctors ( %) (see figure ). pharmacists were seen to refer to their senior colleagues for antibiotic information more than doctors, nurses and other health professionals. more doctors and other healthcare workers (midwives, dispensing technicians and medication counter assistants) sought information from pharmacists than nurses. printed copies of the standard treatment guidelines were mostly used by pharmacists and least used by nurses. assistants) sought information from pharmacists than nurses. printed copies of the standard treatment guidelines were mostly used by pharmacists and least used by nurses. all responding healthcare practitioners admitted being concerned about the emergence of drug resistant infections while . % agreed or strongly agreed that these guidelines are easy to access. a total of . % stated that they preferred their senior's preferences over standard treatment guidelines. only . % preferred to use non-standard treatment guidelines for antimicrobial prescribing while . % felt the standard treatment guidelines did not apply to their patients (table ). all responding healthcare practitioners admitted being concerned about the emergence of drug resistant infections while . % agreed or strongly agreed that these guidelines are easy to access. a total of . % stated that they preferred their senior's preferences over standard treatment guidelines. only . % preferred to use non-standard treatment guidelines for antimicrobial prescribing while . % felt the standard treatment guidelines did not apply to their patients (table ). perception and assessment of the cwpams smartphone app all healthcare workers who had used the app agreed that the app was very useful, relevant to their patient population and considered it the best way to access standard antimicrobial treatment guidelines. in addition, they all felt comfortable using a smartphone on a ward round, admitting that the app increased their awareness of antimicrobial stewardship and encouraged them to challenge antibiotics , , of inappropriate prescribing and to document the indication and duration for antimicrobials on the drug chart. furthermore, participants found the country-specific standard treatment guidelines most useful. this was followed by the who essential medicines list section and the antimicrobial stewardship (ams) resource section. analysis of the cwpams app metrics revealed that the months with the highest downloads and page hits were september, october and november. the increase in september and october can be largely attributed to partnership project visits and antimicrobial stewardship interventions in all four countries. the spike in the month of november can most likely be linked to events during the world antibiotic awareness week in all four countries as well as the app promotion by the commonwealth pharmacists association during the world antibiotic awareness week. pharmacists accounted for the highest number of registered users and had more page hits and downloads than other health care professionals and workers. while this could mean that the app is more common among pharmacy teams, it calls for increased app promotion among doctors and other health professionals, who have also begun to use the app. the variations in the number of page hits and app downloads in each country can be explained by the number of partnerships in per country as ghana and uganda had the highest number of partnerships while tanzania and zambia had the lowest number of partnerships. the use of smartphone mobile apps in healthcare delivery has gained acceptance over the years among patients and health professionals in sub-saharan africa and worldwide [ ] . the cwpams app was developed by the commonwealth pharmacists association to provide easy access to medicine management information for health professionals across ghana, tanzania, uganda and zambia. in addition to providing health professionals with relevant national and international guidelines, notable advantages of the app are its usability without internet access, a feature which suits low and middle-income countries, and its easy adaptability. most recently, the app was updated to provide health care professionals across the commonwealth with links to relevant country-specific and international resources on covid- from the world health organization (who), international pharmaceutical federation (fip) and the africa centres for disease control and prevention, among other relevant sources. the pilot study showed that more than % of patients were content with their health professional's use of smartphone apps while attending to them. age and education level had an impact on the patient's acceptance of smartphone mobile technology as middle-aged patients had the least acceptance while the young and the most elderly had the greatest acceptance. patients with tertiary education had the highest acceptance for these technologies while those with basic primary education had the least acceptance. these results correlate with a study carried out in on the acceptance and use of health technology by community-dwelling elders which revealed that income, education and age were found to significantly affect the acceptance of technology in healthcare. patients with higher education and income used the internet at rates close to or exceeding the general population [ ] . another study also revealed that the acceptance of mobile phone technology among the older population was on the increase as they were found to constitute the fastest-growing group using the internet and computers [ ] . regarding patients' preferences, our survey reveals that more patients preferred their health professionals using a computer/laptop to access information over a smartphone or reference book. this can be explained by the fact that the patients' greatest concern was that smartphones could be a distraction to healthcare provision. this concern corroborates findings from a study by wu et al. which revealed that on an average, physicians' smartphones received . emails and . telephone calls, sent out . emails and initiated . telephone calls within h. the study also revealed that . % of perfusionists admitted that they had used a cellular phone for purposes other than healthcare delivery while performing their duties [ ] . on the contrary, a cross-sectional survey of adult patients in metropolitan academic and private dermatological clinics carried out in revealed that most patients ( . %) considered personal smartphones an acceptable reference tool to provide information in patient care [ ] . to access medical information more than once a day, health care workers mostly use mobile phones ( . %) and printed posters ( . %). these sources were also the most predominantly used daily ( . % and . %), respectively. this supports previous studies which have highlighted an increase in the use of smartphone mobile apps by health professionals [ ] [ ] [ ] . healthcare workers were also found to mostly consult internet search engines ( %), senior colleagues ( . %) and pharmacists ( . %) to access antibiotic prescribing information more than once a day. this demonstrates the need to involve these groups in promoting the app as they have a significant influence on antibiotic prescribing behaviours and healthcare workers' decisions. furthermore, healthcare professionals' responses to the use of the cwpams app was found to correspond with results obtained from a similar study by panesar et al. involving healthcare professionals. both studies show that the health professionals found apps useful and relevant to their patient population. they also agreed that apps encouraged them to challenge inappropriate prescribing [ ] . the concern displayed by healthcare workers for the emergence of drug-resistant infections and the use of the standard treatment guidelines as seen in table was highly impressive. healthcare workers also found the country-specific section of the cwpams app most useful. this correlates with the app metrics from all four countries which revealed that the national prescribing guidelines had the highest number of page hits from may to may . the study highlights the need for more healthcare workers, especially doctors, to use the cwpams app as app metrics and the pilot cross-sectional survey both reveal that more nurses and pharmacists than doctors had used the app. there is also the need for more focused implementation as well as app promotion at all partnership sites and among all health professionals, especially doctors who are prescribers. furthermore, there may be a need for subsequent studies to be carried out within the hospital when a higher number of healthcare professionals have used the app, in order to have a broader perspective from patients and health professionals. it would also be important to incorporate regular reminders about the app into the implementation strategy. a recently published study by lester et al. [ ] highlighted that implementing a locally appropriate, pragmatic antibiotic guideline through an app, supported by a simple educational strategy of weekly 'reminders', led to a significant reduction in third generation cephalosporin usage as well as an increase in the proportion of -h antibiotic reviews. the cwpams microguide antimicrobial prescribing app is the first of its kind to combine country-specific and international guidelines and information on antimicrobial prescribing for ghana, tanzania, uganda and zambia. hence, based on our knowledge, this study on the development, implementation and use of the app in these four countries is novel. one of the limitations is the low sample size for the surveys, which was due to the time constraint in carrying out the survey, limited time spent by patients at the waiting room of a single hospital site and health care workers' busy schedules. however, it is important to note that this section of the full study was intended to be a pilot in one setting and to provide initial descriptive findings. extensive surveying across other sites would enable a test of significance and to confirm trends. in addition, the survey encompassed a wide range of health care workers, including doctors, pharmacists, nurses, midwives and other health care workers. patients' who participated where across a broad range with respect to age and education, providing a wide perspective. the response rate was greater for patients than health professionals, most likely because patients were available to fill questionnaires whilst in waiting rooms compared to health professionals. the proportion of healthcare workers groups that responded to the survey were not comparable. this is due to more nurses and other health care workers being available in the hospital compared to doctors and pharmacists. though not all healthcare workers had used the app, there was an . % response rate from those who had used the app to questions on the use of the app. frequent updates and increased use of the app by health care workers highlight the need for further studies. the cwpams app was developed by the commonwealth pharmacists association using the microguide platform (http://www.microguide.eu). the platform provides a cloud-based service that allows local pharmacists to develop, manage, update and publish clinical guidelines to various apps for any mobile operating system including ios (apple, cupertino, ca, usa), android (google, mountain view, ca, usa), windows devices (microsoft, redmond, wa, usa) among other operating systems. it offers healthcare professionals offline access to clinical guidelines and content autonomously managed by pharmacy teams. it is also available online via https://viewer.microguide.global/cpa/cwpams. the cwpams app contains national and international guidelines listed into various sections including the who essential medicines list, surveillance tools, antimicrobial stewardship training, infection prevention and control (ipc) resources, and country-specific standard treatment guidelines. the app metrics and statistics were derived from routine data collection by horizon strategic partners. the cwpams app was developed for use by secondary care institutions that were part of the cwpams programme in four countries ghana, tanzania, uganda and zambia (s -s , video s ). one of the hospitals in the partnership was used as the pilot study site. the hospital is a secondary health facility with a -bed capacity. cwpams is a health partnership programme funded by the uk department of health and social care's fleming fund to tackle antimicrobial resistance (amr) globally. cwpams will support partnerships between the uk nhs and institutions in ghana, tanzania, uganda and zambia to work together on ams initiatives. this aims to enhance implementation of protocols and evidenced based decision making to support antimicrobial prescribing, as well as capacity for antimicrobial surveillance. further information about cwpams is available via https://commonwealthpharmacy. org/commonwealth-partnerships-for-antimicrobial-stewardship/. cwpams is being run by the commonwealth pharmacists association (cpa) and tropical health education trust (thet). the cwpams app metrics were obtained from data collected by the horizon strategic partners. these assessed the frequency of page hits, guide opens and the number of registered users and downloads. the pilot study was a cross-sectional survey with patients and healthcare workers in one of the hospital sites, six months after the introduction of the app using questionnaires adapted from panesar et al. [ ] . patients' questionnaires comprised of four sections with eight questions using a likert scale and multiple-choice questions. the first section comprised of demographics including age, gender, highest education qualification and occupation. the second section assessed patients' attitudes to health professionals' use of smart phone mobile apps in healthcare delivery. the third section was designed to obtain patients' concerns about the use of these smart phone apps, while the last section requested patients' preferences for health professionals reference ranging from a smart phone mobile app to a tablet, computer/laptop and a reference book. the health care workers' questionnaires comprised of nine sections with questions designed as a likert scale and open-ended questions. the first section obtained healthcare workers' demographics including country, specialty, year of graduation, grade, type of institution and profession and role. the eight sections following comprised of health professionals' attitudes to the use of the cwpams app and current practices. a convenience sample size determination of maximum each was used for the cross-sectional study. app metrics for user engagement evaluating the number of registered users, downloads, guide opens and page hits for various sections of the app from april to may were obtained through the microguide platform. (http://www.microguide.eu). health professionals survey: questionnaires were distributed among healthcare workers comprising of doctors, pharmacists, nurses and other healthcare workers at various points of care in the hospital including consulting rooms, nurses' station, pharmacy sections and wards. a total of questionnaires were distributed to health professionals with returned questionnaires completed anonymously. patients survey: patients' questionnaires were distributed to patients in the waiting room within the consulting area. patients' questionnaires comprised of demographic data and questions regarding attitude to the use of smartphone apps among health professionals over a one-week period. patients' consent was sought for before administration of the questionnaires. a total of questionnaires were distributed to patients based on patients available in hospital during the study period. all questionnaires (s : questionnaires) were completed anonymously with no personally identifiable information documented. study was conducted under service improvement as part of the cwpams project therefore no ethical approval was required but the ghana health service and the ghana amr platform were made aware of the pilot project. microsoft excel was used to analyse the data obtained from the pilot study using descriptive statistics. our study provides insight into the overall perception of the use of mobile apps as a means to improve antimicrobial stewardship, demonstrating general acceptance among patients and healthcare professionals. in general, the patients and healthcare workers surveyed had a positive attitude following the introduction of the cwpams app as a fundamental resource for accessing information on antimicrobial prescribing. hence, increased and more comprehensive use of all sections of the app could contribute to improved antimicrobial stewardship practices among healthcare workers and increased acceptance of the use of smartphone apps among patients. app downloads and utilization were found to be highest during partnership visits and app promotion, highlighting the need for more focused implementation and promotion of the app among all health professionals, especially doctors. further studies will be useful in evaluating the impact of the app on antimicrobial prescribing as well as guide future antimicrobial stewardship interventions. supplementary materials: the following are available online at http://www.mdpi.com/ - / / / /s : s : launch communications presentation, s : ams app-commonwealth pharmacists association (cpa) press release https://commonwealthpharmacy.org/ams-app-cpa-press-release/, video s : commonwealth partnerships for antimicrobial stewardship app https://www.youtube.com/watch?v=mj fa_algci, s : app launch posters, s : questionnaires healthcare workers and patients. antimicrobial stewardship programs in health care systems how primary healthcare workers obtain information during consultations to aid safe prescribing in low-income and lower middle-income countries: a systematic review a mixed methods pilot study to investigate the impact of a hospital-specific iphone application (itreat) within a british junior doctor cohort attitudes and behaviours to antimicrobial prescribing following introduction of a smartphone app medical students and personal smartphones in the clinical environment: the impact on confidentiality of personal health information and professionalism mobile health regulating medical apps: which ones and how much? professional use of mobile devices at a university medical center mobile, social, and wearable computing and the evolution of psychological practice health and the mobile phone therapeutic applications of the mobile phone bewell: a smartphone application to monitor, model and promote wellbeing mobile sensing for mass-scale behavioural intervention emotionsense: a mobile phones based adaptive platform for experimental social psychology research the potential of internet-delivered behaviour change interventions mhealth in africa: challenges and opportunities effectiveness of mhealth behavior change communication interventions in developing countries: a systematic review of the literature mobile phone-based behavioural interventions for health: a systematic review the economics of ehealth and mhealth updates-fda launches med safety app to improve health care delivery in ghana together we unite: the role of the commonwealth in achieving universal health coverage through pharmaceutical care amidst the covid- pandemic recent news-kenya's e-health department has begun piloting a system to curb fake goods older adults and technology use computer use by older adults: a multi-disciplinary review an evaluation of the use of smartphones to communicate between clinicians: a mixed-methods study patient perception on the usage of smartphones for medical photography and for reference in dermatology sustained reduction in third-generation cephalosporin usage in adult inpatients following introduction of an antimicrobial stewardship program in a large urban hospital in malawi this article is an open access article distributed under the terms and conditions of the creative commons attribution (cc by) license the funders had no role in the design of the study; in the collection, analyses, or interpretation of data; in the writing of the manuscript, or in the decision to publish the results. mk and lh work at horizon strategic partners who own and manage the microguide app. key: cord- - bzxdh authors: gånheim, charina; alenius, stefan; persson waller, karin title: acute phase proteins as indicators of calf herd health date: - - journal: vet j doi: . /j.tvjl. . . sha: doc_id: cord_uid: bzxdh the potential for using acute phase proteins (apps) in the assessment of herd health was studied by examining the levels of serum haptoglobin, serum amyloid a (saa) and plasma fibrinogen in relation to clinical findings and leukocyte counts in calves. two groups of calves from conventional dairy farms were studied. the animals were examined times during the first six weeks after introduction into a new environment. haptoglobin, saa and fibrinogen were analysed and weight gain, disease symptoms and treatments were recorded. analysis of antibodies against viral infections was performed. an acute phase reaction (apr) score was established at each sampling by combining the app results and total leukocyte counts. the health status differed between the two groups, although no manipulation of health had been performed, except that the group with a higher incidence of disease had a concurrent experimental infection with lungworm as part of another study. in the group with a higher incidence of disease, the mean weight gain was significantly lower, and the number of sampling days with elevated serum concentrations of apps, and the mean maximum concentrations of haptoglobin and fibrinogen were significantly higher compared to the healthier group. the apr score was significantly higher at days and of the study in the group with a higher incidence of disease. the results indicate that measurement of apps could be a useful tool for evaluation of health in calf herds. an important contributor to beef production is the specialised rearing of dairy calves for slaughter. this production system comprises transportation of young calves from their birth farms to specialised units, and mixing of individuals from several different breeders. stress due to transportation and mixing, in combination with exposure to a variety of microorganisms often leads to outbreaks of infection such as respiratory disease (dyer, ) . important microorganisms include bovine respiratory syncytial virus (brsv), bovine corona virus (bcv), bovine virus diarrhoea virus (bvdv), bovine parainfluenza virus type (piv- ), bovine adenovirus (bav), and certain bacteria, e.g., mannheimia haemolytica and pasteurella multocida (ames, ; bengtsson and viring, ; tråvén et al., ) . disease outbreaks can cause substantial costs for the farmer and result in serious animal welfare problemsan important consideration given the rising concern about animal welfare and food safety. tools for surveillance of herd health status could be a useful component of a quality assurance programme and there is a need to identify new indicators of health and disease. the acute phase response (apr) is a non-specific reaction by an individual to different kinds of tissue damage such as infection, neoplasia or trauma . the tissue damage induces a cascade of events that leads to the production of acute phase proteins (app) in the liver. haptoglobin and serum amyloid a (saa) are important bovine apps, which increase in serum for example during viral and bacterial diseases (murata et al., ; petersen et al., ) , but are absent, or present in very low levels, in healthy animals (godson et al., ; heegaard et al., ; gånheim et al., ) . moreover, sub-clinical inflammatory disorders can induce increase in app concentrations (karreman et al., ) . the concentration of fibrinogen, another app, is increased in the plasma of animals with inflammatory disorders (mcsherry et al., ) and has been used for many years to evaluate inflammatory disease in cattle (eckersall and conner, ) . lungworm infection (gånheim et al., ) , and stress due to poor housing (alsemgeest et al., ) , weaning, transportation and mixing (arthington et al., ) , can also result in increased app concentrations in cattle. apps have been considered both as potential indicators of disease and well-being in individual animals and as indicators of herd health (alsemgeest et al., ; murata et al., ; petersen et al., ) . studies show that saa (karreman et al., ) and haptoglobin (saini et al., ) have been found useful in herd screenings to identify cows with inflammatory diseases. apps may also be an important tool at the slaughterhouse to improve food safety (saini et al., ) . the aim of this study was to further evaluate the potential of using serum apps in the assessment of overall calf herd health by examining the levels of haptoglobin, saa and fibrinogen in relation to clinical findings and leukocyte counts in calves in farms specialised for beef production. two groups of calves differing in clinical health status were studied. the hypothesis was that analysis of one or several app may be used as an objective tool to evaluate animal health and management on this type of farm. two groups of male calves were used, groups a and b. all calves were of the dairy breeds swedish red and white, or swedish holstein, or of crosses between these two breeds. the calves were obtained from different conventional dairy farms, mediated through the swedish meats organization for purchase of live animals. group a consisted of calves from eight different farms. the age of the calves at the start of the study ranged from to weeks (mean weeks, sd . ). at a private commercial farm specialising in beef production, the calves were housed in a stable that had been cleaned and empty for more than two months before their arrival. the calves, weighing - kg, were kept in large boxes on straw bedding. milk replacer was offered via an automated feeder for the first two weeks after arrival. hay and concentrates were fed according to swedish recommendations during the study period of six weeks. group b consisted of calves from eight farms, different from those from which group a calves were selected. the age of the calves at the start of the study varied between and weeks (mean weeks, sd . ) and their weight ranged from to kg. the animals were weaned either before or on arrival at the division of ruminant medicine, department of clinical sciences, swedish university of agricultural sciences. the calves were housed in large boxes on straw bedding in a stable that had been cleaned and empty for more than four weeks before the experiment started. the animals were fed hay and concentrate according to swedish recommendations during the study period of six weeks. these animals were also part of another study, in which they were inoculated orally with lungworm (dictyocaulus viviparus) ( l ) on days and . however, the infection did not induce a patent infection with larvae present in faeces, probably because the larvae were not sufficiently viable. the study protocol was approved by the swedish national board for laboratory animals, uppsala, sweden. jugular blood samples were taken twice weekly from each calf, during the first four weeks following arrival and once weekly for the following two weeks. venoject tubes with edta and without additive (terumo europe n.v.) were used. clinical observations of the calves were performed at the same time, including recordings of any coughing and/or diarrhoea. the farm personnel decided which individuals were to be examined by the herd veterinarian for possible treatment during the observation period. the weight of the calves in both groups was recorded on arrival and again at the end of the study using heart girth measurement in group a, and a scale in group b. within a few hours of sampling, edta-blood was analysed for total and differential leukocyte counts and fibrinogen at the division of diagnostic imaging and clinical pathology, department of biomedical sciences and veterinary public health, swedish university of agricultural sciences. the leukocyte counts were performed using cell-dyn (abbot diagnostic division), and fibrinogen was analysed according to becker et al. ( ) using an automated analyser (konelab , konelab corporation). samples without additive were centrifuged ( g, min) and the serum was frozen at À °c until analysed for haptoglobin, saa, and viral antibodies. analysis of serum haptoglobin and saa was performed using the tridelta phase range haptoglobin assay and saa assay (tridelta development limited), respectively; the working range for the haptoglobin assay was . - g/l, and the intra-and inter-assay coefficients of variation were % at the concentration . g/l . the saa assay had been slightly modified by adding extra standard points. the working range was . - mg/l with samples diluted : . in this range, the intra-and inter-assay coefficients of variation were < %. serum antibody titres against bovine respiratory syncytial virus (brsv), bovine corona virus (bcv), parainfluenza virus type (piv- ), bovine virus diarrhoea virus (bvdv) and bovine adenovirus type (bav- ) were analysed on three occasions per group, i.e. days , and , using elisas (svanova biotech ab for brsv, bcv, piv- and bvdv, bio-x diagnostics for bav- ) and all analyses were performed according to the manufacturer's instructions. seroconversion was defined as a negative optical density value in the first or second sample, converting to a positive value in either the second or the third sample in paired sera. a total scoring of the results from the different apr parameters was calculated. where above normal values of saa, haptoglobin, fibrinogen or total leukocyte count occurred, one point was allocated for each. the sum of the points was calculated for each individual and sampling occasion as well as the mean (sd) value for each group. previously established (gånheim et al., ) basal app threshold levels differing between healthy and diseased animals were used, namely, . g/l for haptoglobin, . mg/l for saa, and . g/l for fibrinogen. a total leukocyte count > . · /l was considered above normal (jain, ) . leukocyte counts, app concentrations, numbers of sampling days with above normal app values and apr scores in groups a and b, were compared using student's t tests. the bonferroni correction was used to avoid mass significances. a p value < . was considered significant. in group a, . % and . % of the calves had signs of respiratory disease and diarrhoea, respectively, on at least one of the observation points (table ) . some calves had both respiratory symptoms and diarrhoea, but not on the same occasion. during the study period, three calves with respiratory disease were treated with antibiotics (benzylpenicillin procaine, ethacilin vet, intervet), based on evaluation of the farm veterinarian (table ) . in group b, all calves showed signs of respiratory disease on at least one of the sampling days, and % of the calves also had diarrhoea at the same time as respiratory symptoms (table ) . almost half of the calves were treated with antibiotics for respiratory disease as described for group a (table ). the proportions of calves in the two groups having respiratory symptoms or diarrhoea, at each sampling occasion, are given in fig. . the proportion was very low in group a, while a higher proportion of sick calves, especially with respiratory symptoms, was obvious in group b during the first two to three weeks after arrival. the average (sd) daily weight gain of the calves in group a was . ( . ) kg during the study period. in group b, the average (sd) daily weight gain for the calves ( . [ . ] kg) was significantly lower. none of the calves in group a seroconverted against brsv, bcv, bvdv, or piv- , during the study period, but % of the animals seroconverted against bav- (table ) . all calves had elevated antibody titres against one or several viruses at the beginning of the study. table numbers (%) of calves with respiratory symptoms (resp), or diarrhoea (diar), or being treated for respiratory disease (treat), or seroconverting against brsv, bcv, bvdv, piv- , bav- , in two groups a (n = ) and b (n = ) of calves over a period of six weeks after housing in group b, one-quarter of the calves had seroconverted against bcv between the first and second sampling, while the remainder of the calves already had high antibody titres against bcv at the start of the study (table ) . a majority of the calves also seroconverted against bav- , while the rest of the group had such antibodies from the beginning of the study (table ) . there were no seroconversions against brsv, bvdv and piv- during the study, but most calves had antibodies against one or several of these viruses on arrival. the total leukocyte counts, and the numbers of neutrophils, lymphocytes and eosinophils for groups a and b are given in fig. . the mean number of monocytes was constantly close to . · /l in both groups throughout the study (data not shown). the proportions of animals with above normal total leukocyte counts were low in both groups (fig. ) . no significant differences between the groups were observed in total leukocyte counts, or in numbers of lymphocytes and neutrophils. however, the eosinophil numbers were significantly higher in group b than in group a on days , and . out of calves in group a, ( %), ( %) and ( %) calves had above normal values of saa, haptoglobin and fibrinogen, respectively, at least once during the study period. in group b, all calves had above normal values of all three app on at least one occasion. the proportions of calves having above normal concentrations of saa, haptoglobin and fibrinogen were numerically higher in group b than in group a at most sampling occasions (fig. ) . in group a, the mean maximum concentration per calf was . g/l (sd . , range . - . g/l), . mg/l (sd . , range . - . mg/l) and . g/l (sd . , range . - . g/l) for haptoglobin, saa and fibrinogen, respectively. in group b, the mean maximum concentrations of haptoglobin ( . g/l [sd . , range . - . g/l]), and fibrinogen ( . g/l [sd . ], range . - . g/l) were significantly higher than in group a. the mean maximum concentration of saa ( . mg/l [sd . ], range . - . mg/l) was numerically higher, but not significantly different, from in group a. the mean numbers of sampling days per calf with above normal saa, haptoglobin and fibrinogen concentrations in groups a and b are presented in table . the numbers were significantly higher in group b than in group a for all three app. the mean apr scores for the two groups are given in fig. . the scores were significantly higher in group b than in group a on days and after arrival. a large proportion of calves in both groups had elevated concentrations of one or more apps at one or several occasions during the study. this is consistent with the expectation that many animals would experience infection after transportation and mixing with animals from other farms. some also had elevated antibody levels against the viruses of interest. the disease and treatment incidences were considerably higher in group b than in group a, as was the proportion of animals with above normal values of apps, and the apr score. however, also in group a, where the health status was considered to be good, many calves had elevated app-values, indicating sub-clinical disease. the mean number of sampling days per calf with above normal app-values was well correlated with the clinical findings in the groups. the difference between the groups was most obvious for haptoglobin, indicating that this may be the most useful predictive app and is in agreement with carter et al. ( ) who concluded that analysis of serum haptoglobin was a better tool for discrimination between calves that became ill and those that did not, compared to other apps. saa is reported to be more sensitive to stimulation (horadagoda et al., ; heegaard et al., ) , and as an increase can be induced also by other factors than disease, such as stress (alsemgeest et al., ) it may be less suitable as an indicator of health problems. the treatment incidence also differed between the two groups. in group b, almost half of the calves were treated with antibiotics, while only / calves was treated in group a. the difference may not only reflect the health status in the groups, but could also be due to different evaluations by the persons handling the animals. however, no deaths occurred in group a, suggesting that all animals in need of treatment received it. the mean weight gain differed significantly between the groups, with the healthier group a having the greatest weight gain. different methods were used in the groups for evaluating weight, indicating these results must be interpreted with care. however, heinrich et al. ( ) demonstrated that, among the different body measures, heart girth is the best estimate for body weight, and sørensen and foldager ( ) and andersson ( ) showed a good correlation between the heart girth and body weight. the total leukocyte counts exceeded normal values for calves of this age (jain, ) only on a few occasions and in a few individuals. the results did not differ between groups indicating that an above normal leukocyte count alone is not a good indicator of disease. the significantly higher number of eosinophils in group b was probably due to the experimental inoculation with lungworm, as eosinophilia is a rather consistent finding during lungworm infection (radostits, ) . this result was table mean (sd) number of sampling days per calf with levels of serum haptoglobin, serum amyloid a (saa) and plasma fibrinogen values above basal levels a in two groups (a and b) of calves sampled on ten occasions during the six week period after housing surprising as none of the animals had a patent infection with the presence of larvae in faeces. however, the eosinophilia indicates that the immune system of the calves had responded to the parasites. thus, the lungworm infection may be one factor responsible for the elevated app levels observed in group b, as lungworm infection can induce production of app (gånheim et al., ) . the seroconversion against bcv and bav- in group b indicates that these infections were probably the main reason for the disease symptoms in this group. it is likely that some calves brought the infection from the farm of origin and spread it in the new group. the four calves that seroconverted against bcv reacted with increases in haptoglobin and fibrinogen and three of them also reacted with an increase in saa. however, as indicated above, the lungworm infection may also have affected the clinical outcome; mixed infections are likely to result in more severe symptoms as was shown when combining viral and bacterial infections (gånheim et al., ) . there may also have been other infectious agents circulating in the group, such as other serotypes of adenovirus or secondary bacterial infections, e.g., with m. haemolytica, or other opportunistic invaders, but other viral or bacteriological examinations were not performed in this study. the apr score, combining the results of the three apps and the total leukocyte counts, was created to give a better overview of the reactions in the groups. other researchers have reported that measurement of a single app is not reliable for health evaluation and recommend a combined analysis of several parameters (toussaint et al., ; young et al., ) . in our study, differences in clinical health status between groups a and b, were consistent with the differences in the apr score. however, the results need to be interpreted with care, as the experimental infection with lungworm may have been partly responsible for the app elevation. the results of the present study support the usefulness of app measurement in monitoring animal health, and supports the proposals made by other authors (alsemgeest et al., ; murata et al., ; petersen et al., ) . at present, routine use of app analyses at the herd level is not realistic because of the high costs involved. however, pooling of serum from several animals in a herd may offer a way to minimize the costs. if less expensive analytical methods become available, the calculation of mean apr scores could be a useful tool for the evaluation of herd health. however, to provide the best evaluation of health status in a herd, or in a group of animals, frequent samplings during the whole rearing period should be made and again this is probably not a realistic approach as it would be too expensive. sampling during the first weeks after arrival in the herd would give an indication of the pressure of infections and/or stress during transportation and mixing and so may not give a fair picture of the management conditions of the actual farm. in our opinion, samplings at, for example, monthly intervals during the middle and later stages of the rearing period may be more relevant if the main interest is to evaluate the animal health conditions of a specific farm. the establishment of suitable app or apr score thresholds for herd health would require analysis of a much larger number of herds. the diagnostic value of acute phase proteins in bovine clinical chemistry. in: blood concentrations of acute-phase proteins in cattle as markers for disease influence of physical stress on the plasma concentration of serum amyloid a (saa) and haptoglobin (hp) in calves dairy calf pneumonia. the veterinary clinics of north america rekryteringskvigor fö r mjö lkproduktionuppfö dningsmodellens betydelse fö r produktionen effect of transportation and co-mingling on the acute phase protein response, growth, and feed intake of newly weaned beef calves respiratory infections -project, panorama and treatment strategies a functional photometric assay for plasma fibrinogen relationship of vitamin e supplementation and antimicrobial treatment with acute-phase protein responses in cattle affected by naturally acquired respiratory tract disease the bovine respiratory disease complex: a complex interaction of host, environmental, and infectious factors bovine and canine acute phase proteins serum haptoglobin as an indicator of the acute phase response in bovine respiratory disease diagnostic significance of the major acute phase proteins in veterinary clinical chemistry: a review the acute phase response in calves experimentally infected with bovine viral diarrhoea virus and/or mannheimia haemolytica the acute phase response of haptoglobin and serum amyloid a (saa) in cattle undergoing experimental infection with bovine respiratory syncytial virus predicting body weight and height in holstein heifers using body measurements acute phase proteins in cattle: discrimination between acute and chronic inflammation hematology of cattle using serum amyloid a to screen dairy cows for sub-clinical inflammation plasma fibrinogen levels in normal and sick cows current research on acute phase proteins in veterinary diagnosis: an overview application of acute phase proteins in veterinary clinical chemistry lungworm infestation in cattle (bovine verminous bronchitis) development of a simple enzyme immunoassay for blood haptoglobin concentration in cattle and its application in improving food safety effect of breed and plane nutrition on the estimation of live weight by heart girth in dual purpose heifers implication of clinical pathology in assessment of animal health and in animal production and meat inspection experimental reproduction of winter dysentery in lactating cows using bcv -comparison with bcv infection in milk-fed calves serum haptoglobin concentrations in a population of feedlot cattle the authors thank the swedish council for forestry and agricultural research for financial support and the staff at the clinic of the division of ruminant medicine and epidemiology for looking after the calves well. key: cord- -vnyv oj authors: chen, yalan; yu, yijun; lin, xin; han, zhenwei; feng, zhe; hua, xinyi; chen, dongliang; xu, xiaotao; zhang, yuanpeng; wang, guheng title: intelligent rehabilitation assistance tools for distal radius fracture: a systematic review based on literatures and mobile application stores date: - - journal: comput math methods med doi: . / / sha: doc_id: cord_uid: vnyv oj objective: to systematically analyze the existing intelligent rehabilitation mobile applications (apps) related to distal radius fracture (drf) and evaluate their features and characteristics, so as to help doctors and patients to make evidence-based choice for appropriate intelligent-assisted rehabilitation. methods: literatures which in regard to the intelligent rehabilitation tools of drf were systematic retrieved from the pubmed, the cochrane library, wan fang, and vip data. the effective apps were systematically screened out through the app markets of ios and android mobile platform, and the functional characteristics of different apps were evaluated and analyzed. results: a total of literatures and apps were included, which were divided into four categories: intelligent intervention, angle measurement, intelligent monitoring, and auxiliary rehabilitation games. these apps provide support for the patients' home rehabilitation guidance and training and make up for the high cost and space limitations of traditional rehabilitation methods. the intelligent intervention category has the largest download ratio in the app market. angle measurement tools help drf patients to measure the joint angle autonomously to judge the degree of rehabilitation, which is the most concentrated type of literature research. some of the apps and tools have obtained good clinical verification. however, due to the restrictions of cost, geographic authority, and applicable population, a large number of apps still lack effective evidence to support popularization. conclusion: patients with drf could draw support from different kinds of apps in order to fulfill personal need and promote self-management. intelligent rehabilitation apps play a positive role in the rehabilitation of patients, but the acceptance of the utilization for intelligent rehabilitation apps is relatively low, which might need follow-up research to address the conundrum. distal radius fracture (drf) is one of the common upper limb fractures, which could be accounted for about . % of upper limb fractures [ ] . it is more common in elderly women; this may be due to the effects of osteoporosis and eventually lead to falls. due to the influence of osteoporosis and the increase of age, slight external force could often result in complex comminuted fracture, and with the possible injury of intercarpal ligament for postmenopausal women. majority of the patients with this type of fracture are mainly middle-aged and elderly women, most of whom still have a misunderstanding that they need to rest and cannot exercise during the recovery phase. exercise intervention has been proven to be one of the most effective ways to improve health and promote recovery [ ] . appropriate rehabilitation exercise can accelerate local blood circulation; accelerate the repair of surrounding soft tissues; prevent muscle atrophy, joint cavity adhesion, and stiffness; and reduce the occurrence of complications such as dysfunction of joints near the fracture site [ ] . elderly patients with fractures have poor compliance with postoperative rehabilitation training. in addition, in the short-term hospitalization period, it is difficult for patients to master the comprehensive knowledge and skills of rehabilitation training and to guarantee the progress and quality of independent rehabilitation training after discharge. moreover, some traditional rehabilitation methods require patients to conduct at a fixed time and place, and some of the method is relatively boring. too much repetitive exercise makes it difficult for patients to persist for a long time and lack motivation. mhealth is a medical and public health practice supported by mobile devices that fills the above gap and promotes unprecedented opportunities for professional clinical diagnosis and treatment recommendations [ , ] . as a software application accompanying mhealth, commonly known as "application" (app), it has become one of the main ways for patients to participate in health and rehabilitation management. it can realize home rehabilitation training for patients, self-diagnose the rehabilitation progress, and obtain professional rehabilitation guidance, which greatly promotes the development of participatory medicine [ , ] . for example, the serious game app could be regarded as effective mean for intervention in home exercise projects [ ] , while the protractor software could measure the angle of arm rotation and abduction for patients and evaluate the degree of recovery by collecting each measurement data. at present, the practical utilization of intelligent rehabilitation assistance tools related to drf is not very extensive. in many cases, patients find it difficult to choose appropriate tools according to the rehabilitation needs when they are faced with a variety of apps on different mobile platforms. therefore, we summarize the up-to-date apps related to drf in both literatures and different mobile platforms. through evidence-based retrieval and data extraction, we systematically analyze the functions, characteristics, and limitations of these apps, to provide evidence-based references for clinical and patients. the intelligent rehabilitation tools on the app market are mainly divided into several categories such as health education and functional exercise. "distal radius fracture", "rehabilitation", "functional exercise" or "health education" were searched in chinese and english separately in app store, google player, and huawei platform. criteria. inclusion criteria include the following: ( ) the study about the intelligent rehabilitation apps or tools for the drf patient, especially intervention research; ( ) apps or tools that assist the intelligent rehabilitation for the drf patient on any platform; ( ) rehabilitation is the main function or included function of the apps or tools; and ( ) the app was applicable from intelligent mobile terminals. exclusion criteria include the following: ( ) apps or tools limited to the operation and use of medical staff, ( ) apps with no score or score less than and download volume less than , ( ) apps or tools for rehabilitation of other fracture types, and ( ) apps or tools without detailed description and usage records. data from the included studies were extracted by the two investigators (j.y. and d.c.) using standardized and piloted design formats. the preset extraction indicators mainly include basic information such as the name of the app, the platform used, and the function. the literature research also needs to extract the clinical research methods and results, and the platform information needs to include user usage, downloads, and ratings. discrepancies in the process of study selection and data extraction were resolved through a group discussion with two other authors (y.c. and g.w.). the extraction of the intelligent rehabilitation apps involved in the literature, classified them according to the functions and types of different apps, and analyzed the advantages and disadvantages. the flow of the search strategy is shown in figure . a total of articles were searched in the four databases. subsequently, non-human research articles were screened out through sifting and screening. the remaining articles were analyzed by abstract and title, then for full-text analysis, and articles were finally included (table ) . ninety-seven intelligent rehabilitation apps were sifted through according to the description by the publisher and the rating. finally, apps related to the rehabilitation of drf were included (supplementary table ). it can be seen from the result that ( %) smart apps are applicable to both ios and android, with ( %) smart apps are suitable for ios system, and ( %) smart apps are suitable for android system (figure (a) ). most of the apps are free. apps. the number of app downloads from the app market computational and mathematical methods in medicine is concisely described in figure (b). the top three apps with more than , downloads are the intelligent intervention app "fisioterapia a tu alcance" and two intelligent monitoring apps "oasis healthcare (honghua medical)" and "gold nurse (gold nurse)." the three software not only have the highest downloads but also got a user score higher than points, which was highly praised by users. apps. intelligent rehabilitation apps were divided into five categories according to different functions: intelligent intervention, angle measurement, intelligent monitoring, and rehabilitation game [ ] (figure (c)). intelligent intervention software was accounted for the most of the results retrieved in the app market. intelligent intervention is mainly used in the later stage of fracture, which is a critical period for functional exercise after reaching the clinical healing standard. such apps would be able to provide patients with functional exercise and develop personalized rehabilitation services by combining physiotherapy (pt) and home exercise programs (heps). the second proportion of the results retrieved in the app market is the angle measurement app which holds the largest proportion among selected literatures regarding intelligent rehabilitation apps. the degree of rehabilitation of drf patients is generally diagnosed by the angle of extension, flexion, pronation, and supination of the wrist and forearm. the angle measurement tool allows the patient to measure and record the rehabilitation angle without leaving home and can make independent judgments or share the data with the doctor in real time so that the doctor can understand the degree of the patient's rehabilitation. the ulnar deviation measurement of the two apps is demonstrated in figure . there are not many rehabilitation games related to fractures in the app store and literature studies. the way of games is conducive to divert the patient's attention, in the process of playing games to enhance wrist finger movement and joint training and reduce the discomfort in the training process. the intelligent monitoring app involves more complicated processes and links, providing patients with an the wrist flexion, extension, radial deviation, and ulnar deviation joint activity were measured by the universal goniometer and g-pro app in iphone , respectively. the g-pro© app has good to excellent reliability (icc ≥ : ), and it is effective at the same time as the general goniometer (r ≥ : ) used to measure wrist rom. patients recovering from wrist injury and professional wrist injury therapists completed a full-level game, respectively, and the experimenter finally scored the software, respectively. revalidate! is a beneficial and interesting experience for the rehabilitation of wrist fractures. it is tailored to patients and provides functional measurements for patients, which may become a very useful exercise tool in future rehabilitation. tilt maze、labyrinth [ ] ios android na na na skilled hand therapy can help recommend appropriate postures and encourage participation in therapeutic games, address specific client deficiencies, and reduce functional problems. rehabilitation assistant [ ] ios android randomized controlled trial ; the control group was treated with preintervention rehabilitation assistant app, while the observation group was treated with postintervention rehabilitation assistant app, and the patient satisfaction was compared the health education effect and health education satisfaction of the rehabilitation assistant app were higher than those before the intervention (p < : or p < : ). computational and mathematical methods in medicine emerging medical service model (figure ). it can combine online consultation with offline care to provide remote services for patients, ease the gathering and intersection of outpatient populations, promote closer communication between doctors and patients, and improve rehabilitation effects. the popularization of mhealth technology has gradually brought a new experience and service mode to the public, which would help to address the disadvantages of the existing medical system [ , ] . this research combines scientific research with the practical app market, comprehensively analyzes the current status of the utilization of drf's intelligent rehabilitation app, and discusses the characteristics of different types of rehabilitation apps. from the research results of the app market, the most involved are intelligent intervention apps, the average downloads are over . the most popular app is the intelligent monitoring app, which scores more than points. in contrast, the most involved in literatures are the angle measurement apps, followed by rehabilitation game apps. this may indicate that angular measurement and rehabilitation game apps have more research value; meanwhile, intelligent intervention and intelligent monitoring are more valuable regarding commercial and economic areas. the app involved in some literatures does not yet exist in the app market, which may indicate that some apps are still in the development stage. however, most of the apps in the app market lack effective evidence and real data support, and the actual medical value needs further verification. studies have shown that the prognosis of patient with drf would benefit from the combination of supervised pt and heps [ ] . intelligent intervention apps realize the combination of the two methods. on one hand, the real-time transmission of smart phone devices realizes the connection between patients and medical service providers, which is conducive to the development of personalized guidance programs [ ] . meanwhile, these apps also provide professional rehabilitation exercise guidance which promotes correct functional exercise [ ] . on the other hand, intelligence software runs in the background to supervise computational and mathematical methods in medicine patients to complete corresponding functional exercises and provide users with timely feedback through charts and other forms of information to improve patients' emotional and psychological conditions in time [ ] . most of the patients with drf would self-diagnose their condition and recovery process by judging proprioceptive recovery such as grip strength, pinch force, and wrist movement. few patients would determine their rehabilitation process by measuring the angle of wrist movement [ , ] . nevertheless, the current understanding and knowledge in regard to wrist proprioception is still not enough [ ] [ ] [ ] . distal fractures of the radius might often involve with the wrist [ , ] , which would limit the range of movement (rom) between the forearm and wrist in multiple motor planes, including wrist flexion and extension, carpal ulnar deflection, and forearm pronation [ , ] . hence, the rom would be more accurate when determining the recovery process of drf [ ] . the angle measurement involved in this article mainly measures the range of motion and angle in two ways. for example, the measurement of ulnar deviation, one is to upload the angle photo or video of the part to be measured to the app and then measure based on the image data; the other is to fix the mobile device on the part to be measured on the wrist or forearm and measure the movement amplitude (angle) in real time (figure ) . the accuracy and precision of these two methods are affected by the accuracy of the method used by the patients. patients with drf usually experience symptoms such as limited wrist movement [ ] , pain, weakness, and even serious complications (nonunion [ ] and malunion [ ] ). professional rehabilitation care is beneficial for pain management and reduction of the occurrence of adverse complications [ ] . intelligent monitoring app helps professional medical staff and patients to establish harmonious relationship [ ] , provide timely and professional medical services, and reduce the occurrence of adverse complications. moreover, it can provide remote consultation and home care. compared with the traditional medical model, it can greatly save patients' waiting time and improve patient satisfaction and rehabilitation flexibility. the advent of the internet era would allow non-textbased interactive information, such as video and images, to computational and mathematical methods in medicine become more acceptable [ ] . rehabilitation games are interactive apps that are based on the internet and visual stimulation which could also combine with wearable sensors. for example, "revalidate!" [ ] is an app that needs to be used in conjunction with wearable devices on the proximal end of the wrist and forearm. by monitoring the parameters of the patient's wrist or upper limb motion range [ ] , it can safely and effectively help the drf patient's wrist rehabilitation [ ] . however, the use of these apps is affected by the patient's cognition and acceptance and requires the guidance of professionals. at the same time, due to cost and geographic restrictions, the promotion of some apps will be restricted by authority. opportunity and challenge. mhealth care can provide personalized precision treatment and better peoplecentered care [ , ] . this home-based rehabilitation model can better meet the needs of patients' autonomous rehabilitation management. at the same time, it can greatly improve the flexibility of rehabilitation management, reduce the medical burden, and better respond to special events such as covid- [ ] . the market prospect of developing intelligent auxiliary rehabilitation tools from the perspective of patients is very broad. this study provides reference for patients with different needs to choose appropriate intelligent auxiliary rehabilitation tools: the services provided by intelligent intervention and monitoring apps are more humanized with easier instruction to operate, which is suitable for middle-aged and elderly patients who cannot master app skillfully; the rehabilitation gaming apps are more interesting, but the operation and use process is relatively complicated, which are more suitable for younger patients; and the angle measurement apps require relevant rehabilitation theory knowledge and are more suitable for patients to utilize under the guidance of medical staff or with the assistance of family members. meanwhile, through this research, detailed information in regard to app instruction and operation modes of different types of rehabilitation software was obtained, which provides a basis for subsequent clinical promotion and utilization. with the continuous deepening of intelligent medical reform, artificial intelligence algorithms have been widely used in auxiliary diagnostic methods of clinical medicine [ , ] . app has gradually become a new trend in medical diagnosis, treatment, prevention, and management. however, the clinical application of intelligent app is still relatively immature, and the practicability and reliability need further verification. for example, in the collection of medical images, noise, image ambiguity, and complex clustering of multiview data [ ] [ ] [ ] [ ] ; permissions and pertinence and limitations of the crowd [ ] ; and cumbersome operation steps may lead to misoperation and privacy protection of information and data [ ] , etc. this study is conducive for medical staff to know the utilization of drf-related apps, understand high-quality intelligent rehabilitation apps, and encourage the practical utilization inside and outside the hospital, so as to promote the expected effect of intelligentassisted rehabilitation tools in clinical and patient home rehabilitation management. with the development of the mhealth medical model, intelligent rehabilitation apps and tools are gradually being used in clinical and patient independent health management. through systematic evidence-based analysis based on literature and app platforms, this study integrates different types of intelligent rehabilitation apps that are suitable for drf rehabilitation and explores its effects on the rehabilitation of such fractures from different perspectives. although the evidence is limited, it can still be clearly shown that app-based rehabilitation intervention, angle measurement, and monitoring management can all improve the effect of rehabilitation training and actively promote patients' self-rehabilitation management. this also enables clinical medical staff and drf patients to make evidence-based choice according to the different characteristics and needs of apps, which could meet the individual needs and improve the effectiveness of selfrehabilitation management. however, due to the restrictions of cost, geographic authority, and applicable population, a large number of apps still lack effective evidence to support popularization. therefore, in future research, large-scale user-centered clinical trials that will be added to evaluate the effectiveness and practicality of apps are particularly important. the data used to support the findings of this study are available from the corresponding author upon request. the authors declare no competing financial interests. the authors' responsibilities were as follows: g.w., y.c., and y.z. designed the research; y.c. and y.y. designed the research; y.c., y.y., and g.w. designed the pilot data extraction table; y.y., z.h., z.f., and x.h. performed literature search and study selection; y.y., d.c., and x.x. conducted data extraction; y.y., y.c., x.l., and g.w. drafted the manuscript. all the authors completely consented with all the data in the study, critically revised the manuscript, and approved the final version. clinical data analysis of incidence and occurrence characteristics of upper limb fracture hand rehabilitation after distal radius fracture rehabilitation strategies for wrist sensorimotor control impairment: from theory to practice mhealth in cardiovascular health care use of mobile health (mhealth) technologies and interventions among community health workers globally: a scoping review the potential of mhealth applications in improving resistant hypertension self-assessment, treatment and control functions of mhealth applications: a user's perspective systematic review on the effects of serious games and wearable technology used in rehabilitation of patients with traumatic bone and soft tissue injuries reliability and concurrent validity of a new iphone(®) goniometric application for measuring active wrist range of motion: a cross-sectional study in asymptomatic subjects self-measured wrist range of motion by wrist-injured and wrist-healthy study participants using a built-in iphone feature as compared with a universal goniometer utility of the iphone gyroscope application in the measurement of wrist motion validity and reliability of goniometric mobile apps: device, application, and examiner factors the validity and reliability of drgoniometer, a smartphone application, for measuring forearm supination face validity and content validity of a game for distal radius fracture rehabilitation using smartphone applications as hand therapy interventions nursing management and effect evaluation of orthopedic rehabilitation assistant app systematic evaluation and research on intelligent assistant rehabilitation tools after fracture surgery a mobile revolution for healthcare? setting the agenda for bioethics interactive telemedicine: effects on professional practice and health care outcomes supervised physical therapy vs home exercise program for patients with distal radius fracture: a single-blind randomized clinical study specialized smartphone intervention apps: review of to nih funded grants action observation in the modification of postural sway and gait: theory and use in rehabilitation capture, transfer, and feedback of patient-centered outcomes data in palliative care populations: does it make a difference? a systematic review the patient-reported outcomes measurement information system (promis) developed and tested its first wave of adult self-reported health outcome item banks contribution of functional parameters to patient-rated outcomes after surgical treatment of distal radius fractures postoperative treatment of distal radius fractures using sensorimotor rehabilitation proprioceptive control of wrist extensor motor units in humans: dependence on handedness proprioception of the wrist joint: a review of current concepts and possible implications on the rehabilitation of the wrist fractures of the distal radius. changing therapeutic strategies notes on fracture incidences as clinical evidence of disturbed bone metabolism with increasing age association of wrist and forearm range of motion measures with self-reported functional scores amongst patients with distal radius fractures: a longitudinal study analysis of the factors affecting the angle and function of wrist joint after distal radius fracture the effect of fracture-related factors on the functional outcome at year in distal radius fractures complications of distal radius fixation mal-union and dysfunction in colles' fracture a guide to improving the care of patients with fragility fractures use of "serious health games" in health care: a review mobile health technologies in cardiopulmonary disease the value of mhealth for managing chronic conditions features and functionalities of smartphone apps related to covid- : systematic search in app stores and content analysis a novel distributed multitask fuzzy clustering algorithm for automatic mr brain image segmentation deep takagi-sugeno-kang fuzzy classifier with shared linguistic fuzzy rules collaborative fuzzy clustering from multiple weighted views fast exemplarbased clustering by gravity enrichment between data objects a novel negative-transferresistant fuzzy clustering model with a shared cross-domain transfer latent space and its application to brain ct image segmentation seizure classification from eeg signals using transfer learning, semi-supervised learning and tsk fuzzy system scaling up mhealth: where is the evidence? mhealth and mobile medical apps: a framework to assess risk and promote safer use this work was supported by the science and technology projects of nantong city (grant number msz ); the natural science research project of xinglin college, nantong university (grant number k ); the college the detailed information of apps related to the rehabilitation of distal radius fracture was contained in supplementary table , including the operation platform, user rating, download volume, payment method, and specific features description. (supplementary materials) key: cord- -vofuo g authors: omae, yuto; toyotani, jun; hara, kazuyuki; gon, yasuhiro; takahashi, hirotaka title: effectiveness of the covid- contact-confirming application (cocoa) based on a multi agent simulation date: - - journal: nan doi: nan sha: doc_id: cord_uid: vofuo g as of aug. , coronavirus disease (covid- ) is still spreading in the world. in japan, the ministry of health, labor, and welfare developed"covid- contact-confirming application (cocoa),"which was released on jun. , . by utilizing cocoa, users can know whether or not they had contact with infected persons. if those who had contact with infectors keep staying at home, they may not infect those outside. however, effectiveness decreasing the number of infectors depending on the app's various usage parameters is not clear. if it is clear, we could set the objective value of the app's usage parameters (e.g., the usage rate of the total populations) and call for installation of the app. therefore, we develop a multi-agent simulator that can express covid- spreading and usage of the apps, such as cocoa. in this study, we describe the simulator and the effectiveness of the app in various scenarios. the result obtained in this study supports those of previously conducted studies. as of aug. , coronavirus disease infection is still spreading in the world. to overcome the spread of covid- , the japanese government gave stay at home order to its citizens from apr. , to may. , [ ] . consequently, the number of total infectors significantly decreased (the number of daily average infectors one week ago before the order was about persons; moreover, the number of daily average infector one week later after issuing the order was about persons [ ] .). besides, as measures to overcome the spreading of covid- , the japanese government is distributing cloth masks, as well as supporting the sterilization of medical institutions and pcr testing [ ] . in addition, the japanese government developed "covid- contact-confirming application (co-coa)." cocoa is an application that can be installed into a smartphone, and by utilizing it, users can know whether they had contact with infectors of covid- . hereinafter, this is called an app. the diagram of the app is shown in fig. . the upside of the figure is used to verify whether the user had contact with the infectors of covid- . if the user is infected with covid- , he/she registers via the button shown in the downside of the figure. if infectors register, other users who are not infected can know whether or not to have contact with infectors for at least two weeks or less via the app. to obtain this information, one is required to install the app and turn on the bluetooth of his/her smartphone. moreover, it is necessary that infectors and other persons approaching within meter and over minutes [ ] . the app's privacy policy and specifications are available on the japanese government website [ ] . the japanese government announced that there have been cases where covid- symptoms do not appear [ ] . roth et al. [ ] reported that even if the infectors of covid- are in their incubation period, it is possible that they can infect others. consequently, it is possible that infectors, who do not show symptoms, can infect many persons. in general, users can know the personal information of infectors via the app. if infectors continue to stay at home, he/she could avoid infecting others. therefore, the following points are important to overcome the spreading of covid- by utilizing the app: ( ) use the app; ( ) users who know about their contact with infectors via the app stay at home, ( ) infectors register that they are infected via the app. moreover, other countries y. omae (e.g., china, india, israel, and so on) also have developed similar apps [ ] . however, to overcome the spreading of covid- , the target values of the described points are unclear. in the real world, we cannot strictly verify the effectiveness of the app because there are various mixed measures for overcoming the spread of covid- . moreover, since human life and death are involved, we cannot conduct a control experiment in the real world owing to infection spread and convergence. thus, we had better conduct experiments in an artificial world, and one of the approaches is applying multi-agent simulation (mas). there are numerous previously conducted studies based on virus-spreading simulations [ ] [ ] [ ] [ ] . however, these studies do not target covid- , and we cannot verify the app's effectiveness. very recently, simulations targeting how to overcome the spread of covid- spreading have been conducted in the year . for example, by applying the susceptible-exposed-infectious-recovered (seir) model to study covid- , hou et al. [ ] observed that a measure of decreasing contact with persons can effectively decrease the total number of infectors at peak time. chatterjee et al. [ ] also developed a seir model for covid- and conducted a simulation experiment using india as a case study. as a result, these authors reported that the measures of avoiding contact with persons, such as lockdown, can significantly reduce the spread of covid- . these studies are beneficial. besides, the described seir model-based covid- simulations [ , ] do not involve the app's effect. to survey the app's effectiveness using england as a case study, hinch et al. [ ] conducted simulations based on a mixed method of sir model (it is not a seir model) and agent-approach simulation. they concluded that the usage of the app by % of the total population can lead to the convergence of covid- . although this is an innovative result, it is difficult to verify its reliability in the case of social simulations. in general, the reliability of simulations is measured by calculating the difference between the data collected in real world and those generated from a simulation model. in addition, takahashi [ ] pointed out that this method is not always possible in the case of social simulation. in the case of the covid- pandemic situation, it is difficult to collect data to calculate reliability in the real world. therefore, we cannot measure the reliability of covid- spreading and the app's effectiveness by calculating the difference between data of the real world and simulator. as an alternative method, takahashi [ ] recommended the comparison of results generated from different simulation models. if different simulation models generated similar results, then we can justify that the simulation result is reliable. further, covid- and the app, such as cocoa, are the latest research topics of the year (cocoa developed by the japanese government was released on jun. , ) . therefore, simulators for verifying the effectiveness of the apps, such as cocoa, are very few as of aug. . thus, applying a measuring method of reliability based on the comparison of differ- ent simulation models is difficult. in conclusion, we observe that now is the time to report numerous simulation cases based on the effectiveness of the app for the convergence of covid- for different methods. in the future, the accumulation of different case studies by various researchers can enable the study of the measurement of simulation reliability. therefore, by improving a simulator developed by omae et al. [ ] , we introduce a multi-agent simulator that can express covid- spreading and the usage of the app, such as cocoa. to overcome the spreading of covid- , we employ the following three points: ( ) install and use the app; ( ) users who have contact with infectors via the app stay at home; ( ) infectors register that they have been infected via the app. furthermore, we include three parameters for the following expression: ( ') the usage rate of the app; ( ') decreasing value of going out probability of persons who have contact with infectors via the app; ( ') infection registration rate of infectors through the app. in this study, we introduce the details of the developed mas and the app's effectiveness of reducing the number of infectors of covid- . in this section, we describe the multi-agent simulator for the spreading of the virus infectors developed by omae et al. [ ] . moreover, the new parameters for expressing the app embedded in the simulator are described. the seir model is one of the methods for simulating the spread of the virus. there are four states (s, e, i, r) in the model [ ] . furthermore, susceptible person (state: s) has a possibility of being an infector by having contact with other infectors. exposed person (state: e) means an incubation period. infectious person (state: i) means an infector of a virus. recovered person (state: r) means the person who has recovered and acquired immunity. in the case of seir model, there have been some simulations of covid- [ , , [ ] [ ] [ ] . thus, we apply the seir model for the simulations of covid- spreading. we note that the basic seir model is in the form of an ordinary differential equation. it can be fast simulated, but including various parameters, such as the expression of cocoa, agents' job, lifestyle, and so on, is difficult. moreover, a mixed model of the seir model and mas has been proposed [ , ] . this is the model of a virus that stochastically spread through the contact of agents. in this study, we use a mixed model of the seir model and mas. an infection transition model is depicted in fig. . the arrows denote a possibility of states' transition and give transition probabilities. if no arrow exists, states' transition cannot occur. additionally, states' transitions are from s to e, from e to i, and from i to r or d. in the basic seir model, state d does not exist. state d means death. the mas developed by omae et al. [ ] included state d because they conducted a simulation to verify the relationship between the number of dead persons and the capacity limit of isolation wards. next, we explain transition probabilities. the stochastic variable for the transition of s, e, i, r, d is defined by p(x t+ |x t ,c, t, h). if c, t, h are omitted, then the transition probability has the same value whatever value we assigned to the variable. here, x t and x t+ denote the infection state of an agent at time t and t + , and they are defined by moreover, c denotes the variable for expressing contact with agents of state i (infectors); c = denotes notcontact; c = means contact with agents of state i (infectors). t denotes the number of days elapsed from changing to other states (a unit is [days]). h denotes the variable for expressing hospitalization of agents of state i (h = : nonhospitalization; h = : hospitalization). in addition, eq.( )-based transition implies that transition probabilities depend on c, t , and h. the strict meaning of contact with an agent and infectors is that they are in contact within euclidean distance for about minutes in an artificial society. first, the transition probability from state s to state s is defined by where β denotes an infection probability of about minutes (the minimum unit of time of the simulator is minutes [ ] ). in the case in which there is no contact with a target agent and infectors, the transition probability is zero. further, in the case of contact with them, the transition probability is decreased. the transition probability from state s to state e is defined as follows: if a target agent had contact with infectors (c = ), the transition probability from state s to state e is β . since state s only changes to state s or e, we have that is satisfied for all conditions of c, t, h. the transition probability from state e to state e is defined as follows: while the transition probability from state e to state i is defined as follows: where is satisfied for all conditions of c, t, h. the transition probability from state i to state i is defined by the transition probability from state i to state r is defined by in addition, the transition probability from state i to state d is defined by where t i→rd denotes the period of required transition time [day] (infection period of virus). if t , which is the number of days elapsed from changing to state i, reaches t i→rd , then the state of an agent changes from state i to state r or d. in the case in which other conditions are present, the target agent maintains state i. additionally, γ represents the fatality rate in the case of nonhospitalization (h = ), while γ represents the fatality rate in the case of hospitalization (h = ). further, the fatality rate depends on the hospitalization of agents. in general, because the fatality rate in the case of nonhospitalization has higher value more than that of hospitalization, we recommend γ > γ ( ) [ ] variables parameters values e max simulation period days e the number of houses numbers e the number of initial infectors persons e locations of companies xy e locations of shops xy e locations of schools xy e capacity limitation of isolation wards beds e basic going out probability prob. e going out time time e stay time of facility time e probability of going to a hospital prob. e decreasing value of e during state i prob. β infection probability prob. γ fatality rate (nonhospitalization) prob. γ fatality rate (hospitalization) prob. t e→i incubation periods (from e to i) days t i→rd infection periods (from i to r or d) days as the fatality rates. since state i only changes to state i, r, or d, the following equation is satisfied for all conditions of c, t, h. the state r denotes that the agents acquired immunity, and the state d means death. therefore, the following equations are satisfied: the described transition probabilities were proposed by omae et al. [ ] . in this study, we also use them for infection simulations. first, we describe parameters of agents, simulation environment, and their initial conditions. the agents live in a -dimensional ( d) space (x and y axes) with minimum and maximum values and as an artificial society. the agents' locations are expressed by utilizing d real numbers from to . the parameters for expressing them are shown in table . additionally, the max simulation period is denoted by e , and the number of houses is denoted by e . we assume that persons (an office worker, homemaker, and student) live in a house. therefore, the total number of populations in an artificial society is × e . next, the number of initial infectors (agents of initial state i) is denoted by e . the state of e persons out of e persons becomes state i (infection). besides, the state of other persons becomes state s. afterward, (x, y) coordinates of the agents' houses and destination facility locations (company, shop, or school) are decided. a coordinate of a house means the location that agents live daily, and its number is e . the destination facility location means the place an individual goes almost daily. it is possible that agents may not go there. the facilities for office workers, homemakers, and students are company, shop, and school, respectively. in a society, there are numerous companies, shops, and schools. therefore, we choose many (x, y) coordinates as variables of e { , , } . in the case of office workers, the company to choose is decided by a uniform random number at simulation start timing. as with office workers, destination facility locations of homemakers and students are selected. further, the destination facility location is one per agent. after the decision, they do not change. moreover, it is possible that agents of state i do not go to the destination facility, such as a company or school, while they go to a hospital to be hospitalized. however, if the capacity limit of isolation wards is reached, they may not be hospitalized even if the agent is an infector. consequently, the capacity limit of isolation wards denoted by e is decided. moreover, we set parameters of β , γ , γ , t e→i , t i→rd described in subsection . . after setting the initial parameters, we start our simulation. further, there are " -day process" and " -step process". we remark that -day process means that the process is conducted at the start timing of a day. in this process, whether or not agents go to the destination facility location (company, shop, and school) is decided based on the basic going out probability e . this value is different for each of the agents. if the agents go to the destination facility location, the going out time e and stay time of facility e are decided. however, the agents of state i go to hospital depending on the "probability of going to a hospital" e . if the capacity limit of isolation wards is not exceeded, the agents of state i are hospitalized. in this case, the basic going out probability e of the agents becomes zero. however, if the capacity limit of isolation wards is exceeded, the agents of state i are not hospitalized. in this case, agents of state i are not isolated in a hospital. therefore, it is possible that they can go outside even if they are infected. then, since they may feel sick, the basic going out probability e is reduced by "decreasing value during state i" e . besides, the -step process means that the process is repeatedly conducted at each minimum unit of time of a simulation. in our simulator, the minimum unit of time is minutes. therefore, step is equal to minutes. agents who decide to go to the facility by a -day process can go out, whereas other agents stay at home. thus, office workers, homemakers, and students go to company, shop, and school, respectively. locations of companies, shops, and schools are respectively denoted by e { , , } , and their coordinates are destinations of agents. the going out time to destination is denoted by e ; at this time, agents go there in the available shortest euclidian distance. after arriving at the destination, agents stay there based on stay time of facility e . afterward, agents go back to their homes. the described process is called a " -step process." since step is minutes, day is equal to steps ( hours). after completing steps, another day is repeatedly started. when the max simulation period e is reached, a simulation is completed. this study aims to verify the effectiveness of cocoa [ ] by applying mas. therefore, we establish parameters to incorporate the app into the existing simulator developed by omae et al. [ ] described in subsections . and . . the newly included parameter is and we call it "app parameter" because it is used to express cocoa. here, p app denotes the usage rate of the app for all the population; p app denotes the decreasing value of going out probability of agents that received notification of their contact with infectors from the app (infectors mean agents of state i); p app denotes the infection registration rate of infectors to the app. a notification process of the app, such as cocoa, to an agent that came in contact with infectors is illustrated in fig. . as depicted in fig. , agent i is of state i (infector) and agent j is the person that received notification of having contact with infectors from the app. condition checks whether or not agent i uses the app. if condition is satisfied, condition checks whether or not agent i registered that he/she is an infector. conditions and are conditional branches of the infectors' side. if condition is true, then condition checks whether or not the agent j utilizes the app. furthermore, if conditions , , and are all satisfied, the app of agent j notifies about having contact with an infector or infectors. consequently, agent j can know that "he/she had contact with an infected person." therefore, the basic going out probability of agent j is reduced by app parameter p app . the higher the p app , the more likely conditions and are satisfied. moreover, the higher the p app , the more likely condition is satisfied. persons who come in contact with infectors should not be allowed to go out. besides, if they do not properly use the app, they may not be aware of having contact with an infector or infectors. thus, if each element of the app parameters p p p app is increased, then we believe that the number of infectors could be decreased. in an artificial society, the reducing period of going out probability of agent j who got contact information via the app is weeks. this is because the contact recording period of the real app developed by the japanese government in the real world is weeks [ ] . moreover, the real app notifies about contact information when a person has contact with infectors for at least minutes [ ] . further, our simulator's minimum unit time is minutes. therefore, if agents have contact with infectors for at least minutes, then the app notifies them concerning this. we examine that the app can effectively reduce the number of infectors. therefore, we conduct simulations to verify the effectiveness of the app, such as cocoa. besides, our simulations' conditions are presented in table . the max simulation period e is days, and the number of houses e is (the total population is ). the number of initial infectors e is persons. further, the basic going out probability e , going out time e , and stay time of facility e of each agent are decided by uniform random number or gaussian random number. a method for giving parameters of going out by applying probability distributions is based on previously conducted studies of mixed models of mas and seir [ , ] . the number of companies, shops, and schools is facilities (the number of total facilities is ). next, we describe the parameters of the infection transition. we recall that the minimum unit of time of the simulator is minutes. therefore, it is desirable to set an actual covid- infection probability β for minutes. however, an actual covid- infection probability is unclear. consequently, we find the infection probability that persons from % to % of the total population become infectors in days when all agents do not use the app. for this result, we set β = . % as the infection probability. afterward, we set parameters t e→i and t i→rd . ohashi's assumption [ ] states that the average incubation period is days and the average infection period is days. parameters t e→i and t i→rd were determined by adding ± days to ohashi's assumption [ ] . thus, the incubation periods t e→i are , , and days. the infec-y. omae registration rate of infected persons , , · · · , [%] a ∼ b: uniform random number from a to b. a ± b: gaussian random number of mean a and std. b. dvp*: decreasing value of going out probability tion periods t i→rd are , , and days. parameters t e→i and t i→rd are decided by uniform random number from the above dates for each agent. since we consider that agents in state i are sick, their going out probability decreases. we set % as the decreasing value of going out probability of state i's agents e . to confirm the effectiveness of reducing the number of infectors of only the app, there are no hospitals in an artificial society. therefore, we set as the capacity limit of isolation wards e , and this implies that agents cannot be admitted to hospitals. as a result, the probability of going to a hospital e and fatality rate (hospitalization) γ in table are empty. next, we consider fatality rate (nonhospitalization) γ . for covid- 's fatality rate, the mitsubishi research institute [ ] reported that the fatality rate of countries that were medically collapse is over %. even though in other cases, the fatality rate is about % (e.g., italy: . %, spain: . %, iceland: . %, singapore: . %, as of apr. [ ] . in the case of italy, the death of out-of-hospital increased because of the covid- outbreak according to baldi et al. [ ] ). therefore, we set % as fatality rate (nonhospitalization) γ . next, we describe the app's parameters p p p app defined by eq. ( ) . our study objective is to verify the effectiveness of the app. therefore, we set many combinations as the app parameters. as demonstrated in table , the values of p app , p app and p app are %, %, · · · , %, respectively. since the value of a parameter is patterns, all parameters' combination is × × = patterns. additionally, because the simulations of infection spread are stochastic events, it is desirable to conduct simulation multiple times per scenario and calculate the average value of the total infectors. if this is not the case, the consequences can be influenced by chance. therefore, we conduct simulations of about times per scenario while changing the random seeds. in other words, the total number of conducted simulation for days is scenarios × random seeds = times. tions now, we checked the number of total infectors of the scenario of all agents who do not use the app. in the results generated by random seed of patterns, there were cases in which the infection did not spread even though the app was not used (the total number of infectors at the end of the simulation is below ). since the random seeds generating the above results are inappropriate to verify the effectiveness of the app, we excluded them from our analysis. thereafter, we calculated the total number of infectors. the total number of infectors means the sum value of state e, i, r, and d. further, the total number of infectors at the end of the days simulation of the scenarios of all infectors who registered that they are infected (i.e., p app = %) is shown in fig. . the vertical and horizontal axes denotes the respective app parameters p { , } app . if either p app or p app is , it implies that the app does not work. therefore, the number of infectors in row and column of p { , } app = % has the same value ( persons). additionally, this number is less than if the app is effective. as illustrated in fig. , as p app and p app increase, the total number of infectors decreases. from the result presented in fig. , we consider the target value of the app usage strategy required to halve the number of infectors compared with when the app is not used. since the total number of infectors is persons in the case in which the app is not used, the standard value is less than × ( / ) = . persons. as shown in fig. , there are no cases of the number of infectors less than . persons in the result of the usage rate of the app (p app = %). therefore, the usage rate of app p app has to be at least %. in the case in which % ≥ p app , the scenarios that the total number of infectors is less than . persons are as follows: for all populations, it is difficult to reach the usage rate ( %) of the app p app . therefore, p p p app = ( %, %, %) or ( %, %, %) is a realistic target value. we confirmed that the total number of infectors become under half value if about half of agents use the app and about half of the frequency of going to the facility. next, we consider the target value for reducing the number of infectors to about / . the standard value is less than × ( / ) = . persons. from the result depicted in fig. , the scenarios that the total number of infectors is less than . persons are as follows: (p app , p app , p app ) = ( %, %, %), . ( ) (p app , p app , p app ) = ( %, %, %), . ( ) (p app , p app , p app ) = ( %, %, %). . ( ) the first case p p p app = ( %, %, %) means that if the usage rate of the app is low, the agents who had being in contact with infectors extremely reduce their going out. the second and third cases p p p app = ( %, %, %) and ( %, %, %) are the relaxation of eq.( )- ( ) , which can reduce the number of infectors by half. as of aug. , the japanese population download rate of co-coa is about % [ ] . therefore, the usage rate of app p app = % is criteria that will be reached in the near future. consequently, the most important scenario is the first case p p p app = ( %, %, %). moreover, the above results are the scenarios in which all infectors who use the app registered that they are infected (p app = %). cocoa does not leak infection information, but infectors may be afraid of their infection information being leaked. as a result of this, it is possible that infectors feel unwilling to register when they are infected. to verify this scenario, we changed p app from % to % in an increment of %. the results obtained by this procedure are shown in fig. . besides, we note that figs. ( ), ( ), ( ), and ( ) are the results of p app = %, %, %, and %, respectively. moreover, as p app increases, the total number of infectors decreases. in the case of infection registration rate p app = %, to reduce the total number of infectors to half, the usage rate of the app p app = % is required. this is a very difficult condition. moreover, in the case of p app = %, %, if the usage rate of app p app is at least % or more, then the scenarios that reduce the total number of infectors to half appears. thus, because rapidly increasing the app's usage rate is difficult, it is important to register infection information. . . results of discussions: the relationship between app parameter p p p app and time series trends of the total number of infectors in subsection . , we discussed the relationship between the app parameters p p p app and the number of infectors at the end of the days simulation. this discussion is important, but it is unclear whether or not the number of infectors increases after the max simulation period. to clarify this viewpoint, it is necessary to check the time series data of the number of infectors in each scenario. however, considering that we simulated many scenarios, it is difficult to show all the time series data. therefore, we calculate the index to determine whether the number of infectors has exponential, linear, or logarithmic growth. afterward, we discuss how to converge infection spread based on this index. now, the amount of the difference of the total number ( ) ( ) ( ) is calculated, where n ip (t; p p p app ) denotes the total number of infectors in t days of scenario using the app parameter p p p app , and ∆n ip (t; p p p app ) denotes the differential value. moreover, we develop a linear regression model with the following intercept: ∆n ip (t; p p p app ) = wt + b, . . . . . . . . ( ) where here, ∆n ip (t; p p p app ) denotes an estimated value of ∆n ip (t; p p p app ), and t max denotes the max simulation period (t max = days). then, the coefficient of linear regression model w is the index for expressing time series trend for the total number of infectors n ip (t; p p p app ). the meaning of the index is shown in fig. . the upper side in fig. demonstrates diagram of n ip (t; p p p app ). the left and center or right sides of fig. demonstrate exponential and linear or logarithmic growth, respectively. the underside in fig. represents regression value ∆n ip (t; p p p app ) calculated by utilizing eq. ( ) . as depicted in fig. , w > , w = or w < represents exponential, linear or logarith-mic growth, respectively. therefore, since the exponential growth of the number of infectors (w > ) causes infection pandemic, it should be avoided. however, because the logarithmic growth (w < ) leads to the convergence of infection, it is a desirable case. we calculate coefficient w of all the scenarios using the app parameters p p p app . the results are shown in fig. . the vertical and horizontal axes represent w and scenarios, respectively. the three numbers of the horizontal axis are app parameters p first, we discuss the case of p app = % shown in fig. ( ) . in this case, if p app = % and p app is % or less, the signs of coefficient w are positive. it illustrates an exponential growth. in contrast, if p app = % and p app is % or more, coefficient w is usually approximately . therefore, to avoid infection pandemic during the periods of low app usage rate, it is important to keep p app at % or more. second, we discuss the case of p app = % shown in fig. ( ) . in this case, the exponential, linear, and logarithmic growth are mixed. in the case of p app = % and p app = %, there is an exponential or linear growth. in addition, if p app is % or more, the growth of infectors is almost logarithmic. thus, when the usage rate of app p app and decreasing value of going out p app are % or more, scenarios of convergence of infection appear. however, if p app and p app are % or more and p app is %, then coefficient w is approximately . in this case, the number of infectors keeps increasing in proportion to time. therefore, it is important to call on infected persons to register and increase p app . finally, we discuss the case of p app = %, %, and % shown in fig. ( )- ( ) . in this case, the growth of the total number of infectors is almost logarithmic. therefore, if about % or more of all population uses the app, then the infection pandemic may converge. as of in japan, the rate of spread of smartphone is . % [ ] . therefore, if about % of smartphone users do not use the app, it is difficult to reach the usage rate of the app, which is about % of all the population. it is difficult to achieve this condition. thus, knowing the condition for overcoming the spread of covid- is important. besides, the result of overcoming covid- by applying the usage rate of the app, which is % or more, is similar to those of hinch et al. [ ] and kurita et al. [ ] . to overcome the spread of covid- , hinch et al. [ ] reported that the usage rate of the app, i.e., about % of the target populations is required. table of kurita et al. [ ] shows that the number of covid- reproductions is less than . in many cases when the usage rate of the app is about % or more. this means that the spread of covid- converges. therefore, our study supports the results of hinch [ ] and kurita [ ] . it is noteworthy that similar results were obtained using various methods for verifying the effectiveness of the app, such as co-coa. in this study, we utilized the mas to verify the effectiveness of an app, such as cocoa. as basic trends, as app parameter p p p app increases, the total number of infectors decreases (see figs. and ) . therefore, the usage rate of the app, decreasing value of the going out prob- ability of persons who had contact with infectors, and the registration rate of infectors can effectively reduce the spread of covid- . moreover, if the usage rate of the app is % or more, the time series trends of the number of infectors in many scenarios have logarithmic growth (see fig. ). our study's result supports some previously conducted studies [ , ] . in the case of the simulation task that cannot verify reliability in the real world, the accumulation and comparison of case studies using various simulation methods is important. thus, this study's result is beneficial. as of aug. , , the installation rate of the co-coa of all the japanese population is about % [ ] . besides, since one is required to always turn on his/her smartphone's bluetooth, the actual usage rate of the app will be lower than the installation rate. as of aug. in japan, it is very important to call on everyone to install cocoa. we observed that one of the reasons that many peoples do not install cocoa is because they do not know its effectiveness in the reduction of infectors. to solve this problem, it is very important that researchers report the effectiveness of this app even if the simulation environment is performed in an artificial society. moreover, we hope that the accumulation of case studies, such as our study, can lead to an increase in the usage rate of the app. besides, this study reported the effectiveness of only the app. as mentioned in section , our simulator can include the capacity limit of isolation wards. in our future studies, we will report the effectiveness of multiple measures, such as mixing the app and capacity limit of isolation wards to overcome the spread of covid- . finally, we will announce the effectiveness of the app and other measures to everyone and request them to install and use the app. we believe that apply consistent efforts can help us to overcome the spread of covid- . basic policies for novel coronavirus disease control by the government of japan (summary ministry of health, labor, and welfare of the japanese government ministry of health, labor, and welfare of the japanese government ministry of health, labor, and welfare of the japanese government covid- infection control team of the japanese government ministry of health, labor, and welfare of the japanese government "occurring of subclinical pathogen carrier of covid- transmission of -ncov infection from an asymptomatic contact in germany contact-confirming application trends in each country on agent-based approach to influenza and acute respiratory virus infection simulation sis evolutionary game model and multi-agent simulation of an infectious disease emergency an agent-based spatially explicit epidemiological model in mason pandemic simulations by made: a combination of multi-agent and differential equations, with novel influenza a (h n ) case the effectiveness of quarantine of wuhan city against the corona virus disease (covid- ): a well-mixed seir model analysis healthcare impact of covid- epidemic in india: a stochastic mathematical model effective configurations of a digital contact tracing app: a report to nhsx state-of-the-art of social system research -methods of evaluation and analysis ( )-resolutions of models and validation a prediction method for viral disease outbreak using a multi-agent simulation including capacity limitation for isolation wards and stay-at-home orders modified seir and ai prediction of the epidemics trend of covid- in china under public health interventions seir modeling of the covid- and its dynamics stability analysis and numerical simulation of seir model for pandemic covid- spread in indonesia epidemic prediction of covid- infection covid- policy analysis report : each country's infection status from the analysis of lethality out-of-hospital cardiac arrest during the covid- outbreak in italy labor and welfare of the japanese government communications usage trend survey effectiveness of co-coa, a covid- contact notification application key: cord- -lpmsalux authors: alqahtani, amani s.; bindhim, nasser f.; tashani, mohamed; willaby, harold w.; wiley, kerrie e.; heywood, anita e.; booy, robert; rashid, harunor title: pilot use of a novel smartphone application to track traveller health behaviour and collect infectious disease data during a mass gathering: hajj pilgrimage date: - - journal: j epidemiol glob health doi: . /j.jegh. . . sha: doc_id: cord_uid: lpmsalux this study examines the feasibility of using a smartphone application (app) to conduct surveys among travellers during the hajj pilgrimage, where the use of apps has not been evaluated for infectious disease surveillance. a longitudinal study was conducted among pilgrims at the hajj using an iphone app with separate questionnaires for three study phases covering before, during, and after hajj. forty-eight pilgrims from countries downloaded the app. respondents were aged between and (median ) years and . % ( / ) were male. of these, % ( / ) completed the first phase, % ( / ) completed both the second and third phases, and of these reported meningococcal vaccination, with % ( / ) receiving other vaccines. all ( ) reported hand hygiene use and % ( / ) wore a facemask at some point during the pilgrimage. four ( %) reported close contact with camels. respiratory symptoms commenced from the th day of hajj, with sore throat ( %) and cough ( %) being the most common. three participants ( %) reported respiratory symptoms after returning home. conducting a prospective survey using a smartphone app to collect data on travel-associated infections and traveller compliance to prevention is feasible at mass gatherings and can provide useful data associated with health-related behaviour. pilot use of a novel smartphone application to track traveller health behaviour and collect infectious disease data during a mass gathering: hajj pilgrimage the annual hajj pilgrimage to mecca, saudi arabia, is a striking example of intensely crowded human activity where - million pilgrims assemble from over countries. incidence of acute respiratory tract infections (ari) is high [ , ] . moreover, emergence of middle east respiratory syndrome coronavirus (mers-cov) in saudi arabia and other countries poses a new public health challenge [ ] . in order to reduce the risk of ari among hajj pilgrims, the saudi arabian health authority recommends a range of infection control measures [ ] , however, compliance to these measures is highly variable [ ] [ ] [ ] [ ] . several studies have addressed pilgrim knowledge, attitude, and practice (kap) towards preventive measures and infectious diseases, finding that their understanding about the potential severity of respiratory infection and the need for protective measures was inadequate [ ] [ ] [ ] [ ] [ ] [ ] . gautret et al. [ ] found that < % of french pilgrims were aware of social distancing, available treatment options, and facemask use as precautionary measures against the spread of respiratory infections. other studies assessed pilgrim knowledge of emerging infections, such as mers-cov and ebola, and found that % of pilgrims were not aware of mers-cov circulation in arabia and about % had no accurate knowledge of ebola transmission. however, longitudinal studies examining these questions before, during, and after travel are lacking. conducting longitudinal studies among travellers during mass gatherings involves many challenges, including requirements of a large sample size and high response rate, as well as continuous follow up throughout the course of travel with real-time data capturing. conducting such studies using conventional ''pen and paper"-based methods requires significant time and resources. smartphones are increasingly becoming an integral part of modern life, making it possible to conduct prospective surveys among hajj pilgrims through their use. several studies have demonstrated their usefulness in conveying health messages in a variety of contexts and audiences, with high response and retention rates and fewer dataentry errors during descriptive studies and randomised controlled trials [ , ] . thus, smartphones may provide better platforms to conduct prospective surveys among hajj pilgrims than conventional ''pen and paper"-based methods [ ] [ ] [ ] [ ] [ ] . additional advantages include constant internet connection, location-detection services, and user proximity making it an ideal tool for collecting infectious disease data during mass gatherings. data concerning smartphone usefulness in infectious disease research at mass gatherings are very limited [ ] . therefore, we conducted a pilot study using a smartphone app to examine its feasibility to track not only hajj pilgrim kap regarding preventive measures, but also symptom onset and participation in high-risk activities before, during, and after hajj . a prospective cohort study was conducted among hajj pilgrims at three time points, including before, during, and after hajj (between september th and october th). this involved using three sets of questionnaires in english, including a pre-hajj questionnaire composed of questions, seven identical pages of hajj questionnaires (containing five questions per questionnaire) each to be completed daily over a week during the peak hajj period, and a post-hajj questionnaire composed of six questions. we developed an iphone application (app) called 'hajj health diary', utilising the 'health monitor' app template [ ] and released it in the apple app store on september , . users started by registering their device in our online secure research database and were assigned a unique identifier for their device. this method was used successfully in previous studies [ , , ] . the app determined user location through the smartphone location service and recorded it in our research database each time the participant used it. this study and the materials described below were approved by the human research ethics committee at the university of sydney (project no: / ). the terms and conditions of downloading the app were communicated before obtaining participant consent. participants aged years and older and participating in the hajj pilgrimage in mecca were included. attendance was confirmed by tracking their location during hajj. participants aged less than years or whose stay in mecca during the hajj period could not be confirmed through location-tracking services were excluded from the study. the app consists of three main screens ( fig. ) and an 'about' screen, which includes the participant information sheet and consent form. the first screen (first phase) is the pre-hajj questionnaire, including data on participant demographics, pre-existing chronic diseases, vaccinations received before travel, factors influencing vaccination decision and uptake, perception of the risk of respiratory infection during hajj, willingness to participate in highrisk activities, such as drinking unpasteurised milk, and awareness of official health recommendations provided by saudi arabian authorities. this phase lasted from september to , . the second screen (second phase) included the hajj questionnaire, which consisted of seven pages of identical questionnaires, one page for each day of the 'peak' hajj period. these questionnaires asked about development of respiratory symptoms and their adherence to preventive measures. given that the 'peak' hajj period in lasted from october to , the app pushed a local notification each day to the user to complete the questionnaire. if the user did not complete it on the same day, they were reminded to do it the next time they opened the app. therefore, the first default questionnaire needed to be completed first before going to the next. the participant could not start completing the hajj questionnaire before completing the pre-hajj questionnaire and not before october , . the third screen (third phase) includes a post-hajj questionnaire, including questions about participant use of infection prevention methods, involvement in high-risk activities, and development of ari symptoms week after the conclusion of hajj. as the hajj pilgrims usually spend up to weeks in saudi arabia before returning to their home countries, the app pushed a daily reminder to complete the post-hajj questionnaire from october to , . if the participant did not have an internet connection, data were stored locally on the device and automatically transferred to our database as soon as an internet connection was available. to understand usage behaviour, we also collected data entry date and time and transferred it to the database. participants had to complete all questions before submitting any of the questionnaires and could not submit any questionnaire more than once, even if they deleted and re-installed the app. we used both active and passive recruitment strategies. first, we recruited australian pilgrims through distribution of brochures during pre-hajj travel seminars in sydney. second, we released the app to the apple app store globally (and exclusively), expecting that some users searching for hajjrelated apps would find it. this was likely given that our app was the only one that would appear under the search term 'hajj health' in that outlet. in the pretravel questionnaire, we included a question on how the participants heard about the app. we identified a priori that there might be potential challenges, including loss of internet connection, app de-installation and re-installation, advertent or inadvertent omission of survey questions, and failure to follow the recommended order while completing the surveys. to minimise these pitfalls, the following measures were taken: if the participant lost the internet connection, the data would be stored locally on the device and transferred to our database as soon as the internet connection became available. if the participant de-installed the app and re-installed it, the app would not allow resubmission of the same questionnaire, requiring the participant to start from where they finished before de-installing the app. this would also maintain the sequence of questionnaire completion. to avoid any omission or delay in completing the questionnaires, the app would push a daily reminder to complete the current survey. the app was downloaded by pilgrims from countries (table ) . of them, % ( / ) completed the first phase (pre-hajj questionnaire) and of those, % ( / ) completed all three study phases (fig. ) . of the participants who completed the pre-hajj questionnaire, ( . %) opened the app in saudi arabia at least once while having internet connection. the respondents were aged between and (median ) years and . % ( / ) were male. a large portion ( . %, / ) was university educated and . % ( / ) were employed. sixteen ( %) participants had pre-existing chronic diseases, including five with diabetes ( . %), three each with hypertension, bronchial asthma, and hypercholesterolemia ( . %), and one with heart disease ( . %). the participants stayed in mecca for a median of days (range, - days), . % ( / ) attended hajj for the first time, and only . % ( / ) were aware of annual saudi arabian health recommendations for hajj pilgrims. in terms of how the participants heard about the app, . % ( / ) reported first seeing it in the apple app store, . % ( / ) heard about it in pre-hajj seminars, . % ( / ) from the study researchers, and . % ( / ) from other sources. regarding convenience of using the app, . % ( / ) found it very convenient, . % ( / ) found it slightly convenient, . % ( / ) found it a little inconvenient, and . % ( / ) found it very inconvenient. those who found it inconvenient (to any degree) left the survey incomplete. concerning usage behaviour, the number of participants who completed the hajj questionnaire on the day the reminder was pushed, i.e., the st day, was , but dropped to on the nd day and remained at from the rd day onward (fig. ) . those who did not complete the hajj questionnaire on the specified day completed it within the next - days. all participants who completed the hajj questionnaire subsequently also completed the post-hajj questionnaire. of those who completed the study phases, all reported receiving the compulsory meningococcal vaccine. the main factors driving meningococcal vaccine uptake were severity of the disease % ( / ) and effectiveness of the vaccine % ( / ). only ( %) pilgrims received other vaccines before hajj ( had influenza vaccine and one pneumococcal vaccine). three of these had comorbidities. forty-four percent ( / ) of participants were unconcerned about catching influenza while at hajj and % were unconcerned about developing a cough. forty percent ( / ) of respondents expressed concern about contracting mers-cov during hajj, at least to a modest extent, while the rest did not. however, % ( / ) of participants were willing to visit a camel farm during hajj and % ( / ) were willing to drink unpasteurised camel milk if offered in saudi arabia. in practice, % ( / ) of pilgrims ( american, canadian, and australian) actually reported coming into contact with camels, including visiting a camel farm, taking photographs with camels, and drinking their milk ( ) . the onset of respiratory tract symptoms began from the th day of the peak hajj period and continued over the next several days. sore throat ( %) and cough ( %) were the most frequently reported symptoms (fig. ) . after returning home from hajj, % ( / ) of participants reported developing a cough and sore throat within week and, among these, one pilgrim from australia reported having had contact with camels during hajj. sixty-four percent ( / ) reported wearing a facemask during hajj, with uptake highest on day ( table ). protection from infectious agents and air pollutants was the main reason for mask use. difficulty in breathing and a feeling of suffocation were commonly cited as barriers to the use of facemasks. on the other hand, all participants practiced hand hygiene at some point (mostly during the st days) during hajj. respondents stated that hand hygiene was easy to implement, convenient, and believed it to be effective in preventing infections. overall, this pilot study indicates that conducting a prospective survey using a smartphone app to collect data on travel-associated infections and traveller compliance to prevention is feasible, given that the response rate was > %. this survey also demonstrates that many pilgrims partake in activities that may increase risk of acquiring emerging infections. of the people who downloaded the app, ( %) participated in the first survey and of these, ( %) went on to complete it. previous paperbased cross-sectional surveys reported response rates ranging between < % and > % [ , , ] . the studies with high response rates involved recruitment with continuous follow up of worshippers throughout their pilgrimage. this approach requires significant resources and greater investment of time and cost [ , ] . other paper-based surveys where pilgrims were not followed up continuously had response rates as low as . % [ ] . the strengths of our study include its low cost, ability to reach far and wide to allow real-time analyses and longitudinal follow-up, and ability to capture data daily during the peak hajj period, something not accomplished in other studies. this pilot survey reveals that all pilgrims complied with hand hygiene. this is supported by a review by benkouiten et al. [ ] , which found that hand hygiene was the most popular nonpharmaceutical preventive measure among hajj pilgrims. ease of use and participant belief regarding its effectiveness against infection were important driving factors. in this study, we found that respiratory symptoms commenced on the th day of tent stay during hajj and continued thereafter, with cough and sore throat being the most commonly reported symptoms. this is likely because the incubation period of most commonly circulating respiratory viruses is about - days [ ] . used in combination with a geographic information system as a tool for syndromic surveillance, this novel method can be used to detect real-time clusters of respiratory infections at hajj and other mass gatherings. although mers-cov has been circulating in saudi arabia since , no case of hajj-associated mers-cov has been reported [ ] . evidence suggests that mers-cov can be transmitted to humans through close contact with an infected camel [ ] . interestingly, our study identified that some pilgrims had close contact with camels, including visiting camel farms, photo opportunities with camels, and drinking their raw milk. through this survey, we identified one australian pilgrim who had close contact with camels and subsequently developed respiratory symptoms within week of returning home. further follow up of the case was not possible, however, given that no mers-cov case was reported in australia, it is highly unlikely that the person had mers-cov. therefore, this pilot study suggests that smartphones could help detect patients with potential emerging infectious diseases. electronic surveillance to identify outbreaks of infectious diseases at a mass gathering has been attempted previously [ ] . for instance, surveillance using electronic medical records deployed during the winter olympic games helped detect an influenza outbreak, which was subsequently described with the aid of laboratory diagnosis [ , ] . digital interfaces, including smartphones, were applied at the london olympics to identify illnesses and injuries among athletes [ ] . our study, the first of its kind at hajj, demonstrates the feasibility of a smartphone app in a prospective survey of pilgrim illness and adherence to preventive measures throughout the course of travel. owing to delays in development, testing, and app store approval, the app was only released a few weeks before hajj, limiting the amount of time it was available to respondents prior to their journey. we speculate that earlier app release will result in greater numbers of participants. because this survey was conducted only in english, multilingual applications could have expanded participation into diverse language groups. data on pilgrim demographics show that - % of pilgrims are older than years of age [ , ] , while most smartphone users are aged - years [ ] , which might have impacted study outcomes. however, since smartphone use is gradually becoming ubiquitous, respondent demographics are likely to be less important in the future. finally, we designed the app only for iphone users, thus excluding users of other smartphone platforms, such as android or windows mobile. extending availability of this app to other platforms will likely increase participation rates. in conclusion, this pilot study demonstrates that smartphone apps can be used to conduct surveys to prospectively gather data concerning onset and progression of symptoms and location information during mass gatherings. such data collection can potentially reinforce education associated with disease prevention behaviours, thus improving public health. a larger study with multilingual apps for both iphone and android smartphones is planned for hajj . mass gathering medicine: hajj and umra preparation as a leading example prevention of influenza at hajj: applications for mass gatherings travel implications of emerging coronaviruses: sars and mers-cov health conditions for travellers to saudi arabia for the umra and pilgrimage to mecca (hajj) vaccinations against respiratory tract infections at hajj protective practices and respiratory illness among us travelers to the hajj the prevalence of acute respiratory symptoms and role of protective measures among malaysian hajj pilgrims detection of respiratory viruses among pilgrims in saudi arabia during the time of a declared influenza a (h n ) pandemic hajj pilgrims' knowledge about acute respiratory infections australian hajj pilgrims' infection control beliefs and practices: insight with implications for public health approaches australian hajj pilgrims' knowledge, attitude and perception about ebola french hajj pilgrims' experience with pneumococcal infection and vaccination: a knowledge, attitudes and practice (kap) evaluation health knowledge, attitude and practice among iranian pilgrims pro-smoking apps for smartphones: the latest vehicle for the tobacco industry? depression screening via a smartphone app: cross-country user characteristics and feasibility confirming the one-item question likert scale to measure anxiety adherence to a smartphone application for weight loss compared to website and paper diary: pilot randomized controlled trial smartphone versus pen-and-paper data collection of infant feeding practices in rural china the world's first application of participatory surveillance at a mass gathering: fifa world cup health monitor project who uses smoking cessation apps? a feasibility study across three countries via smartphones circulation of respiratory viruses among pilgrims during the hajj pilgrimage protective measures against acute respiratory symptoms in french pilgrims participating in the hajj of non-pharmaceutical interventions for the prevention of respiratory tract infections during hajj pilgrimage incubation periods of acute respiratory viral infections: a systematic review has hajj-associated middle east respiratory syndrome coronavirus transmission occurred? the case for effective post-hajj surveillance for infection evidence for camel-to-human transmission of mers coronavirus new digital technologies for the surveillance of infectious diseases at mass gathering events hospital electronic medical recordbased public health surveillance system deployed during the winter olympic games illness and injury in athletes during the competition period at the london paralympic games: development and implementation of a web-based surveillance system (web-iiss) for team medical staff causes of mortality for indonesian hajj pilgrims: comparison between routine death certificate and verbal autopsy findings digital industry association for australia. australian mobile phone lifestyle index professor robert booy has received funding from baxter, csl, gsk, merck, novartis, pfizer, roche, romark, and sanofi pasteur for the conduct of sponsored research and travel to present at conferences or consultancy work. all funding received is directed to research accounts at the children's hospital at westmead. dr. anita e. heywood has received grant funding for investigator-driven research from gsk and sanofi pasteur. dr. harunor rashid received fees from pfizer and novartis for consulting or serving on an advisory board. the other authors have no competing interests to declare. key: cord- -byophdo authors: zahid, talal; alyafi, rusha; bantan, noor; alzahrani, rana; elfirt, eman title: comparison of effectiveness of mobile app versus conventional educational lectures on oral hygiene knowledge and behavior of high school students in saudi arabia date: - - journal: patient prefer adherence doi: . /ppa.s sha: doc_id: cord_uid: byophdo objective: this study aimed to evaluate the impact of two different oral health education approaches, a mobile application (the brush dj app) and conventional educational lectures, on the oral hygiene knowledge and behavior of high school children. methods: the research was a cross‐sectional study of students from two public schools in jeddah city, saudi arabia. an eighteen-item questionnaire was used for this purpose. those who completed the baseline questionnaire were allocated to one of two groups: ( ) mobile application and ( ) educational lecture. a follow-up survey was later conducted at three months, which repeated eight of the eighteen questions asked in the baseline survey. the change in oral hygiene attitude and behaviors was compared across both groups. results: the brush dj app was found to be equally effective compared to educational lectures in changing oral health knowledge, attitude and behavior. both groups showed significant improvements in almost all aspects of oral health, except for the frequency and duration of tooth brushing in the app group. there was no change in twice daily tooth brushing of app users, and less than % reported brushing their teeth for minutes. a statistically significant change, however, was noted among lecture group participants in these two areas of oral hygiene routine. the app was also found to be more difficult in usability than educational lectures (p = . ). conclusion: the brush dj app may be a valuable tool to improve oral health knowledge, attitude and behavior. however, the app needs some improvements. the content and features of the app need to be structured in a way that it allows for personalization and is more interactive, practical and user-friendly. oral diseases are major public health problems of considerable social and economic burden, owing to their high prevalence. according to the global burden of disease study , about half ( %) of the world's population is affected by oral diseases. two oral conditions have been reported to account for most of the global oral health burdens: tooth decay (dental caries or cavities) and periodontal (gum) disease. other oral conditions that commonly impact the overall well-being and quality of life include dental trauma, oral cancer, tooth wear (dental erosion, attrition and abrasion), edentulism, cleft lip and palate and oral manifestations of hiv. on a global scale, tooth decay of permanent teeth was found to be the most prevalent of all diseases and periodontal disease was the th most prevalent condition. statically, however, the scenario is far worse for underprivileged people living in developed and developing countries. in most industrialized countries, tooth decay has been reported to affect around - % of schoolchildren as well as adults. notable causes for the high prevalence of dental caries and periodontal diseases are poor oral hygiene, inadequate fluoride exposure and tobacco use. the development of dental caries in the oral cavity can be substantially reduced by ensuring a constant low-level exposure to fluoride. this can be achieved by drinking fluoridated water, using a fluoride-containing toothpaste ( to ppm) or applying a topical fluoride gel. , on the other hand, the prevention of periodontal diseases can be largely achieved by maintaining proper oral hygiene such as daily brushing and flossing. hence, most clinicians now suggest twice-daily tooth brushing with an antimicrobial toothpaste containing fluoride to prevent tooth decay and gum diseases. however, there is also a need to educate people concerning the benefits of healthy oral habits to maintain good oral hygiene. proper knowledge and awareness related to oral health are essential for developing healthy oral hygiene behaviors. this has been demonstrated in earlier studies, which reported a direct association between increased oral health knowledge and better dental care. [ ] [ ] [ ] [ ] [ ] consequently, to effectively prevent and control the prevalence of oral health burdens, it is important to develop healthy oral habits at an early stage of life. hence, school children are the ideal target population given the fact that healthy hygiene behaviors that stem from the school-age years usually carry over into adulthood. however, although a larger body of literature has been published in many countries to evaluate the oral health knowledge, attitudes, and behavior among students, to date little attention has been given to develop effective oral disease preventive programs for this key target group. mhealth or mobile health is an emerging sub-segment of ehealth that involves the use of mobile communication and/or wearable devices to improve the practice of medicine and public health. several recent studies including systematic reviews have suggested the use of mobile devices as an invaluable adjunct, which could help improve the oral hygiene compliance in different age groups. [ ] [ ] [ ] [ ] in addition, it has been shown that the use of mhealth with a conventional oral health education program is more effective in improving compliance among adolescent patients than verbal instructions of oral hygiene alone. , to date, however, only a limited number of studies have been conducted to evaluate the quality and effectiveness of numerous mobile applications developed for improving oral hygiene behavior. this study aimed to assess and compare the impact of two different approaches of oral health education, mobile application and educational lecture, on the oral hygiene knowledge and behavior of high school children. the present study was designed as a quasi-experimental study. the baseline survey was done to assess the knowledge and attitude of high school students towards oral hygiene. a follow-up survey was later conducted at three months to determine whether there was any improvement in oral health knowledge, attitude and behavior. the study protocol was reviewed and approved by the research ethics committee at the king abdulaziz university. the survey was conducted from november to march , using a structured questionnaire that was developed after a thorough literature review. a small working group committee independently reviewed and validated it. study participants were recruited from two public schools (a female governmental school and a male high school) in jeddah city, saudi arabia. the principals of the two schools were contacted, and the permission to visit and conduct the survey was obtained. student participation was completely voluntary. all students who were willing to take part in the baseline and follow-up survey and owned a smartphone, tablet or other smart devices were included. the rationale of the study was explained to prospective students beforehand. written informed consent was obtained from parents, guardians or caregivers of each participant before participation. we performed an a-priori sample size (n) calculation of subjects, considering the frequency and duration of toothbrushing as the main outcomes, fixing an absolute error (d) of % and at type i error (z -α/ ) of %, and with an expected proportion (p) of % estimated from a pilot study. we recruited subjects additionally to address loss to follow-up during the study. the questionnaire an eighteen-item questionnaire was used for the baseline survey. questionnaires were distributed manually in randomly selected classes. collected data included demographics as well as information related to oral hygiene knowledge, attitude, and behaviors. participants who completed the initial survey (baseline) were allocated to one of two groups using simple randomization: (a) mobile application group and (b) education lecture group. the improvement in oral hygiene behavior was compared across both groups. eight of the eighteen questions used in initial survey were repeated in the follow-up survey to assess the change in knowledge and attitudes towards oral hygiene. additional questions were included in the followup survey only to assess the efficacy of the method used (application vs lecture). we used two different approaches of oral health education to determine the oral hygiene knowledge and behavior of high school children: ( ) mhealth and ( ) conventional dental education lectures. in this study, the mobile application used was the brush dj app (ios version . . / android version . . , ben underwood). we selected this application for multiple reasons: it is free and user friendly, listed in the apps library of the national health service (nhs) uk, and reported to be a promising tool that motivates an evidence-based oral hygiene routine. participants of the mobile app group were briefed about the various features of the app and how to install and use it. they were also instructed to use the app twice daily for three months. for the conventional education group, a minute lecture session on good oral hygiene practices was delivered using whiteboard, markers, presentation slides, as well as dental teeth models. a dental hygienist carried out the lecture session. participants of both groups were also supplied with additional learning and/or instruction materials in form of handouts. categorical variables were summarized as counts and percentages. mean and standard deviation were used to summarize the distribution of age. bar plots were used to visualize the responses to attitude and knowledge questions. statistical significance was assessed using chisquare test (or chi-square test of independence where appropriate). the % of positive answers for each question was used as an indicator for the knowledge and attitude towards oral hygiene. mcnemar's test was used to test whether the change in the % of positive answers was significantly different between baseline and follow-up survey within each group. hypothesis testing was performed at . level of significance. a total of students completed the baseline survey questionnaire. of these, respondents were planned to be allocated to the mobile application group and were to the educational lecture group for the follow-up survey. the baseline demographics and the oral health attitudes were not significantly different between the two groups [see table ]. overall, the male to female ratio was comparable in the study cohort ( . % and . %, respectively). the mean age of the included participants was . ± . years. figure outlines the baseline knowledge, attitude and behavior towards oral health among study participants. less than half of the study participants reported using mouthwash while only . % used it correctly (ie, once per day every two weeks) and about % flossed their teeth regularly. when asked about the oral hygiene behaviors, > % of the population reported brushing their teeth twice a day, almost one-third reported using a toothpaste with fluoride and < % reported changing it every three months. nearly % reported having knowledge about the best brushing technique while only . % answered the correct duration for teeth brushing. the frequency of dental visits was largely on demand. only . % reported having a routine dental check-up (at every months). around % of the participants had awareness about the negative impact of poor oral hygiene and % were aware of the main causes of tooth staining. of the students recruited in the baseline survey, ( . %) responded to the follow-up questionnaire ( in the app group and in the educational lecture group). eight questions from the baseline questionnaire were repeated in the follow-up survey. patient preference and adherence : submit your manuscript | www.dovepress.com of the four questions added in the follow-up survey, no significant difference between the two groups was found in three responses [ table ]. overall, almost half of the study participants reported benefitting from the lecture/app to schedule their appointments, around % liked the way of teaching, and nearly % noticed a change in their teeth and gums after using the app/lecture. however, of the two interventions, participants in the app group encountered significantly more difficulties with their intervention compared to those who received the lectures (p = . ). less than % of the participants in the app group reported brushing their teeth until the app music ended at minutes, while more than % of those in the lecture group reported brushing their teeth for minutes as instructed. statistical comparison between the groups was not performed due to the different number of choices across each group. more than % in the app group reported learning the right ways to clean their teeth or gum after watching videos in the app [ figure ]. similarly, around % mentioned taking advantage of the reminder icon in the app. only eight questions were repeated in the follow-up survey. analysis of baseline and follow-up data revealed that the knowledge and attitude of participants towards oral health improved significantly in both groups for almost all aspects except for the frequency of tooth brushing in the app group [see table ]. in the follow-up responses, the use of mouthwash, floss and toothpaste with fluoride, the frequency of toothbrush changing at every months, knowledge regarding the best brushing technique and awareness about the routine dental visit (at every months) were increased among participants of both groups. post-test results showed that the percentage of correct answers was comparatively higher in the app group than the dovepress lecture group for question number and . the opposite was observed for question numbers , and where participants from lecture group gave more correct answers than those who used the app. no statistically significant differences were found between the two groups for the remaining questions. statistical analysis showed that there was no statistically significant interaction between gender and time (p > . ) for the frequency of brushing, frequency of changing the toothbrush, frequency of using mouthwash, best brushing technique, type of toothpaste used, and frequency of dental visit (data not shown). however, the frequency of using floss increased to a higher extent in females ( . % to . %) compared to males ( . % to . %) although these differences were statistically significant only at . level. these non-significant results can be explained by the fact that the percentage of changes was similar across males and females. in the current study, the effectiveness of the brush dj app, which is listed in the nhs choices health apps library, was compared with the conventional means of oral hygiene education. the app was found to be equally effective compared to educational lectures in changing oral health knowledge, attitude and behavior. participants from both groups showed significant improvements in almost all aspects of oral health. the only exception was the frequency of tooth brushing in the app group. our results confirm the findings of earlier studies that increased oral health knowledge and awareness has a positive impact on healthy oral hygiene practices. [ ] [ ] [ ] [ ] [ ] however, it is to be noted that health promotion programs such as oral health education alone may not be sufficient to develop healthy oral hygiene behaviors. such programs can temporarily improve oral health behavior and attitude irrespective of the educational approach; , still, other potential confounders (eg, socioeconomic condition, family situations, peer and social influences, local customs, cultural values and availability of resources) may adversely hinder the development of healthy oral practices. hence, there has been a need for well-structured oral health educational interventions that involve psychological and behavior-change strategies and target a broader goal of making an actual change in attitudes, behaviors, intentions, beliefs and lifestyle. the benefits of using technologies as an educational tool have been highlighted in dental literature. [ ] [ ] [ ] [ ] [ ] smartphones and other mobile devices may be a valuable tool for health promotion, as they are more readily accepted among young people than traditional means of dental education. , , in fact, in clinical settings, dental education apps have been found to improve patient-provider communication. besides, while the conventional dental education programs involve workforce utilization and are difficult to organize, the use of dental education apps may provide an effortless means of delivering health information to a wider audience due to the widespread adoption of mobile devices and their powerful technological advances. , , , in addition, these apps may not only increase knowledge and awareness about maintaining good oral health but also motivate their users to follow an evidence-based oral hygiene routine. in this study, a higher trend of correct answers was observed among students who received educational lectures than those who used the app. our findings can somewhat be considered similar to the findings of a recent randomized controlled trial where the effect of the whiteteeth app was examined on oral hygiene behavior in adolescents. the authors reported that although the mobile app incorporated many behavior change techniques, its effects in changing tooth-brushing frequency and duration were similar to that of usual care. however, the effect of mhealth on school children may also depend on age. in a recently published study on children of to years, zotti et al found mobile apps to be more effective, engaging and fun compared to verbal oral hygiene instructions. the higher trend observed in our study may partly be due to the personalized nature of the educational lectures, as students received individual level oral health education and motivation from a dental professional. it is undeniable that individual-level communication with dentists will have more influence on high school students in making healthcare decisions than an app that attempts to replace a direct patient-provider discussion. , on the other hand, oral health education apps appear to be largely less appealing among school students, as children of this age group tend to use mobile apps mostly for entertainment or gaming purposes rather than education. this has been demonstrated in two recent studies where the authors found that most of the currently available oral hygiene apps lack user engagement and need improvement in terms of aesthetics and information accuracy. , strategies that can be implemented to improve user engagement with an app include ease of use, attractive user interface, unique smartphone features (eg, real-time visualized brushing instructions), and tailored design and information. , , , nevertheless, as suggested by several recently published studies, , , incorporating mobile apps with a standard oral hygiene program may be a more plausible approach for oral health promotion among adolescents than educational lectures alone such as via distance motivational tutoring by a dentist or an educator. another possible reason for the lower trend in the app group might be the differences in correct answers between the two groups at baseline. in fact, the app group participants gave significantly higher correct answers than the lecture group for four of the eight questions repeated in the follow-up survey. nevertheless, the actual reason for this needs further investigation, as the findings could help improve the app and allow inclusion of effective behavior change techniques. in our analysis, it appeared that the brush dj app needs some improvements in several aspects. while the app primarily aims to motivate users to brush twice a day and for minutes by playing music, statistical analysis of baseline and follow-up data revealed no change in twice daily tooth brushing of app users. in addition, less than % reported brushing their teeth for minutes. in contrast, a statistically significant change was noted in these two aspects of oral hygiene routine in lecture group participants. moreover, our findings also sharply contrasted with the results of underwood et al where around % of app users reported brushing at least twice a day and % reported being motivated by the app to brush their teeth for longer. these differences could be in part attributed to the non-user-friendly nature of the app, as around % of app users were not fully satisfied with the way of teaching and % reported encountering difficulty in using the app. in this investigation, however, these app users were not asked why. this needs to be investigated in future studies to help improve the app. there are several limitations to this study. firstly, the results of this study cannot be generalized to saudi young population due to the small sample size and recruitment of participants from two specified schools in jeddah city, saudi arabia. secondly, all information collected in this survey is based on self-reported data of participants. this may have introduced bias due to the chances of falsepositive responses from participants. hence, the data presented on oral health knowledge, attitude and behaviors could be an under-or over-estimation. thirdly, due to logistical constraints, no dental assessment was done to assess the actual oral health status of study participants. this could have allowed for an objective assessment of responses by the participants. other limitations of this study include the cross-sectional design, no adjustment for potential confounders such as socioeconomic condition and social factors, and short time period between baseline and follow-up survey. further research is warranted with a randomized control design to obtain more meaningful outcomes. future studies should also adjust for potential variables and include objective assessment of oral health status before and after the intervention. the use of mobile apps in dentistry may become an alternative to the conventional method of dental education; however, they cannot be a substitute for a direct patient-provider communication. these apps may at best act as a pedagogical enhancing tool in dental education. considering this limitation, such educational apps need to be built through an interdisciplinary collaboration among dentists and other professionals (eg, teachers and psychologists). the developers of these applications also need to involve end-users to design and develop the apps in such a manner that they are more practical and user-friendly. alternatively, if these apps include a feature like distance education or counseling by a doctor or an educator, they can more readily be integrated in the routine dental practice, as this would enable patients to get oral health advice directly from dentists during this covid- pandemic. the present study was conducted to evaluate the effectiveness of two different oral disease preventive approaches, a mobile application (brush dj app) and educational lecture (conventional method), in school children. the use of both the mobile app and educational lecture significantly improved oral health knowledge, attitude and practices among study participants. however, the app was found to be less effective than educational lecture to motivate an evidence-based oral hygiene routine. the brush dj app may be a useful tool to improve oral health knowledge, attitude and behavior. however, it needs some improvements. the content and features of the app needs to be more interactive, practical and user-friendly. the authors report no conflicts of interest for this work. global, regional, and national incidence, prevalence, and years lived with disability for diseases and injuries for countries, - : a systematic analysis for the global burden of disease study oral health, health, and health-related quality of life effective use of fluorides for the prevention of dental caries in the st century: the who approach prevention of dental caries through the use of fluoride-the who approach fluoride and oral health oral health knowledge and habits of senior elementary school students community-based population-level interventions for promoting child oral health el metwally a. oral health knowledge, attitude and behavior among students of age - years old attending jenadriyah festival riyadh; a crosssectional study oral health behaviour and social and health factors in university students from low, middle and high income countries oral hygiene facilitators and barriers in greek years old schoolchildren assessing the impact of oral health on the life quality of children: implications for research and practice the emerging field of mobile health whiteteeth") on improving oral hygiene: a randomized controlled trial a systematic review to assess interventions delivered by mobile phones in improving adherence to oral hygiene advice for children and adolescents effectiveness of a digital device providing real-time visualized tooth brushing instructions: a randomized controlled trial effect of mhealth in improving oral hygiene: a systematic review with meta-analysis usefulness of an app in improving oral hygiene compliance in adolescent orthodontic patients implementing post-orthodontic compliance among adolescents wearing removable retainers through whatsapp: a pilot study the use of a mobile app to motivate evidence-based oral hygiene behaviour oral health knowledge, attitudes and practice behaviour among secondary school children in chandigarh effectiveness of oral health education intervention among female primary school children in riyadh, saudi arabia comparative clinical study testing the effectiveness of school based oral health education using experiential learning or traditional lecturing in year-old children effectiveness of oral health education in children-a systematic review of current evidence the smartphone in medicine: a review of current and potential use among physicians and students healthcare in the pocket: mapping the space of mobile-phone health interventions smartphone interventions for long-term health management of chronic diseases: an integrative review preferences related to the use of mobile apps as dental patient educational aids: a pilot study apps for oral hygiene in children to years: fun and effectiveness the impact of patient-centered communication on patients' decision making and evaluations of physicians: a randomized study using video vignettes mobile apps for oral health promotion: content review and heuristic usability analysis patient focused oral hygiene apps: an assessment of quality (using mars) and knowledge content patient preference and adherence is an international, peer-reviewed, open access journal that focusing on the growing importance of patient preference and adherence throughout the therapeutic continuum. patient satisfaction, acceptability, quality of life, compliance, persistence and their role in developing new therapeutic modalities and compounds to optimize clinical outcomes for existing disease states are major areas of interest for the journal. this journal has been accepted for indexing on pubmed central. the manuscript management system is completely online and includes a very quick and fair peer-review system, which is all easy to use. visit http:// www.dovepress.com/testimonials.php to read real quotes from published authors. key: cord- -tlba flz authors: westgard, christopher; fleming, w. oscar title: the use of implementation science tools to design, implement, and monitor a community-based mhealth intervention for child health in the amazon date: - - journal: front public health doi: . /fpubh. . sha: doc_id: cord_uid: tlba flz it is essential to analyze the local context and implementation components to effectively deliver evidence-based solutions to public health problems. tools provided by the field of implementation science can guide practitioners through a comprehensive implementation process, making innovations more adaptable, efficient, and sustainable. it is equally important to report on the design and implementation process so others can analyze, replicate, and improve on the progress made from an intervention. the current study reports on the design and implementation of an mhealth intervention to improve child health in the amazon of peru. the study aims to provide insight into how an implementation science tool can be used to improve implementation and reporting of an evidence-based intervention in a global health setting. methods: implementation of a community-based mhealth intervention is analyzed and reported through the lens of the active implementation frameworks (aif). the aif is used to analyze the design, implementation, adaptation, and monitoring of the intervention. the implementation process is categorized in the four stages of implementation. the results of the analysis and subsequent implementation activities are reported. results: the exploration stage was used to learn about the local context in the amazonian communities and identify an evidence-based solution to address poor child health. several potential solutions were combined to create an innovative mhealth tool. during the installation stage, the stakeholders worked together to improve the intervention and plan for implementation through human-centered design. the providers in the field were trained and data was gathered to monitor implementation. during initial implementation stage, electronic tablets were distributed to community health agents and continuous quality improvement activities allowed for rapid improvements to be implemented. the intervention moved on to full implementation stage as acceptance and fidelity approached %. conclusion: the aif highlighted several potential barriers to implementation that may have been overlooked without the guidance of a science-based implementation tool. reporting on the implementation process shows how implementation science tools can be used to foresee and address potential threats to successful implementation. the results of this study provide insight into the components of implementation in amazonian communities, as well as the process of using implementation science tools in any global health setting. many public health interventions that have been proven to be effective in controlled settings are not creating the expected impact when replicated in community settings ( ) ( ) ( ) ( ) ( ) ( ) . there are interventions that have been effective at improving child health and development in low-resource community settings, however, replicating and scaling these interventions have been challenging ( , ) . for example, home visits by health promotors have been shown to be effective, though outcomes vary greatly ( ) ( ) ( ) ( ) . progress to improve and scale evidence-based interventions to address poor childhood development has been slow, partly due to difficulty adapting interventions to diverse contexts and a lack of reporting on the implementation process conducted by researchers ( , , ) . the implementation process is complex and influenced by diverse factors. prior to implementing an innovative program in a new context, it is essential to determine if it can be effective and if adaptations are needed to enhance its potential impact. understanding the context helps to improve the fit of the innovation and implementation strategies, thus improving feasibility, acceptability, and sustainability ( , ) . implementation science proposes various theories, models, and frameworks (called tools henceforth) that can be used to improve diffusion of evidence-based interventions, adapt innovations to local contexts, better understand the implementation setting, and evaluate the implementation process ( , ( ) ( ) ( ) ( ) . however, few studies have been conducted that report on the use of the tools in global health settings and the resulting implementation process ( ) . the current study reports on the design and implementation of an mhealth intervention to improve child health in the amazon of peru. the study aims to provide insight into how an implementation science tool can be used to improve implementation and reporting of an evidence-based intervention in a global health setting. reporting on the implementation process is expected to show how implementation science tools can be used to foresee and address potential threats to successful implementation. this report addresses the need for critical reflections from practice-based settings to give insight into the barriers and facilitators of effective implementation in community-based settings ( ) . the current study utilizes an implementation science tool to systematically design, implement, monitor, adapt, and report on a community-based mhealth intervention for child health. the study utilizes a systematic method for choosing the most appropriate implementation science tool for the initiative. the tool is used to ensure the key components to effective implementation are considered and supported. the tool is also used to guide reporting of the implementation process to ensure all relevant activities are described here. the implementation process is categorized into four stages of implementation to display the challenges to implementation and the solutions that were provided. the analysis focuses on the use of information gathering to identify and improve an intervention and the implementation process, the implementation outcomes (fidelity, acceptability, adoption) and training for quality improvement. the analysis of the process and outcomes are reported in the results. this article describes the implementation of an intervention to improve the impact of community health agent (chas) programs on child health and development outcomes. the study took place in the northern amazon region of peru, in the department of loreto. in loreto, % of children under have anemia, % under have chronic malnutrition ( ), and infant mortality rate is deaths per , live births ( ) ( ) ( ) . delay in early childhood development was reported to be experienced by . % of children in the region ( ) ( ) . many of the illnesses can be mitigated by better practices in the household that lead to better sanitation practices, nutrition, and disease prevention ( , ) . however, caregiver's knowledge of practices to maintain a healthy family are limited as they transition from traditional practices to modern medicine ( ) ( ) ( ) ( ) . to improve health in the communities, the population must understand the causes, consequences, and treatments of poor nutrition and infection ( , ( ) ( ) ( ) . in peru, cha programs are widely used but greatly fragmented, with each level of government (national, regional, local) operating distinct programs. although they share the objective of improving maternal and child health, each program has a different system of operations, incentives, supervision, material, etc. while some communities have cha programs from all three levels of government, others have none. the cha programs often lack effective job aids to guide health education and data collection. to address these concerns, elementos, a peruvian research organization, used an implementation science tool to guide the process to identify a potential solution, co-create the design, and implement the innovation. the objective was to improve child health and development, by improving the capacity of cha programs to conduct health promotion and surveillance. during the pilot study, which is the focus of the current paper, the innovation was tested in a randomized control trial, in communities, with chas, serving children. it was provided to established cha programs for them to use during their regularly scheduled home visits with caregivers of children - years of age. the communities are only connected by rivers, approximately h by boat ( . h by speed boat) from the department capitol of iquitos. each community has a population between and , people. the communities have sporadic cell phone signal and at least h of electricity per day. the study protocol is described in westgard et al. ( ) . the first step for the implementation process was to choose the appropriate implementation science tool. a list of potential tools and their level of analysis is included in the online tool, dissemination and implementation models in health research and practice ( ) . the list is extensive with little information on how each tool can be used. previous knowledge of implementation science and its tools or additional reading is necessary for the list to be meaningful. the authors of the current study utilized their knowledge of implementation science, along with additional study of the various tools, to select the five most promising tools for the project. the five tools were: ( ) active implementation frameworks (aif) ( ); ( ) consolidated framework for implementation research (cfir) ( ); ( ) the exploration, preparation, implementation, sustainment framework (epis) ( ); ( ) interactive systems framework for dissemination and implementation (isf) ( ); and ( ) the theoretical domains framework (tdf) ( ) . to compare the tools and select the most appropriate for the project objectives, the theory, model, and framework comparison and selection tool (t-cast) was used ( , ) . the t-cast helped the authors systematically think about the strengths of each tool as they relate to each criterion that is important for successful implementation. the criteria were chosen from a list, based on the project's objectives. the tools were scored across the following criteria: the tool includes relevant constructs, provides a step-by-step approach for applying it, provides an explanation for how constructs influence implementation, focuses on relevant implementation outcomes, addresses a relevant analytic level, proposes testable hypotheses, and contributes to an evidence base. based on the score of each criterion, the authors were able to differentiate the most appropriate tool. the aif scored the highest in the evaluation with an average score of . . it was therefore selected to be the tool utilized to guide the implementation research and practice. the average score of each tool is in table . the aif is comprised of five distinct frameworks. the active implementation frameworks include; ( ) usable innovations, ( ) implementation stages, ( ) implementation drivers, ( ) implementation teams, ( ) improvement cycles. through the application of the aifs, users are guided through the stages and key activities for successful implementation, supporting careful consideration of the implementation setting and components of the intervention [see figure ; ( , ) ]. the stages of implementation include ( ) exploration of the local context and identification of innovations that can create positive change, ( ) installation of the capacity and resources needed to introduce, improve and sustain an innovation; ( ) initial implementation, during which performance data is used to rapidly improve both the innovation and implementation supports and strategies; and ( ) full implementation, where high quality implementation and program outcomes are realized and sustaining performance is a core focus. the current study reports on the process used to move from exploration to full implementation of the innovation in the communities. the implementation stages framework, of the aif, is used to organize the description of the evolving implementation process. the other components of the aif are presented within the various stages of implementation ( , ) . the implementation process during each stage is described with the aim of reporting on the key components addressed to accomplish implementation. key challenges that arose and the decisions that were taken to address them are presented. the results of the analysis of the implementation process is presented in the sections below, representing the four stages of implementation. the authors explore the critical components of project implementation as they relate to each stage of implementation. the exploration stage involved understanding the needs of the communities, identifying evidence-based practices that can address their needs, and determining the right fit between potential solutions and the local context. the work done during this stage improved the chances for success of the program by checking to ensure the local population wanted the intervention and believed it could work within their reality ( , ) . the research team conducted formative research in the communities to better understand the needs and priorities of the families. this involved the following activities: • interviewed regional health directors, program coordinators of municipalities, and community leaders to learn about their most pressing health issues, their priorities, and key barriers to progress. a common consensus among all stakeholders was the problem of child malnutrition and poor early childhood development. • conducted a social determinants study to better understand the drivers of poor child development in the communities. the study found that poor sanitation and nutritional practices were associated with an increase in developmental delay, and contact with a chas was associated with a decrease in developmental delay ( ) . • conducted a study to better understand the barriers to utilization of local health services for maternal and child health. the study identified key reasons why some mothers do not attend health-checkups for their child nor give micronutrient supplements. long wait times, closures, and a mistrust of health center personal were among the top reasons ( ) . • conducted a performance evaluation of chas in the communities. through observations of home visits by chas, the study found that many chas lacked the capacity and material to transmit the knowledge needed by the caregivers to conduct healthy maternal and child health practices ( ) . the studies and informant interviews identified poor health behaviors within the household as a key driver for child health and development. unhealthy behaviors that were taking place included: drinking untreated water; early cessation of breast feeding; poor diet; poor handwashing practices; unsanitary toilets; and low use of nutrient supplements and deworming medication ( , , ) . caregivers often lacked a good understanding of the causes, consequences, and treatment of common childhood illnesses. the local stakeholders agreed that health promotion and education were greatly needed, and that chas are a strong potential mechanism to provide that service. evidence from the studies suggested that improved performance of chas could improve the knowledge and practices of caregivers, and thus improve the child health and development outcomes. following the decision to focus on chas to address poor child development, additional research was conducted on the policies and operations of the cha programs in the region. the research team conducted interviews with representatives of cha programs at the national, regional, and local level. it was soon discovered that representatives at each level operated a distinct cha program. the programs share the objective of improving maternal and child health, however, each program has a different system of operations, incentives, supervision, recruitment, and material. while some communities have chas from all three levels of government, others have none. the research team studied the operations of the cha programs to better understand how an intervention could be designed and implemented to improve cha performance and impact. the research revealed several of the barriers expressed above: chas struggle to remember and transmit the knowledge needed to teach caregivers, they lack direction to choose which health messages should be shared at each home visit, and they lack material to help transmit the information. additionally, there is a lack of supervision and any control of the quality of the home visits. fidelity of the cha program suffers from a lack of a responsive supervisory system. the supervisors and representatives of the health centers and ministry of health have little way of determining if home visits are being conducted as intended. a landscape analysis was conducted to identify evidencebased interventions with the potential to improve cha performance in the amazonian communities. potential interventions were identified by reading scientific literature, expert interviews, and assessing the tools shared by chw central and the community health worker assessment and improvement matrix ( ) . a list of potential interventions was evaluated through a policy analysis to determine which best satisfied the selection criteria and showed most promise to be successful in the low-resource community setting. the search for potential interventions and the comparison process was dynamic, with new interventions being added and deleted over several months. the analysis revealed that multiple interventions had the potential to create positive impact in the cha setting. by utilizing mobile information and communication technology (ict), several of the intervention components could be combined into one innovative intervention. several studies have shown that mobile icts can improve the performance of chas in their ability to perform health promotion, collect and report timely information regarding family health, provide health services such as vaccines, and refer families to appropriate local health services ( , ( ) ( ) ( ) ( ) ( ) ( ) ( ) . additionally, when a mobile ict tools are used by a cha, the device can increase the confidence the caregivers have in the messages being transmitted and increase the confidence the chas have in their own work ( , , , , ( ) ( ) ( ) . through implementation science, innovations in mobile icts and strategies for child health and development can be extended to low resource settings to empower local actors and spread the benefits of advancements in technology ( , ) . the evidence-based interventions that showed promise to improve cha performance included: conducting surveillance of maternal and child health indicators with a mhealth tool, utilizing animated videos to deliver health messages to encourage behavior change, harnessing health behavior theory for the creation of health messages, and improve self-efficacy of chas by providing dynamic tool ( , , , - , - , , , ) . the intervention components were combined to create an innovative tool that supports cha programs. the innovative tool was titled, the child health education and surveillance tool application (the chest app). a video of the app can be viewed online ( ) . the chest app is an android-based application downloaded onto an electronic tablet. the chest app provides the steps for the cha to follow to conduct an effective home visit with caregivers. it was designed to improve the capacity of chas to transmit knowledge of healthy child-rearing practices and conduct disease surveillance. the chest app provides the following functions: ( ) collect child health indicators at the household level and upload the data to the server; ( ) select appropriate health messages to deliver during the home visit based on the age of the child; ( ) share animated videos, images, and statements that reinforce the health messages; ( ) calculate and display the anthropometric and nutritional status of children; and ( ) organize the case load of children and maintain schedules for home visits and health check-ups ( ) . a full description of the intervention can be seen in the study protocol that was published in ( ) . the theory of change for the intervention is displayed in appendix in supplementary material. once the chest app intervention was defined, it needed to be assessed to determine fit and feasibility for success in the local context. the intervention was assessed alongside the implementation setting to determine the probability of success. the exercise was supported by the hexagon exploration tool, of the aif. the hexagon exploration tool guides selection and evaluation of potential interventions for an implementation setting by promoting the consideration of key program and implementation site indicators [appendix in supplementary material ( )]. the chest app was assessed with the hexagon exploration tool by considering the six key components for successful implementation, as displayed in table . the exercise confirmed the potential for success of the intervention in the communities and promoted further consideration of important components of implementation. after exploration, efforts shifted to preparing for implementation. an implementation team at elementos was created to assist the actors in the field. the implementation team created the initial plan for implementation, prepared the local actors, and readied the tools, and material for the intervention. they conducted the training, monitoring, and quality improvement cycles. the team consisted of an implementation scientist, nurse, nutritionist, and anthropologist. before going to the field to prepare the local actors, the implementation team was trained on the use of the chest app, how to coach the chas, how to conduct an effective home visit with the tool, and how to identify and report challenges experienced by the chas. the chest app was developed throughout the months of the installation phase. a prototype was needed to show the local actors what the intervention would look like. however, the final form of the app was unknown at the beginning of development because the design needed the input from the end-users and the implementation team needed to further understand the workflow of the chas. the multidisciplinary team designed and created the app, the health messages and images, and animated videos to include in the app. to prepare for implementation, the team needed to determine where and when implementation would take place. to determine the location of the pilot, meetings were conducted with directors of cha programs at the levels of government (national, regional, and local) to present the chest app, document the system of operations of each cha program, and assess interest multiple studies have shown that mhealth tools can improve cha performance in similar low-resource community settings, including health education with videos and digital surveillance tools ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) usability previous studies and stake-holder interviews suggested that the technology could be used in the local context. additionally, the acceptability and usability of the chest app was confirmed through informal interviews with the local populations. the chas and local supervisors expressed their preferences for how the tool should be designed to fit the needs of their program. elementos had the resources to design and implement the chest app thanks to funding from grand challenges canada, saving brains grant ( ) . elementos created the capacity to conduct the project by hiring a multidisciplinary team of specialists; an implementation scientist, nutritionist, anthropologist, community psychologist, epidemiologist, nurse, communicator, and software engineer. together they developed the material (a guidebook of health messages and animated videos), the app, and the implementation protocol. the chest app was developed from open source code from opensrp, ( ) which includes code and forums for support. technical assistance and development was also provided by the unc chai core team ( ) capacity to implement the communities have established cha programs that have the capacity and interest to receive and integrate the chest app into their normal activities. the cha programs are supported by funding from their municipality, which includes pay for a program supervisor. additional implementation support was provided by elementos by providing an implementation team that visits the communities for continuous training and support for year. fit with current initiatives the tool was created to integrate within the established cha programs with minimal interruption of their current activities. the tool was expected to improve the ease and effectiveness of their current initiatives. multiple studies by the research team identified the need; reflected by the high rates of malnutrition, misunderstanding of health topics by caregivers, and poor performance of chas. in receiving the intervention. the cha program coordinated by the regional ministry of health has little formal structure or supervision. the chas conduct occasional campaigns (such as malaria prevention) and some conduct home visits as their own independent initiative. the chas did not receive incentives or regular supervision. their home visits are thus infrequent and unpredictable, making their program a poor fit to receive the chest app. at the national level, the cha program, "cuna mas, " conducts home visits in areas of extreme poverty ( ) . the chas receive a modest stipend for their work, close supervision, and continuous training. based on the structure of the program, they were an excellent fit to receive the chest app. the directors of the program were excited to receive the chest app. however, working with the national government provided difficult. significant turnover of staff in the ministry delayed communications and the formulation of a formal agreement. ultimately, the team was not able to formulate a formal agreement with the ministry in time to implement. they decided to implement the intervention with the local government cha programs. the district-level municipalities operate their own cha program, which vary by districts. all include home visits with children under , an incentive package (stipend or gift baskets), and a program supervision. the districts that were approached by the implementation team (amazon, indiana, las amazonas, fernando lores, and punchana) were enthusiastic to receive the chest app intervention. they established communications with each district and began learning about the specific activities of the cha program in each district. to determine when implementation should take place, the implementation process had to adapt to the political activities in each district. the districts had recently conducted elections, so their new representatives were adjusting programs and budgets, including the cha program. many were replacing chas and supervisors with their own contacts, freezing the program until a new budget could be released, or changing specific activities of the chas. therefore, installation and initial implementation had to be delayed for the programs to stabilize. continuous communication with the program coordinators and policy makers in the districts made it possible to continue to improve the intervention and implementation process during the delay. although this created delay for the start date of the chest app intervention, it helped troubleshoot and avoid several potential issues that may have arose. in each community, leaders were selected to head the initiative in their group. the program supervisor was included as a leader and the supervisor choose one cha to join the leadership. the supervisor was the primary person of contact in each community. the cha leader helped lead group discussions, trainings, and share the opinions of the chas to the implementation team. the cha leaders also played an important role in setting the general mood of the group. when the leader decided to accept the intervention and dedicate themselves to learning the new skills, the rest of the group followed even if they were initially hesitant. to prepare for the evaluation of the intervention and to adjust the program to the local setting, extensive data collection was conducted. the data provided a baseline for adaptive monitoring, evaluation, and learning. process data was collected on the procedures of the chas and supervisors to be able to monitor changes that may occur after implementation of the intervention. the indicators included the number of home visits conducted, number of children visited by each cha, the reporting procedures of the cha and supervisors, the health indicators they reported, time delay for the indicators to be reported to the health authorities, and acceptability expressed by all stake-holders. this information allowed the implementation team to later assess acceptability, adoption, and fidelity of the new intervention. this helped determine if doses and quality are changing overtime and identify opportunities to adapt the intervention to achieve greater effectiveness. an assessment form was created for the implementation team to measure acceptability of the intervention. acceptability was measured by interviewing the chas, supervisors, and caregivers. during each visit to the communities by the implementation team, an interview was conducted with one of each actor. the implementation team filled out the acceptability assessment form with each. the questions pertained to what they liked and disliked about the intervention and suggestions to improve it. the chest app includes a method to track adoption of the tool into the cha program. the cha records child health indicators with the app during the home visits. the supervisor connects the tablet to a wifi hotspot and uploads the data from the app to the server. the server can be accessed by the supervisor, health center personal, municipality, and implementation team. they can see if the chas are conducting the appropriate number of home visits with the tool and collecting the required information. the supervisor was trained to upload the data from the tablets and interpret the data to determine if the frequency of home visits by the chas matches what is expected of them. in this way, adoption can be tracked by all parties. fidelity of the intervention was accessed by the supervisor through observations of home visits by the chas. an assessment form was created by the implementation team and supervisors to assess fidelity. the fidelity assessment form included the steps needed for a quality home visit and scoring system for each step. the supervisor kept a record of fidelity scores for each cha and scheduled trainings with the chas based on their scores. data was collected on the intermediary/mediator variables to test the theory of change of the intervention. the information was gathered from household surveys conducted by the implementation team. the intermediary variables included; performance of chas, knowledge of caregivers, childrearing practices, and use of health services. testing for change along each step of the theory of change helped to determine opportunities to adjust and improve the intervention or implementation process. data was also collected on the primary outcome indicators; hemoglobin levels, anthropometrics to estimate malnutrition, and early childhood development scores. the data was collected to determine the effective size of the intervention and report the implementation outcomes. the chas and program supervisors were trained on how to use the chest app to support their work. the training was designed to teach them how to operate the tablet and application, how to use the tool to improve the interaction with caregivers during home visits, and how to use the information gathered by the tablet to improve their impact. the training lasted days. the first day was focused solely on the use of the app. the implementation team sat down with groups of chas to show them how to use the app. then, the chas spent the day practicing, working in groups to help each other resolve problems and remember the steps. the cha leaders were the first to answer questions from the others before a member of the implementation team stepped in to help. the second day of training included simulations of home visits. one cha conducted the home visits with the chest app while another cha pretended to be a caregiver in her home. the implementation team conducted several simulations for others to watch to show how the home visits could be more dynamic with the use of the tool. the chas mimicked the behaviors of the implementation team and greatly improved how they conduct home visits. the third day of training was one-on-one with a member of the implementation team and each cha. the member of the implementation team accompanied the cha on a home visit with a caregiver in their community. the implementation team member gave advice to the cha after the home visit on how it could be improved to better transmit the knowledge displayed in the app. at times, the supervisor joined home visits with the cha and implementation team member. at this time, the supervisor was trained in how to assess fidelity with the fidelity assessment form. before and after the home visits, the implementation team showed the supervisor how to score the home visit on the fidelity assessment form, creating a common standard for a quality home visit. through the conversations and observations of quality home visits, the supervisor learned how the intervention is intended to be delivered. the chas were initially nervous to use the new tool, albeit excited by the novelty. many of the chas had never used a touchscreen device before. at the end of the -day training, all chas were able to conduct a home visit on their own with the chest app. however, ∼ % of the cha needed additional practice with the app to become faster and more comfortable. a total of chas were training, in groups with an average size of chas. during the workshops, the implementation team worked together with the chas and supervisors to identify opportunities to further adapt the chest app to match their needs. the team conducted human centered design exercises to surface challenges they anticipated from using the new tool and elicit suggestions for how it can be improved. the research team noted the difficulties and suggestions that were expressed and took them back to the developers so they could make quick, incremental improvements. for example, the language used in the app needed to be updated to include more localized terminology. also, the images used to indicate if a child has chronic or acute malnutrition were removed because they caused confusion. the option in the app to record the child's id number was made optional because we learned some children do not have a government-issued id. the initial implementation stage began by distributing the tablets with the chest app to the chas and supporting the integration of the new tool into their normal activities. a total of chas in five communities received a tablet. the chas immediately began using the tool to help choose which child to visit and guide them through their home visits. they collected data on child health indicators while in the homes and coordinated with the implementation team to upload the data to the server. this stage of implementation was about testing and improving the functionality of the chest app and the implementation process. the implementation team continued to work with the chas to conduct improvement cycles on the intervention, further train the chas in the use of the chest app, support the program supervisors on downloading the data and making data-based decisions, and communicate with the authorities of the municipalities to share the advancements and value of the chest app for their program. the cha leaders agreed to meet with the chas that were struggling to use the app comfortably. initially, all chas in each community met - times per week to practice using the app. the cha leaders and supervisor organized the meetings and assisted those that needed help. the meetings became less frequent as they mastered the new tool. members of the implementation team from elementos visited each community bi-weekly during the first months following implementation, and then once a month thereafter during the first year of implementation. the cha leaders played an important role during the meetings with the implementation team. they voiced the concerns they had about the tool, requested changes, and gave feedback about the general mood of the chas in using the tool. the mood was very positive, as the chas liked the new tool and the prestige it gave them when they visited the homes. the families that receive visits from a cha with the chest app were the ultimate recipients of the intervention. their experience with the chas changed due to the new tool. they now have the opportunity to see the health status of their child displayed in the app with stop-light indicators (red or blue), view images and videos that explain topics of health and development, and hear the cha give guided messages to promote behavior change. to measure the effectiveness of the chest app intervention, implementation outcomes were tracked and evaluated throughout implementation. the implementation outcomes represented how well the intervention was delivered and received. the outcomes that were tracked included adoption, fidelity, and acceptability. the implementation team and local authorities used the chest app as a decision-support data system to assess adoption. adoption was accessed by analyzing the number of children the chas visited and the number of home visits conducted per month. the number of home visits per month was consistently rising or staying consistent (depending on the community), over the first months of implementation. at month of implementation, the implementation team noticed a sharp drop in number of children visited with the chest app in a community. the change signaled a reduction in adoption of the intervention and the need for the implementation team to visit the community to troubleshoot the situation. the team found that several of the chas were released from the cha program due to budget cuts. the team worked with the supervisor and municipality to adapt their program to work with fewer chas, prioritizing the children with poor nutrition status, and visiting the healthy children less frequently. this allowed the chas that remained to continue to visit all the children in the program. identifying the problem was possible due to the integrated mechanism in the chest app to monitor use of the tool. the fidelity assessments conducted by the supervisors provided the information needed for targeted training and quality improvements. by observing random home visits with the chas, the supervisor identified which chas were having problems conducting the home visits as intended. the chas that scored poorly on parts of the fidelity assessment form received support from the supervisor and cha leader to improve their performance on those specific steps of the home visit. the assessment allowed the supervisor to identify which aspects of the home visit were not being delivered with fidelity and focus on those aspects during the on-going training. the information gathered during the fidelity assessments also improved quality improvement efforts. the supervisor and implementation team found that many chas were having trouble remembering to gather child health indicators during the visit. they conducted the educational portion without conducting the surveillance portion of the home visit. with this information, the team made adjustments to the chest app, making it required to click through the surveillance section of the app before advancing to the educational section. this proved to be effective at ensuring the surveillance was conducted and that the intervention maintained high fidelity. the implementation team evaluated acceptability of the intervention immediately after implementation and during the following months. the information recorded in the acceptability assessment form provided valuable information to continuously improve the quality of the intervention. overall acceptability of the intervention increased over time. most of the suggestions for improvement occurred during the first weeks of implementation. after adjusting the program based on their suggestions, acceptance, and positive feedback were expressed by all chas and supervisors. evaluation of acceptability by the caregivers revealed that the families found the home visits to be more appealing with the chest app. caregivers, children, and other family members became more interested and attentive during the home visit than before. the chest app made the caregivers feel more confident in the information the cha presented and could more easily understand the messages. the suggestions gathered during the acceptability assessments provided opportunities to improve the quality of the intervention. the team synthesized the requests for changes and adjusted the chest app when appropriate. a change was made to the app because some communities were not able to upload the data from the tablet to the server due to insecure assess to cell phone data. the developers added to the chest app the ability to transfer data directly from the tablet to a local computer with a cable. by adding a direct transfer function, the program coordinators were able to extract the data as they needed, without cell phone signal or direct assistance from the implementation team. additionally, a function to erase a case/child from the caseload was added to the chest app interface. the chas expressed the need for the function due to children frequently aging out of their program or moving away. the chas needed the ability to delete cases without the assistance of the implementation team. the solution seemed obvious once the chas explained the need, however, the problem did not occur to the design team until then. once a high level of acceptability, adoption, and fidelity were reached and maintained, the program began the full implementation stage. after months of implementation support, the intervention was operating with % adoption across all active cha programs involved in the pilot. when a cha conducted a home visit, they used their chest app. also, acceptability and fidelity were high, and supervisors continued fidelity checks and quality improvement efforts without outside support. one community canceled their cha program, and thus stopped using the chest app. the municipality canceled support for the program due to budget restrictions. they anticipate re-activating the program in the coming months. this is an important detail when assessing sustainability of the intervention when implemented at the district level. after months, the research team ended their monthly visits to the communities. the chas and supervisors did not need ongoing training outside of their own local support. elementos was able to scale-back resources (staff and travel expenses) invested to support the chest app program on the ground. however, elementos was not yet able to stop all involvement in the programs. the supervisors in three of the communities were not yet able to upload, download, organize, and interpret the data obtained with the chest app. a member of the implementation team continued communication with the supervisors of each program to assist with the task of data management. on a monthly basis, the supervisors connected the tablets to a cell phone hot spot or directly to a computer to upload the data to the server. they signaled to the team at elementos that new data was uploaded. at elementos, the data was then downloaded and organized in a user-friendly report and sent back to the supervisors. the supervisors then shared the report with the municipality and local health post. to determine if the chest app program (intervention and implementation process) is cost-effective and should be sustained and scaled, an evaluation of the process and impact is needed. follow-up surveys were planned to be conducted and compared to baseline to determine impact after months of operation. the follow-up surveys were delayed due to the covid- pandemic, still pending at the time of writing this manuscript. assessment will include measurement of the performance of the chas, knowledge evaluations, and surveys of household practices to measure the impact of the chest app on chas and caregivers. the improvements in knowledge and practices of caregivers are expected to reduce anemia, reduce chronic malnutrition, and increase early childhood development scores. sustainability of the intervention has been measured by tracking the adoption and fidelity scores over time. both adoption and fidelity were high during the first phase of implementation and has maintained after external support from the research team was withdrawn. sustainability will be tracked for an additional year to ensure the intervention can be further maintained before scaling. an important component to assess sustainability of the intervention is the cost. the primary expense of the chest app intervention is the cost of the tablets. for the pilot project, the tablets were provided by elementos. the municipalities committed to buying new tablets for the intervention to scale to additional communities in their district and to replace old tablets as they become unusable. their commitment to dedicate sufficient budget to the cha program to buy tablets is necessary for the intervention to be sustained. therefore, sustainability of the program is determined by state actors, and not outside support or funding. the program is expected to be continued as long as child health and development remain a top priority. the cha program with the chest app is expected to be scaled to additional districts and regions of peru once sustainability is confirmed. the intervention and implementation process were created so they can be replicated and expanded without a decrease in impact (voltage drop) ( ) . training chas in new communities can be done with trainers of relatively low expertise. the cha leaders of past intervention communities can take a lead role in training new communities. the data support can be conducted by a central supporting agency, such as the regional ministry of health or a non-governmental organization. each municipality can manage the supervision and evaluation of adoption and fidelity. therefore, the program can replicate with little additional cost and demand for outside support. the educational material that is included in the chest app was created to match the reality of the amazon region. the food sources, infections, sanitation challenges, etc., matches the experience of amazonian communities. to scale the program, the educational material will need to be adapted to match the diverse contexts in peru, such as the high mountains and coastal plains, and include messages in local languages. with the chest app, modifying the material to match the local reality is feasible and economical. once the material is developed and translated, it can be uploaded to the tablets remotely. additionally, new educational material can be added to the chest app as the program advances or to match diverse health challenges that arise. the updates can be distributed without purchase or deliver of new material, only adjust the app's code and connect the tablets. this is the first study to examine the use of the aif to analyze and report on the implementation process of a global health intervention for child health and development. by reporting on the process, the reader can learn about the implementation context in the amazon of peru and how the tool can be applied to analytically assess key components of implementation. the aif guided the research team to focus on important components of implementation, thus further dedicating resources and analytical consideration during implementation to increase probability of successful of the intervention. the key components of the implementation process included information gathering to conduct improvement cycles, the implementation outcomes (fidelity, acceptability, adoption) to monitor progress and sustainability, and training for continuous quality improvement. analyzing the various components gave great insight into the behavior of the participants and local system. understanding the perspectives and behaviors of the providers, end users, and program coordinators on the ground is a valuable part of the implementation science approach, and essential to create longlasting behavior change ( ) . the chest app innovation and the implementation strategy went through several adaptations to better fit with the local context. the implementation science approach was extremely beneficial to guide the multiple design iterations and rapid-cycle problem solving. the results were greatly improved promotional material, app design, and implementation process. it is important that researchers specify and report on the process used to design, adapt, and implement an intervention ( ) . details of the implementation process are needed for others to evaluate, replicate, improve, and scale the intervention ( , ) . this study reports on the implementation process and key components that were assessed during the design and implementation of the intervention. the consolidated advice on reporting ecd guidelines (c.a.r.e guidelines) describe which implementation components should be reporting when conducting implementation research on early childhood development interventions ( ) . this study reports on those components, including previous evidence of intervention, rational, context of implementation, description of recipients, adaptations that occurred, personnel, methods to assess fidelity, and others. an additional study will be published following the final evaluation of the chest app intervention that reports on process and clinical. the study contributes to the knowledge base by demonstrating how an implementation tool can be applied in practice in global health. the scientific community has indicated the need for greater reporting on the delivery of public health interventions, especially those in global health ( ) ( ) ( ) ( ) ( ) ( ) . activities conducted during the design, implementation, and evaluation of an intervention should be reported so the scientific community can learn what works and what does not. this study provides information on the implementation of a child health and development intervention in a community-based setting. the chest app intervention was analyzed and reported through the lens of the aif. the aif assisted the research team to consider components of implementation that are often neglected, such as choosing the right solutions that fits local context, information gathering for data driven decision making and adaptations, and monitoring implementation outcomes. the analysis and activities that took place during each stage of implementation are described so others implementing a similar intervention can reflect on the experience and improve their own implementation process. this report contributes to the pool of knowledge needed to improve impact and scale of global health, communitybased interventions. publicly available datasets were analyzed in this study. this data can be found here: https://figshare.com/articles/indiana_ resultados_xlsx/ / . cw was involved in the design, implementation, data collection, data analysis, and writing of the manuscript. wf was involved in the analysis and writing of the manuscript. all authors contributed to the article and approved the submitted version. the supplementary material for this article can be found online at: https://www.frontiersin.org/articles/ . /fpubh. . /full#supplementary-material diffusion theory and knowledge dissemination, utilization, and integration in public health advancing a conceptual model of evidence-based practice implementation in public service sectors implementing innovations in global women's, children's, and adolescents' health: realizing the potential for implementation science outcomes for implementation research: conceptual distinctions, measurement challenges, and research agenda implementation strategies: recommendations for specifying and reporting what implementation evidence matters: scaling-up nurturing interventions that promote early childhood development strategies to avoid the loss of developmental potential in more than million children in the developing world strategies for reducing inequalities and improving developmental outcomes for young children in low-income and middle-income countries comprehensive review of the evidence regarding the effectiveness of community-based primary health care in improving maternal, neonatal and child health: strategies used by effective projects lay health workers in primary and community health care for maternal and child health and the management of infectious diseases community-based intervention packages for reducing maternal and neonatal morbidity and mortality and improving neonatal outcomes effectiveness of community health workers delivering preventive interventions for maternal and child health in low-and middleincome countries: a systematic review review of implementation processes for integrated nutrition and psychosocial stimulation interventions barriers and facilitators to implementing an evidence-based woman-focused intervention in south african health services fostering implementation of health services research findings into practice: a consolidated framework for advancing implementation science making sense of implementation theories, models and frameworks beyond "implementation strategies": classifying the full range of strategies used in implementation science and practice implementation research: a synthesis of the literature situación epidemiológica de las enfermedades diarreicas agudas situación epidemiológica de las infecciones respiratorias agudas (ira), neumonías sob (asma) en el perú. lima: ministerio de salud del peru developmental delay in the amazon: the social determinants and prevalence among rural communities in peru a systematic approach to behavior change interventions for the water and sanitation sector in developing countries: a conceptual model, a review, and a guideline health promotion by social cognitive means health, healthcare access, and use of traditional versus modern medicine in remote peruvian amazon communities: a descriptive study of knowledge, attitudes, and practices health service utilization, perspectives, and health-seeking behavior for maternal and child health services in the amazon of peru, a mixed-methods study impact of health education on soil-transmitted helminth infections in schoolchildren of the peruvian amazon: a cluster-randomized controlled trial the role of health education and sanitation in the control of helminth infections anemia infantil: retos y oportunidades al using of health belief model to promote preventive behaviors against iron deficiency anemia among pregnant women the health belief model mhealth tool to improve community health agent performance for child development: study protocol for a cluster-randomised controlled trial in peru dissemination & implementation models in health research and practice bridging the gap between prevention research and practice: the interactive systems framework for dissemination and implementation developing theory-informed behaviour change interventions to implement evidence into practice: a systematic approach using the theoretical domains framework t-cast: an implementation theory comparison and selection tool framework comparison and selection tool (t-cast) implementation science: international encyclopedia of the social & behavioral sciences: second edition module : implementation stages | nirn. national implementation research network an overview of active implementation frameworks | nirn. national implementation research network the stages of implementation completion for evidence-based practice: protocol for a mixed methods study observational measure of implementation progress in community based settings: the stages of implementation completion (sic) performance evaluation of community health workers: case study in the amazon of peru anemia and malnutrition in indigenous children and adolescents of the peruvian amazon in a context of lead exposure: a cross-sectional study health and nutrition of indigenous and nonindigenous children in the peruvian amazon community health worker assessment and improvement matrix (chw aim): a toolkit for improving community health worker programs and services i am not telling. the mobile is telling": factors influencing the outcomes of a community health worker mhealth intervention in india app-supported promotion of child growth and development by community health workers in kenya: feasibility and acceptability study midwives' cell phone use and health knowledge in rural communities preventing maternal and newborn deaths globally: using innovation and science to address challenges in implementing life-saving interventions mobile health (mhealth) approaches and lessons for increased performance and retention of community health workers in lowand middle-income countries: a review impact of m-health application used by community health volunteers on improving utilisation of maternal, new-born and child health care services in a rural area of uttar pradesh, india the effect of a community health worker utilized mobile health application on maternal health knowledge and behavior: a quasi-experimental study the aceh besar midwives with mobile phones project: design and evaluation perspectives using the information and communication technologies for healthcare development model adapting a health video library for use in afghanistan: provider-level acceptability and lessons for strengthening operational feasibility improving diabetes health literacy by animation impact of a video-based health education intervention on soil-transmitted helminth infections in chinese schoolchildren | infontd avilable online at saving brains -unlocking the potential for development available online at available online at active implementation hub. national implementation research network. frank porter graham child development institute available online at the dynamic sustainability framework: addressing the paradox of sustainment amid ongoing change designing a behavioral intervention using the com-b model and the theoretical domains framework to promote gas stove use in rural guatemala: a formative research study appendix | the hexagon exploration tool. key: cord- -z k cujs authors: liu, yali; avello, maria title: status of the research in fitness apps: a bibliometric analysis date: - - journal: nan doi: . /j.tele. . sha: doc_id: cord_uid: z k cujs fitness applications have undergone considerable development in the last few years and becoming popular and significant in both academic and practical areas. however, contributions to the systematic mapping of this field continue to be lacking. this paper constitutes the first bibliometric study in this field to better understand the current state of research. we examined records from databases scopus and web of science (core collection) using several bibliometric analysis methods. all the records on this emerging topic were published between and . we processed these records using statistical analysis and science mapping. the bibliometric analysis included the year of publication, journal name, citation, author, country, and particularly, research methodology. additionally, we used the vosviewer software to perform bibliometric mapping of co-authorship, co-citation of authors, and co-occurrence of keywords. this field of study, it was found, is currently in its precursor stage, contributing primarily to the fields of medicine, computer science, and health sciences. the united states appeared to have made the largest contribution to this field. however, author productivity, number of citations, and number of core journals all indicated a high degree of fragmentation of research in this filed. remarkably, scientific research in this area is expected to progress tremendously over time. overall, this study provides basic data and research classifications for the initial phase of research and research direction for future research in this area. with the global outbreak of the covid- pandemic in , almost every country is facing problems concerning the shortage of medical and healthcare resources, and people have become more aware of the importance of following a healthy lifestyle and incorporating physical exercise into their daily lives. as the most downloaded type of mobile health applications (mhealth apps), fitness apps can help people manage their nutritional intake, assist their participation in fitness and physical activities, and promote a healthy lifestyle. therefore, these apps are gradually occupying the commercial mobile app market (beldad & hegner, ) . nowadays, fitness apps are rapidly developing in the commercial application market and are attracting the attention of academia (beldad & hegner, ) . numerous studies have implemented empirical protocols to verify the results of using fitness apps for improving the level of physical activity and/or diet in users (schoeppe et al., ) . however, from the academic side, it is still a novel and young area of research. as a diverse field of research that is related to an emerging phenomenon, and with the integration of new technologies, the research available on fitness apps is still scarce. both empirical research and theoretical orientation reviews, mostly focus on summarizing the functions and features of fitness apps and user perspectives. as a result, there appears to be a lack of more macro and objective quantitative research in this field. and the various types of literature are not as substantial or abundant compared to other mature areas of research. it is necessary to carry out a bibliometric study to know the main empirical and theoretical orientations in this case. the data obtained from the bibliometric analysis will be essential to assess the intensity and orientation of new lines of research (bartoli & medvet, ) . moreover, it is essential to classify the existing research in the research field to track the research progress and research trends in the field (gaviria-marin et al., ) . bibliometrics study can achieve this objective. it helps display past academic research activities and achievements visually. to our knowledge, there is no bibliometric study in the field of fitness app research, even though this type of literature has been used widely in other fields in recent years (zanjirchi et al., ) . bibliometrics can supplement existing experiments and review studies, help researchers identify hidden research lines, hot issues, and research methods in the field, and reduce the problems of neglecting certain excellent articles due to the deviation of researchers' subjective judgments (zanjirchi et al., ; veloutsou & mafe, ) . therefore, this study offers a bibliometric study of the advancements in research on the mobile-fitness app. it is based on data from a bibliometric analysis. it seeks to assess the intensity and research topics dominant in the scientific community when it comes to this emerging phenomenon, focusing explicitly on the fitness segment of mhealth. this study also aims to provide relevant data and bibliometric indicators for the initial stage of fitness application research and provide primary data for advancing future research in this field. the data used in this study is obtained from two leading databases for scientific research: scopus and web of science. the research is organized as follows. first, a research background is provided. second, the research methods and the sources of research data are outlined. third, the results are presented and discussed. finally, the main conclusions, limitations, and further opportunities for research are stated. nowadays, mobile apps pertain to a wide range of topics and areas of users' personal and social lives and fulfill various purposes. the use of advanced medical information systems and telematics applications is one of them, which has resulted in the increased availability of medical services at lower overall costs (kao et al., ) . medical and sanitary institutions have begun to appreciate the potential of mhealth apps for communication with patients as well as for the utilization of mobile devices that are specifically designed to monitor specific biomedical data. mhealth is defined as the provision of medical care and health-related services through mobile communication devices that enable user-interaction capability (cummiskey, ; lupton, ) . "mobile health (mhealth) has become an essential field for disease management, assessment of healthy behaviors, and for interventions on healthy behaviors" (mas et al., , p. ). there are two main areas of implementation of mhealth apps: in professional medical practices (both on the side of doctors and patients; e.g., skyscape, mysugr), and selfmonitoring of healthy habits (e.g., myfitnesspal). the first area has a field of an app exclusively in the healthcare field, involving the relationships between doctors and their patients. the second area represents fitness apps, which is the subject of this study, is concerned with the personal monitoring of the activities of individuals within the framework of adopting healthy lifestyles or disease prevention habits, and this category is often implemented through commercial apps that are developed without the supervision of medical administrations. the term "fitness" has a wide semantic field: on the one hand, it refers to the practice of physical exercise to obtain or maintain good body shape and composition; on the other hand, more generally, it refers to a good state of vitality and physical well-being (corbin et al., ) . since the s, academic as well as medical attention to health-related physical fitness (hrpf) has increased considerably. fitness is understood within the hrpf framework, which is defined as a set of people's abilities to perform certain physical activities, their energy level to perform daily tasks, and their capacity to reduce the risk of diseases related to sedentarism (cheng & chen, ). the who warns of the development of non-communicable diseases, the pathologies of which are associated with unhealthy lifestyles and diets, as these diseases currently constitute a serious cause of death worldwide (who, ) . in particular, the who has established a set of minimum criteria for physical activity for different age groups as well as balanced dietary patterns to maintain optimal health conditions such that people can achieve a reduction in risk factors for non-communicable diseases, including cancer, cardiovascular ailments, and diabetes. the high rate of obesity is one of the most worrying factors for health globally, particularly in developed countries, but also in emerging countries, with a drastic growth among children (anderson et al., ) . for this reason, the who recommends avoiding a sedentary lifestyle and following balanced diets for all age groups. interventions for population self-management, based on changes in lifestyle, are effective in reducing risk factors and the incidence of non-communicable diseases (burke et al., ) . the use of applications on mobile devices has become a key factor in helping and advising people on the adoption of healthy lifestyles in the st century. although some clinicians lack confidence in the protocols and recommendations of fitness apps, these fitness apps have a great potential to be effective due to their ability to educate a large portion of the population on healthy habits at a low operating cost (blackman et al., ) . the methodology used in this research work is depicted in figure . it consists of four steps. we identified "fitness app" as the field for this study with the aim of finding as many articles as possible on fitness-related apps closer to health behaviors than to a professional medical approach. however, in the compilation of the final set of articles, we also included those that, without being strictly articles on fitness apps, contained relevant keywords linked to the subject of study, even though they were papers dealing with other types of mhealth apps. the data was obtained from two databases: scopus and web of science core collection (wos). these two databases are currently the leading sources for indexing scientific articles and allow for the collection of data from a large number of journals (adriaanse & rensleigh, ) . scopus owns high-quality and reliable coverage and complete data for each reference. it is the largest abstract and citation database for peer-review literature (zanjirchi et al., ) . the wos is also recognized by the scientific community as a digital bibliometric platform with high-quality literature, which can also provide metadata for bibliometric analysis and covers a wide range of disciplines (gaviria-marin et al., ; hew, ) . the combination of more than one database for mining scientific data can provide more robust results for the bibliometric analysis (de oliveira et al., ) even though it makes it necessary to integrate the information from both databases with different structures and review the articles one by one. mining the data is the most basic and crucial step to obtain valuable and credible research results. the search for this study was conducted in april and included all relevant publications until the end of december . the study focused on scientific research related to personal care applications of fitness, using the keywords "fitness app" and its plural form in english for searching through titles, abstracts, keywords, or topics. our search criteria are detailed in table . these two keywords represent the technological concept (app) associated with the lifestyle (fitness), whose specific relationship makes the object of the present investigation. no more keywords related to the fitness industry were used (e.g., weight loss/running, dieting) since we wanted to examine which other specific categories were reviewed under the category of fitness apps in general. our search does not have a low-time frame limit, and the aim is to learn about the starting time of research in this field. title-abs-key ("fitness" and "app" or "apps") publication period* until article, review, and conference paper article, review, proceedings paper, and meeting abstract language english english *no low time frame limit was set, but articles published before , while containing relevant keywords, were seen not to be relevant to the field. after searching in the two databases separately, we performed a manual review of the titles and abstracts (also full text if necessary), excluding articles whose topics did not meet the criteria of the study, and subsequently removing duplicate literature. when the same article appeared in both databases, we opted to keep the references in scopus because scopus provides broader bibliographic information than wos. the search returned records. we decided to keep the conference papers and meeting abstracts due to the youth and relative novelty of the field of study. after filtering out the irrelevant and incomplete records, we ended up with a total sample of records. the records were then analyzed using bibliometric analysis. bibliometrics is "the quantitative study of physical published units, or bibliographic units, or of the surrogates for either" (broadus, , p. ) . the bibliometric analysis allows us to understand the intensity of the research available on a topic as well as the different research fields explored by the academic community. the variables analyzed for the bibliometric study were the year of publication, author, country of institutional origin, language of publication, type of document, journal, number of citations, area of research, topics analyzed, and the research method used. additionally, bibliometric mapping was also conducted. the construction of bibliometric maps has always received attention in bibliometric studies (van eck & waltman, ) . we used vosviewer software to present the relation of co-citation, co-occurrence of keywords, etc. finally, we summarized the current research hotspots and trends in this field, based on the content of these articles and the information presented by the keywords of their authors, to inform and inspire further studies. the first article on fitness apps was published in , and until , the intensity of research was very low. . % of the articles are published from onwards. in , there was a significant increase in the number of publications, doubling the number of . these results represent a price's index of . % until the end of . price's index (price, ) refers to the percentage of references less than five-year-old. as the price index's value is relatively high, this area is considered to be novel and dynamic. price 's law ( ) proposes that the development of the scientific field follows an exponential growth, which doubles in size every - years. the development of the scientific field goes through four stages: the precursor stage, the exponential growth stage, the consolidation of the body knowledge stage, and the decrease in the production stage. as shown in figure , publications in related fields underwent a growth process from to . a linear mathematical adjustment of the measured values provided us with a correlation coefficient r = . , which implies that . % of variance failed to explain this fitting. in contrast, a mathematical adjustment to the exponential curve provides a coefficient r = . , indicating an unexplained variance of . %. this reveals that the data analyzed is more consistent with a linear fitting rather than an exponential one. while the third stage of growth also showed a linear trend, the first contribution in this field was produced in , and the exponential growth trend stage was not detected. so, research in this field is still in its precursor stage. additionally, the number of publications in - was close to % of the total, exhibiting rapid growth. although there was a small decline in compared to , we expect the scientific production in this field to enter the exponential growth stage in the coming years. articles on fitness apps are published in a wide range of journals, from medical and health-related ones to computer science-related ones. out of the records, were published in academic journals, and were published as conference proceedings. the publication source also indicates a great dispersion: there were journals and different conference proceedings in total. among all the relevant journals, eight journals have published three or more articles. however, only nine conference proceedings had more than one article. compared to other fields of study, this number seems very small and indicates a low level of source concentration. table presents the field's most productive and influential journals, and table outlines the nine most productive conference proceedings. besides, % of the publications were from conference proceedings. the first and second positions by the number of publications came from the field of computer science. the high proportion may be explained by the fact that, although the importance of conference proceedings in areas such as the natural sciences is decreasing, they still play an important role in computer science, with nearly % of citations also distributed in the proceedings (michels & fu, ; lisée et al., ) . it also shows the importance of the development of fitness apps in the domain of computer applications. braford's law ( ) is a tool used in bibliometric studies to evaluate the concentration/dispersion factor of a set of publications. in essence, it allows the determination of the most productive nucleus in a particular subject. it postulates the existence of a small nucleus of journals that address the topic more broadly as well as a vast peripheral region that is divided into several zones with journals that have a decreasing representation in the subject studied (alvarado, ) . the number of journals in the core and the number in the successive zones are in a ratio of : n: n . therefore, journals included in the core have a comparatively high concentration of publication, while those involved in the surrounding areas are increasingly dispersed. thus, we can see that there is an unequal distribution of articles in the journals. a large number of articles are found in a small number of journals. as shown in figure and table , within the core of the ring, only journals contained one-third of all published articles ( records). zone comprises journals, and zone comprises journals. zona contains a much smaller number of journals than the theoretical value ( ). this result suggests the innovative and youthful nature of the field under study, which has not been considered in depth by many journals. the number of citations is an important indicator of the influence and the attention presented by the scientific community. according to the results shown in table , a total of articles received more than citations-all from academic journals. this number is relatively low compared to other more mature fields of research. the most cited article ( citations a total of , authors have contributed to this field. the average number of authors per article was . , which indicates the trend towards multi-author contributions in the field and a wide dispersion of research. table summarizes the first authors in the list, with more than two contributions. in those cases where the information was not available at scopus, we used the information provided by wos. the most productive authors in terms of the number of articles published are oyibo k. and vassileva j., both from the university of saskatchewan (canada), with contributions. third and fourth-ranked gay v. and leijdekkers p. are co-authors. in the scope of the subject of our study, they co-authored a total of six articles. the work of the most productive authors does not attract the highest number of citations. the author, with the highest number of citations in the fitness apps field, is west j.h. his six articles have garnered a total of citations. three of them are ranked in the top ten most influential papers in table . they were all published in the journal with the most contributions in the field, journal of medical internet research. the author with the highest h-index ( ) is salmon j., from deakin university, whose research pertains to the fields of medicine, health professions, and nursing. however, the total number of citations for his three articles was only . no other author had an h-index above . the high inconsistency in the number of citations, the number of author contributions, and the h-index show that no scholar or team of scholars has yet had a decisive influence on the field, which is also related to the fact that the field is still in the precursor stage of research. additionally, the authors in table are not widely dispersed in terms of institutional affiliation, with several authors (and close rankings) being from the same institution. this suggests that a high proportion of the top productive authors are co-authors, as evidenced in figure . it highlights that only four authors did not co-author papers with others. the remaining authors make up the remaining nine clusters. moreover, members in each group usually come from the same institutions or countries, with less cross-national/interregional cooperation. the authors' productivity data are much lower than the values suggested by lotka's law ( ) . this law states that the number of authors making n contributions in a given period is approximately equal to the number of authors who make /n contributions. generally, the application of lotka's law gives the theoretical result that about % of authors make only one contribution in their field of study. in the field of research on fitness applications, the value of lotka's law is . %. this confirms the huge dispersion of the field, which can be explained either by the novelty of the phenomenon or by a multidisciplinary approach. additionally, the analysis of co-citation of authors shows the structure and connections of the co-cited authors, i.e., "which authors are cited together more frequently" (gaviria-marin et al., , p. ) . figure shows the results of the analysis conducted using vosviewer, and the number of citations for each author is indicated by the size of the colored dot. authors with more than citations were clustered in five groups. some of these authors did not contribute directly to our field. however, their articles are frequently cited by other authors in the fitness app research field. authors in cluster mainly tend to focus on research in the areas of social sciences, business, management and accounting, and mathematics. sub-topics of interest to them include behavior change, physical activities, etc. authors in cluster primarily devote their research to the field of biochemistry, genetics and molecular biology, and health professions. physical and health education is also one of the sub-topics they are interested in. in cluster , the main research interests include psychology, and besides, the authors have contributed to the areas of computer science, nursing, and decision making. the main research interests of the authors of cluster lie in the arts and humanities, social sciences, computer science, and psychology. they have also undertaken certain interpretative explorations of technological acceptance. cluster consisted of only two authors, richard m ryan and edward l. deci. they are also co-authors of articles with fairly high citations, and both of them have an h-index of no less than . their main areas of research are psychology, in which self-determination theory and motivation are also a point of interest. out of the records did not specify the country/region of origin. of the remaining records, the countries that contributed the most were the united states ( . %), the united kingdom ( . %), and australia ( %). it should be noted that almost half of the studies were carried out in english-speaking countries. among the asian countries, china, india, and south korea stood out. national/regional contributions are double counted when authors of the same article are affiliated with institutions from different countries. our results show that the main research areas of study are medicine ( . %), computer sciences ( . %), behavioral sciences ( . %), computer medicine ( . %), and psychology ( . %). most articles contribute to more than one field. it seems that research in fitness apps has flourished through its study in the medical area, followed by its computational features. however, the study from the point of view of consumer behavior, integrated into the field of social sciences, seems not to have taken off yet. we predict significant growth in this domain as fitness apps become more popular, and communication through social networking sites goes viral, particularly among young people. the applied research methods allow the collection of empirical data to contribute to scientific knowledge. it is an important variable to understand the empirical orientations of research in this field of knowledge. total* . % *out of the total articles, articles ( . %) used multiple methods. of these, articles used two methods and one article used three methods. the most frequently used research method was the experiment. the experimental design was used in . % of all research. most of them were "in the wild" experiments, implemented on a small group of participants (n < ) who were asked to use a fitness app, developed expressly for the research, for a short period. the second most used research method was the survey ( . % of the articles), which allowed the evaluation of the user perspective and behavior with self-reported data. the third-ranked research method was content analysis. the articles that used this method analyzed and evaluated the total or partial functionality of a range of fitness-related apps, their technical characteristics and the attributes that make them more valued by users, more effective in changing consumer behavior, etc. for example, cowan et al. ( ) calculated a theoretical score for each of the health and fitness applications to determine whether the applications included relevant aspects of the behavioral change theory. the content analysis articles allow us to understand how fitness-related apps have evolved over the years and how researchers' focus has changed over that same period. by reviewing relevant articles, we found that behavior change techniques, gamification features, and consumer engagement strategies have been attracting attention, as shown in figure . the analysis of the frequency of appearance of the keywords allows the reader to approach the main topics analyzed in the articles in this field. the analysis of the keywords selected by the authors allows the determination of which relationships are established between a field of research and others close to it (durán-sánchez et al., ) . as shown in table , the terms "physical activity" and "mhealth" appear in . % of all the contributions. both keywords are the conceptual core of fitness app research. physical activity is also related to the terms "exercise" ( . %), "obesity" ( . %), and "weight loss" ( . %). portability is a concept associated with new devices for self-monitoring of activity: the terms "wearables" and "fitness tracker(s)" appeared in . % and . % of articles, respectively. the principle of playful functions is reflected in the term "gamification," with . % of the articles, which is a factor that can increase user adherence to the programs. figure maps the correlation between the keywords. to make the map clearer, with more focus on the core of the field of study, we removed the keyword "app" and its various related forms from the mapping analysis. the most frequent keywords were located in five differentiated clusters. cluster , which we named "digital mhealth" is mainly related to mhealth and ehealth (electronic health). they are platforms for fitness apps. also included in this group are keywords such as privacy and security, which are all related to the technology and device issues of fitness applications. cluster , which we named "mhealth and fitness trackers," is pretty similar to cluster , with only an emphasis on fitness trackers and persuasive technology as well as health apps and wearable electronic devices. cluster , which we named "physical activity, motivation, and social support," comprises keywords such as physical activity, exercise, physical fitness, etc. social support and motivation are also included in this group, which may be since these two are also important factors that support people to stick to physical activity (tang et al., ) . cluster , which we named "generalistic keywords," is more macro in nature and contains a wide range of topics such as fitness, mobile, and public health. cluster , which we named "behavior change and gamification," includes keywords such as behavior change, gamification, wearables, and self-determination theory. finally, based on all the information obtained as well as our thorough review of the contributions that are part of this bibliometric study, we now describe the main topics of research on the subject of fitness apps: ) analysis of the quality and performance of the use of the apps concerning the objectives of the users. the performance is measured through an evaluation of different indicators, such as the level of physical activity or weight loss. in this criterion of research, the use of innovative features is particularly important. for example, mata et al. ( ) tested the performance of the training planning function of the relevant apps and confirmed the high performance of these app-generated training and nutrition plans through expert validation. ) analysis of the benefit of the use of fitness apps for the chronically ill. patients affected by severe chronic diseases can undergo improvement in their general condition through lifestyle improvements. for example, bonato et al. ( ) analyzed the possibility of using an app for monitoring physical exercise routines for people affected by hiv. the apps are used to encourage patients to exercise to improve their general condition. examination of the use of fitness applications to encourage people with a specific need due to their socio-demographic profile to follow the minimum physical activity requirements established by the who. this includes the specific physical exercise needs that can be implemented through apps for the elderly (mas, palou & conti, ) , children (tripicchio et al. ) , or people with disabilities (pérez-cruzado & cuesta-vargas, ). increasing user motivation is an integral part of a significant number of articles. very high abandonment rates are observed in the use of these applications, and there is a lack of user engagement (bardus et al., ) . among the factors that may influence the use of the apps, some researchers are interested in the aesthetics of the user interface (bardus et al., ) , social relations (lewis et al., ) and the personalization (zhou et al., ) . some articles focus on the problems related to fitness apps and the adherence to hegemonic beauty canons. in this line of research, honary et al. ( ) concluded that the use of these apps might increase social pressure to achieve unrealistic beauty ideals and could thus increase the incidence of eating problems, such as anorexia or excessive physical exercise. another issue of concern relates to the privacy of and the large amount of personal data collected by these apps (adhikari et al., ) . wearable devices provide more accurate and convenient data for measuring people's daily activity levels. however, they are usually associated with relevant mobile apps for health data visualization and analysis. for example, lee et al. ( ) this study aimed to present in detail the current state of research on fitness applications through an exhaustive bibliometric analysis and bibliometric mapping. the social function and health potential of fitness apps represent a recent and growing phenomenon, which justifies an increase in the intensity of scientific research in recent years. . % of the contributions were published onwards when the usage of these apps had already been an important trend in the commercial market for several years. several bibliometric indicators (e.g., distribution of years of publication, price's index, author productivity, bradford's law, h-index, number of citations, source of publication, research areas, research methods, etc.) were analyzed to understand the main features and patterns of research on fitness apps. moreover, the scientific mapping analysis of the co-occurring keywords, co-authors, and co-citing authors provided an additional analysis from a timedepth perspective. in general, it is important to note the great dispersion of research, with a very high number of authors who have only made one contribution being a characteristic of a field of research that has not yet reached maturity. research in this field is still in its precursor stage. moreover, many of the studies have a relatively high number of co-authors. this situation is reflected in the indicator of author productivity, which is relatively low (oyibo, k. and vassileva, j. being the most active author with eight published articles). however, the most productive authors are not the most influential authors. west. j.h. has gained citations for his four articles, ranking first for this field of study. this dispersion of research is also reflected in the source of the publications. although there is a specialized journal in mhealth (jmir mhealth and uhealth), it can be found that submissions on fitness apps are distributed across a large number of academic journals and conference proceedings. with this data and support from the analysis of scientific mapping, it can be concluded that authors or prestigious journals have not been integrated and the research references in this field are relatively fragmented, partly due to their novelty and multidisciplinary requirements but also due to the technical orientation of the developers to circumvent the basic health, social, and behavioral aspects of health, society, and behavior. as in many other areas, the united states remains a prominent contributor in this area. china and india are the most productive in developing countries. these two countries are increasing their productivity and expanding their influence in various fields of scientific research at present. the most common research method used in this field is the experimental procedure that measures behavioral changes or changes in health indicators after a period of use. the second most used method is the survey, followed by the analysis of content. a considerable amount of literature is related to medicine, computer science, and healthcare. many authors have also focused on this main area of research. additionally, physical activity was the most frequently occurring keyword. "behavior change" linked to "physical activity" is also an important keyword. specifically, it refers to concepts such as behavior change theory, behavior change techniques (e.g., goal setting, self-regulation), etc. however, relatively few studies on consumer behavior from a social science perspective have been found. it seems that consumer-related research has mainly focused on analyzing the optimization of the functionalities of mobile applications from a medical or computer science point of view and neglected the aspects intrinsic to consumer behavior such as the motivations for using fitness apps, the attitude towards them, or how social networks influence the choice of the app to be used. the fact that the keyword "motivation" appears only times and all after is a clear indication of this finding. based on the generalization of all the information obtained and the review of the abstract and some of the full text, we found that the performance and function of fitness apps, the benefits for chronic disease treatment, the influence of using fitness app for public health, and factors of motivations of using fitness apps are currently popular research topics in this field. future research could build on these directions and incorporate relevant issues from a social science perspective (e.g., consumer motivations, consumer engagement, consumer behavior, etc) to further investigate on fitness applications. this article is useful in understanding the early state of research in the fitness app field. however, it is necessary to consider several limitations. one of the limitations of this study is the delimitation of the sample search criteria. in essence, the concept of fitness serves as a central reference for the applications that users utilize to perform selfmonitoring of health-related factors, particularly the level of physical activity. the control of "diet" is another health factor that overshadows and is superimposed on the concept of fitness, but one that could also be considered as a separate field in future studies, or add it to the keyword search scope for getting more comprehensive results. a novel preventative solution for effective asthma management: a practical evaluation security and privacy issues related to the use of mobile health apps web of science, scopus and google scholar: a content comprehensiveness comparison the reliability of a smartphone goniometer application compared with a traditional goniometer for measuring ankle joint range of motion el crecimiento de la literatura sobre la ley de bradford understanding recent trends in childhood obesity in the united states a review and content analysis of engagement, functionality, aesthetics, information quality, and change techniques in the most popular commercial apps for weight management bibliometric evaluation of researchers in the internet age expanding the technology acceptance model with the inclusion of trust, social influence, and health valuation to determine the predictors of german users' willingness to continue using a fitness app: a structural equation modeling approach smartphones and health promotion: a review of the evidence assessing the internal and external validity of mobile health physical activity promotion interventions: a systematic literature review using the re-aim framework an extensible approach for integrating health and activity wearables in mobile iot apps a mobile application for exercise intervention in people living with towards a taxonomy of mhealth sources of information on specific subjects toward a definition of "bibliometrics self-monitoring in weight loss: a systematic review of the literature developing a mobile app-supported learning system for evaluating health-related physical fitness achievements of students. mob gamification use and design in popular health and fitness mobile applications definitions: health, fitness, and physical activity apps of steel: are exercise apps providing consumers with realistic expectations? there's an app for that smartphone use in health and physical education do physical activity and dietary smartphone applications incorporate evidencebased behaviour change techniques? m-health y t-health bibliometric method for mapping the state-of-the-art and identifying research gaps and trends in literature: an essential instrument to support the development of scientific projects gamification for health promotion: systematic review of behaviour change techniques in smartphone apps knowledge management: a global examination based on bibliometric analysis hall of fame for mobile commerce and its applications: a bibliometric evaluation of a decade and a half understanding the role of healthy eating and fitness mobile apps in the formation of maladaptive eating and exercise behaviors in young people healthcare delivery of the future: how digital technology can bridge time and distance between clinicians and consumer. pwc health research institute integrating a mobile health applications for self-management to enhance telecare system social support patterns of middle-aged and older adults within a physical activity app: secondary mixed method analysis conference proceedings as a source of scientific information: a bibliometric analysis just a fad? gamification in health and fitness apps the frequency distribution of scientific productivity quantifying the body: monitoring and measuring health in the age of mhealth technologies efectos de un programa de entrenamiento presencial vs prescripción a través de una aplicación móvil en personas mayores a cross-domain framework for designing healthcare mobile applications mining social networks to generate recommendations of training and nutrition planning systematic analysis of coverage and usage of conference proceedings in web of science smartphone applications to perform body balance assessment: a standardized review improving adherence physical activity with a smartphone application based on adults with intellectual disabilities (appcoid) gamification and behavior change techniques in diabetes self-management apps little science, big science citation measures of hard science, soft science, technology, and nonscience technological innovation mediated by business model innovation: app developers moving into health apps to improve diet, physical activity and sedentary behaviour in children and adolescents: a review of quality, features and behaviour change techniques behavior change with fitness technology in sedentary adults: a review of the evidence for increasing physical activity how can weight-loss app designers' best engage and support users? a qualitative investigation technology components as adjuncts to family-based pediatric obesity treatment in low-income minority youth software survey: vosviewer, a computer program for bibliometric mapping brands as relationship builders in the virtual world: a bibliometric analysis there's an app for that: content analysis of paid health and fitness apps non-communicable diseases four decades of fuzzy sets theory in operations management: application of life-cycle, bibliometrics and content analysis evaluating machine learning-based automated personalized daily step goals delivered through a mobile phone app: randomized controlled trial key: cord- -eqn kl p authors: drissi, nidal; ouhbi, sofia; janati idrissi, mohammed abdou; ghogho, mounir title: an analysis on self-management and treatment-related functionality and characteristics of highly rated anxiety apps date: - - journal: int j med inform doi: . /j.ijmedinf. . sha: doc_id: cord_uid: eqn kl p background and objective: anxiety is a common emotion that people often feel in certain situations. but when the feeling of anxiety is persistent and interferes with a person's day to day life then this may likely be an anxiety disorder. anxiety disorders are a common issue worldwide and can fall under general anxiety, panic attacks, and social anxiety among others. they can be disabling and can impact all aspects of an individual's life, including work, education, and personal relationships. it is important that people with anxiety receive appropriate care, which in some cases may prove difficult due to mental health care delivery barriers such as cost, stigma, or distance from mental health services. a potential solution to this could be mobile mental health applications. these can serve as effective and promising tools to assist in the management of anxiety and to overcome some of the aforementioned barriers. the objective of this study is to provide an analysis of treatment and management-related functionality and characteristics of high-rated mobile applications (apps) for anxiety, which are available for android and ios systems. method: a broad search was performed in the google play store and app store following the preferred reporting items for systematic reviews and meta-analysis (prisma) protocol to identify existing apps for anxiety. a set of free and highly rated apps for anxiety were identified and the selected apps were then installed and analyzed according to a predefined data extraction strategy. results: a total of anxiety apps were selected ( android apps and ios apps). besides anxiety, the selected apps addressed several health issues including stress, depression, sleep issues, and eating disorders. the apps adopted various treatment and management approaches such as meditation, breathing exercises, mindfulness and cognitive behavioral therapy. results also showed that % of the selected apps used various gamification features to motivate users to keep using them, % provided social features including chat, communication with others and links to sources of help; % offered offline availability; and only % reported involvement of mental health professionals in their design. conclusions: anxiety apps incorporate various mental health care management methods and approaches. apps can serve as promising tools to assist large numbers of people suffering from general anxiety or from anxiety disorders, anytime, anywhere, and particularly in the current covid- pandemic. education, and relationships [ , ] . the exact causes of anxiety disorders are still unknown. according to the national institute of mental health, it is likely to be a combination of genetic and environmental factors [ ] . other possible factors that can lead to susceptibility include brain chemistry, personality type, exposure to certain mental and/or physical disorders, trauma and stress [ ] . the covid- outbreak, in addition to being a public health emergency, is also affecting mental health in individuals on a global scale causing people to suffer from stress, anxiety, and depression [ , ] . the pandemic is also triggering feelings of fear, worry, sadness, and anger [ , ] . quarantines, self-isolation, fear of the unknown, loss of freedom and other factors are causing psychological issues in people around the world [ , ] . these situations and circumstances can trigger several anxiety disorders, mainly separation anxiety disorder which is defined as fear of being away from home or loved ones, illness anxiety disorder which is defined as anxiety about a person's health (formerly called hypochondria) [ ] and panic attacks that are affecting a large number of people because of excessive worrying. psychiatric patients are additionally at a higher risk of experiencing symptoms related to psychological issues caused by the pandemic [ ] . people with preexisting anxiety disorders are showing aggravation of their conditions, for example, many people with ocd are developing new fixations on the covid- virus and are experiencing compulsive cleaning [ ] . due to the covid- pandemic, social interactions have significantly decreased in several parts of the world. while this may have provided relief to some people with social anxiety, it is possible this lack of interaction may have negative consequences in the longer term [ ] . returning to work after a period of lockdown, while still in the state of pandemic, is also causing the workforce to exhibit symptoms related to ptsd, stress, anxiety, depression and insomnia [ ] . the current covid- situation is also affecting the mental well-being of health care workers, who are at a high risk of psychological distress [ ] , especially those who are experiencing physical symptoms [ ] . the situation is further worsened by the recommended avoidance of inperson contact and fear of infection, as people with anxiety and other mental disorders might not be able to consult with a mental health professional. there are various barriers to mental health care delivery, such as cost, stigma, lack of mental health care professionals, and distance from health care services [ , ] . mobile mental health or m-mental health, which uses mobile technologies for providing mental health services, has the potential to help overcome mental health care delivery barriers, as it provides anonymous access to care, low to no cost care, and remote communication. smartphones can be a convenient tool to reach a large number of people from different parts of the world. there are many mobile applications (apps) for mental health problems such as ptsd [ , ] , stress [ ] , depression [ ] and alcohol dependence [ ] , as well as other health issues such as obesity, that apps can help with, especially due to lack of exercise during circumstances similar to the current lockdown [ ] . smartphone apps have high rates of acceptance among the general public, and especially in young people [ ] due to its cost effectiveness [ ] . many studies have reported that apps have shown positive results in the treatment and management of anxiety [ , , ] . this study aims to analyze the functionality and characteristics of highly j o u r n a l p r e -p r o o f rated anxiety apps to identify users' preferred features and management methods delivered for anxiety with a smartphone or a tablet. for the purposes of this study, only free apps were selected, as recent statistics in march showed that . % of android apps and . % of ios apps were freely available worldwide [ ] . a total of apps, android apps, and ios apps were selected. the anxiety management approaches used in these apps among other aspects of functionality have been extracted and analyzed. this section presents the methodology that was followed in order to select and analyze android and ios anxiety apps. this paper follows the quality reporting guidelines set out by the preferred reporting items for systematic reviews and meta-analysis (prisma) group to ensure clarity and transparency of reporting [ ] . google play repository and the app store were used as sources to select anxiety apps. both app repositories are very popular with a high number of available health care apps: more than , apps are available in the google play store, and more than , apps are available in the app store [ ] . a general search string, composed of only one word "anxiety", was used. it was automatically applied to the titles and descriptions of android and ios apps. j o u r n a l p r e -p r o o f each app from the search result was examined by the first author to decide whether or not to be included in the final selection. the second author revised the final apps selection. the following inclusion criteria (ic) were applied: • ic : anxiety related apps in google play store and app store. • ic : apps that have a free version. • ic : apps that have + stars rating. ic reflects a level of user satisfaction with the app. the focus is on highly rated anxiety apps so as to discover the functionality features and characteristics that provide high user satisfaction. the following exclusion criteria (ec) were applied to the candidate apps to identify the final selection that would be included in this study: • ec : apps that have less than raters. • ec : apps that could not be installed. • ec : apps that crashed and could not be used after installation. apps that match any of the ec were excluded from the selection. ec is based on the heuristic guideline by nielsen [ ] , which recommends having five evaluators to form an idea about the problems related to usability. the apps' selection process was established as follows: . the search string was used to identify candidate apps in the google play store and app store in order to create a broad selection from which to choose from. . ic were used to identify relevant apps. . apps that met one or more of the ec were excluded. the above actions were carried out in march . a final selection of android apps and ios apps was identified after application of ic and ec. fig. presents the selection results. data collection was carried out using the data extraction form presented in table . each app was installed and assessed to explore its functionality features and characteristics. the devices used for the apps' assessment were: oppo a (android ), and ipad (ios ). a template was designed in an excel file to provide basic information about the apps as well as specifying their main features and functionality characteristics. some of these characteristics and functionality features were retrieved from the app's description available in the app repository. this section presents and discusses the results of this study. a total of apps, android apps, and ios apps were identified as both free and highly rated apps. tables a. , a. , a. , a. , a. and a. in appendix present general information about the apps such as name, link, rating, number of raters, number of installations (not available for ios apps), and date of latest update. the majority of the selected apps ( %) offer in-app purchases for paid features and functionality. these apps are free to download and use, but many of their proposed functionality features are not available without purchase. thus, it can be said that users may not fully benefit from the app unless they purchase these specific features. however, it should also be noted that in-app purchases are a way for many developers to monetize their work j o u r n a l p r e -p r o o f apps general information: -name of the app. -date of the latest update. -users rating (scored out of ): to report the level of user satisfaction from the apps. -number of raters: to report the number of raters satisfied with the app. -number of installations (not available for ios apps): to identify the most installed apps. -in-app purchase: to identify whether free apps charge users for certain functionality features. -management method: to identify management and treatment methods for anxiety that could be delivered through an app, and the most used ones in the available apps. -intervention approach: to identify approaches that could be transmitted through an app, and the most followed approaches in the available apps. -targeted mental problem/symptoms: to identify anxiety related issues addressed by the apps and issues that might be managed with similar management methods and approaches as ones for anxiety, as well as to identify problems that could be treated and managed through apps. -involvement of mental health care professional: this information was extracted from apps' descriptions in-app repositories and from apps' content. we consider mental health care professionals to be those professionals with a mental health background including psychiatrists, therapists, counselors and experts in psychological issues or management methods. -physical health information such as hr and bp: to identify whether the app relies on physical indicators to assess the mental status of the user. -authentication method: to identify if the app provides users with the option to keep their personal health data inaccessible to other users of the same device. -gamification features: to identify whether gamification features are included in the app to encourage and motivate the users to keep using it. -social features which might include: links to communities, associations, and centers; interoperability with other apps or websites; the possibility to share content via social networks (sn); and contact information in case of emergencies. -languages: identify the availability of the apps in multiple languages, which reflects the degree of internationalization of the app. -offline availability: identify whether the app can be used without internet access. [ ] . the free version of the app is used by many developers as an advertisement tool to attract users into purchasing and unlocking more features [ ] . free apps with in-app options are becoming the norm in-app markets. in , in-app purchases accounted for more than % of ios app revenue in the us and % of revenue in asia [ ] . the majority of the selected apps ( %) updated their functionality and content in the three first months of . this could be linked to the current covid- pandemic situation. on december st, the who china office was informed of a number of pneumonia cases from an unknown cause, that were later linked to the coronavirus [ ] , which has now spread to all regions of the world [ ] . to limit the spread and risk of the virus, the who advised the public to practice social distancing and to stay home [ ] . many countries have declared obligatory lockdowns and people were quarantined, which has created a state of fear and worry that has elevated many individuals' anxiety and stress. various existing anxiety apps have, thus, been updated to include covid- related content. table presents various management methods identified in the selected anxiety apps with meditation and breathing exercises being the most common. the main goal of meditation is to help the user enter a deep state of relaxation or a state of restful alertness. it helps to reduce worrying thoughts, which play a key role in symptoms of anxiety, and bring about a feeling of balance, calmness, and focus [ ] . several studies have presented evidence supporting the use of meditation in anxiety treatments [ , , ] . one study reported that it was beneficial for a group of chinese nursing students j o u r n a l p r e -p r o o f meditation a , a , a , a , a , a , a -a , a , a , a , a -a , a , a , a , a , a , a , a , a , a , a , a , a , a , a , a , a , a , i , i , i -i , i , i , i , i , i , i , i , i , i , i , i , i breathing exercises a , a , a , a , a , a , a , a a , a , a , , i i i i games a , a , a , a , a , a , a , a , a , a , a , a , a , a , a , i , i , i , i , i -i , i , i , i , i , i -i , i , i , i assessment tests a , a , a , a , a , a , a , a , a , a , a , a , a , a , a , i , i stories a , a , a , a , a , a , a , a , a , i , i , i , i , i , i , i mindfulness practices a , a , a , a , a , a , a , a , i , i , i , i , i , i , i , i guided relaxation a , a , a , a , a , a , a , a , a , a , i , i , i , i community chats with app users via the app a , a , a , a , a , a , a , a , a , a ,i yoga and physical exercises a , a , a , a , a , a , a , a , a , a , a , a , a , a , a , a , i motivational and inspirational statements a , a , a , a , a , a , a , a , a ,i , i , i online therapy and coaching a , a , a , a , a , a , a , a i , i recommending activities and tips a , a , a , a , a , a , a , a ,i interactive messaging a , a , a , a in reducing anxiety symptoms and lowering systolic bp [ ] . another study reported that it showed improvements in the reduction of anxiety for breast cancer patients [ ] . a meta-analysis of controlled trials for the use of meditation for anxiety also reported a level of efficacy of meditative therapies in reducing anxiety symptoms [ ] . additionally, meditation has been shown to be effective in managing various types of anxiety such as panic disorder and agoraphobia [ ] . breathing exercises are another mechanism that can help to relax and relieve stress. while practicing deep breathing, a message is sent to the brain to calm down and relax. biochemical changes subsequently decrease hr and bp and help the person to relax [ ] . studies have shown that breathing exercises can improve cognition and overall well-being [ ] , while also reducing anxiety [ , , , , ] . breathing exercises can also have a positive impact on psychological distress, quality of sleep [ ] , depression [ , , ] , everyday stress, ptsd, and stress-related medical illnesses [ , ] . breathing exercises are also used to help with asthma, which was the case in a and a . however, it should be noted that such exercises may help patients whose quality of life is impaired by asthma, but they are unlikely to reduce the need for anti-inflammatory medication [ ] . many of the selected apps provided educational content about anxiety and other mental issues, symptoms, and management methods, either in the form of courses, articles, videos, or others. educating users about anxiety can help to reassure them and provide them with the necessary knowledge by answering questions and correcting misinformation that they might have. educating users about the provided management method and its benefits may also increase their trust in the management approach and their willingness to try it. mental assessment tests have been provided by some apps to give the user an idea about his/her mental status, anxiety, stress and/or depression levels. relaxing music and sounds, is a noninvasive and free of side-effects ap-proach that has been used in apps as a management method. it has been shown to be an effective tool for the reduction of anxiety, stress, and depression [ , ] . it has also shown positive results in the prevention of anxiety and stress-induced changes like hr and bp [ ] . developers should take into account the type of music and sounds used, as well as the accompanying environment, as they both affect the effectiveness of this method [ , ] . thirty-one apps provided journaling and writing diaries to help users plan their day, track their mood, and express their thoughts, feelings, and emotions. securing the privacy and confidentiality of users' information is critical in such apps. all selected ios apps providing journaling provide authentication methods, while only % of android apps with this functionality provide users with the same level of authentication. eleven apps provide the user with the possibility of communicating with other users. in these apps, users are able to share their experiences, talk about their issues, help each other, and relate to others who are undergoing similar problems as their own. in the current covid- pandemic, being in a state of isolation but having the ability to connect with an online community can be very helpful. the idea of enabling interaction with a community of people with similar issues is quite interesting and can be extremely helpful, especially given that people with anxiety often tend to avoid direct communication [ ] . for users who prefer communication with mental health care professionals, there are ten apps available that provide online therapy and coaching, enabling users to communicate with mental health care professionals, without having to travel, while also avoiding obstacles like stigma and distance. selected apps offering online therapy services charge fees for these services. these apps also provide information on the mental health care professionals' credentials. this information is important as it allows the user to check whether these professionals are appropriately accredited and decide which mental health care professional is most suited for his/her needs. thirty-three apps provide users with games like coloring books, puzzles, and slime simulations, as management methods for anxiety. these games help the user to relax, and to take his/her mind off worrying thoughts or feelings. games are usually enjoyable and entertaining and this may motivate users to continue using these apps. the variety of management methods identified in the selected apps points to the high potential of apps usage for coping with anxiety. developers have integrated various promising and effective management methods in their apps' functionality features. users can access these features at any time and in any place. this could be beneficial for users with anxiety disorders, especially in situations where immediate help is needed (e.g., during panic attacks), or in cases where mental health care professional cannot be reached due to circumstances like distance or the current global lockdown situation. table presents the selected apps which state the use of specific intervention approaches for anxiety management. the most used ones included mindfulness, cognitive behavioral therapy (cbt), and hypnosis. mindfulness was the most adopted management approach. it is defined as "bringing one's complete attention to the present experience on a moment-tomoment basis" [ ] . mindfulness practices allow practitioners to shift their concentration to their internal experiences occurring in each moment, such as anxiety and mood problems [ , ] , and improving an individual's internal cognitive, emotional, and physical experience [ ] . some findings suggest that mindfulness can be more complicated than it might seem, as many el-ements like attention emotional balance, differences in emotion-responding variables, and clinical context can influence its effect [ , , ] . therefore, these elements should be taken into account while developing mindfulnessbased anxiety apps. cbt is a form of psychological treatment, mainly based on efforts to change thinking patterns [ ] . many studies have supported the effectiveness of cbt-based interventions for the treatment of anxiety, and have reported on the long-term positive effect it has on both children and adults [ , ] . a study examining available evidence on cbt have yielded positive results and confirmed its effectiveness for anxiety disorders [ ] . cbt has also been used in the treatment of some specific anxiety disorders like ptsd [ ] and ocd [ ] . it has also been proved effective for depression, alcohol and drug use problems, eating disorders, and severe mental illness [ ] . cbt and mindfulness-based therapy can also be useful in reducing anxiety during the covid- pandemic [ ] . hypnosis is a therapeutic technique designed to bring relaxation and focus to the mind [ ] . many studies have reported the effectiveness of hypnosis for the treatment of anxiety. one study stated that it can reduce anxiety among palliative care patients with cancer [ ] , and another reported on its considerable benefits to terminally ill patients [ ] . hypnosis is also used to treat and manage stress and phobias [ ] , as well as sleep and physical symptoms [ ] . other approaches have also been identified in the selected apps as shown in table , but it should be noted that a few of them were not based on scientific approaches. table presents the different health issues besides anxiety that were addressed by the selected apps. all selected apps addressed general anxiety. some apps addressed specific types of anxiety like social anxiety, separation anxiety, performance anxiety, ocd, ptsd, and panic attacks. focus and concentration a , a , a , a , a , a , a , i , i , i self-esteem and confidence a , a , a , a , a , a , a , i , i , i , i , i pain a , a , a , a , a , a , a , i , i mood a , a , a , a , a , a , a , a , a some apps addressed other mental and physical issues, which usually occur with anxiety like stress [ , ] , sleep issues [ ] , and depression [ , , ] . some apps used management methods to treat addiction-related issues, eating disorders [ ] , phobias, [ ] , and asthma [ ] . the majority of the apps do not use physical health information. hr and bp are impacted by anxiety and stress [ ] . both can be used by apps to indicate the anxiety level of the user [ ] . yet in our selection only two apps provided this functionality feature (a and a ). a collects data on hr variability, using the photoplethysmogram (ppg) technique to get insights on the user's health, including stress, energy, and productivity levels. the app also allows the user to manually enter bp as a convenient way of journaling. it should be noted that a provides cardiovascular tests, including hr and peripheral blood circulation, as an app purchase option. only % of the selected apps reported involvement of mental health care professionals as presented in table . apps providing online therapy specified information about the therapists that the user can contact. this information includes their specialty, experience, and diplomas. some apps shown in table provided names of the professionals involved in their co-creation. providing names gives the user the possibility to look online for the credentials of the involved professionals and might increase the user's trust toward these apps. we cross-checked the names displayed in table and found them to be legitimate. table table presents the authentication methods identified in the selected apps. the majority of the selected apps ( %) do not require authentication. the absence of authentication might give the user a sense of anonymity. however, authentication can help the user ensure the privacy of his/her data. the app a requests a nickname and a password, ensuring security and confidentiality as well as keeping the anonymity of the user, since it does not use any information or sources that could reveal the identity of the user like facebook account, google account, or email. nickname and password a gamification is the use of game elements in non-gaming systems which are mainly used to improve user experience and user engagement [ ] . table presents the different gamification methods identified in the selected apps. note that some apps use more than one gamification method. the majority of the selected apps used gamification features to encourage and motivate the user. creating a fun, interactive user experience with the adoption of game elements can create an enjoyable user experience, which can further reduce boredom and motivate users keep using the app. this can also increase user engagement, leading to users providing more accurate information about their mental health status and to increased benefit for the user from the provided mental health care management method. gamification is a widely used approach that has shown effectiveness with anxiety and other mental health problems, such as depression and ptsd for military personnel [ , ] , and aggression for veterans [ ] . combining j o u r n a l p r e -p r o o f game a , a , a , a , a , a , a , a , a , a , a , a , a , a , a , a , i , i , i , i -i , i , i , i , i , i -i , i , i , i graphics a , a , a , a , a , a , a , a , a , a , a , a , a unlocking new features a , a , i , i , i score and points a , a , a , i stickers, awards and stars a , a , a , a , i game elements and knowledge on game players' behaviors with known mental health care management methods is an interesting approach that can result in the creation of effective anxiety apps. table presents the different social features provided by the selected apps. many apps provide social and communication features, which allow the user to connect with communities of app users as well as with centers and associations, or with others to share content and progress. those social features could prove to be beneficial to the user. for instance, sharing progress and content from the app via social networks (sn) and emails helps provide social support to the user from family and friends. social support is significantly associated with well-being and absence of psychological distress [ ] . it has a favorable effect on certain psychological issues [ ] , and can serve as a mediator to stress and anxiety caused by life events [ ] . providing social support is also among the behavioral change techniques implemented in m-health apps to promote app usage [ ] . additionally, providing contacts in case of emergencies is crucial and might help the user in critical situations j o u r n a l p r e -p r o o f where he/she feels the need for immediate help. links to associations, websites, and centers can provide the user with more helpful resources. social features are very important as they help the user connect with others in a beneficial way. emergency contacts' information a , a , a , a , a , a , a , a , a group treatment i , i table presents the languages available in the selected apps. the majority of the apps ( app) are available only in english, which can be explained by the fact that the search string applied in app repositories was in english. only one app (a ) automatically translates its content to the device's preferred language. while the rest of the apps are available in more than one language. availability in multiple languages can help reach a larger number of users. i , i -i , i , i , i , i , i , i , i , i -i , i -i , i , i , i -i more than one language a , a , a , a , a , a -a , a , a , a , a , a , a -a , a , a , a , a , a , a , a , i , i , i , i , i , i -i , i , i , i , i , i , i system's languages a j o u r n a l p r e -p r o o f table shows whether an app requires internet access to function or not. internet access is required to install and create accounts for all apps, but once that is done, many apps function without internet access. offline availability is an aspect that will help users benefit from the app without necessarily being in a setting with internet access. this will decrease the app's limitations and make it more accessible to users. however, some of the management methods identified do require internet access, like online therapy and communication with communities of app users. additionally, offline availability may require downloading more data that could be permanently stored, which may affect a phone's memory and performance. some apps were only partially available offline, resulting in limited functionality when internet access was not available. other apps only made downloaded data available offline, meaning the user chooses and downloads content that he/she wants to be available while offline. these are convenient solutions to offline availability that do not compromise on app functionality. this study is subject to limitations, such as: (i) missing terms (e.g., stress, depression) in the search string that might have resulted in the selection of relevant apps, as usually an app targets more than one mental health issue. however, the search string used identified any app that mentions anxiety in its title and/or description, therefore this can alleviate the threat of missing relevant apps; and (ii) the first author conducted the search and applied the ec and ic to the initial selection. however, the final selection has been reviewed by the second author. with the current development in mobile communication and the wide ownership of mobile devices, m-mental health seems to be one of the most promising ways to deliver care to people in need regardless of their situation. under certain circumstances like the current covid- pandemic, the use of mobile communication and apps for anxiety might become a necessity. panic attacks can mimic covid- symptoms, which might worsen the condition of people with anxiety disorders [ ] . having an app on hand that can ease anxiety in such circumstances is useful. this study highlights the functionality and characteristics of anxiety apps that are well rated by users. we plan to build on the reported findings to develop a reusable requirements catalog for anxiety apps. mental health care professionals and people with anxiety disorders will be involved in the co-creation of this catalog. the catalog will also include software quality requirements based on the iso/iec standard and recommendations from the uk national health service (nhs) and the health insurance portability and accountability act (hipaa) on health apps. since the reusable requirements catalog for anxiety apps will be based on functionality of existing highly rated apps, as well-being based on inputs from mental health care professionals and people suffering from anxiety, it could be used to assist developers to select relevant requirements for anxiety apps. apps could therefore be designed based on the catalog to assist people dealing with anxiety. requirements from the catalog could also be used to generate checklists for audit and evaluation purposes [ ] , either to evaluate apps or to compare their functionality and characteristics. the findings from this study may also assist researchers and developers interested in the field of m-mental health, especially in the sub-field of anxiety, to have an overview of the characteristics and functionality of existing highly rated apps for anxiety. our findings could also assist mental health professionals to find anxiety apps that could be integrated in their mental health care process, as well as assist people suffering from anxiety to find mobile apps best suited for their needs. during the covid- pandemic, mhealth can also help disseminate health information among health personnel and community workers [ ] . all authors contributed to the creation of the manuscript. nd: design, conception, acquisition and interpretation of data, classification of selected apps, drafting of the manuscript, revision. so: design, conception, statisti- j o u r n a l p r e -p r o o f what was already known on the topic: -anxiety disorders are a common mental issue. -there are many barriers to mental health care delivery, mainly cost, stigma and distance from health professionals. -apps were found to be effective tools to deliver mental health care, and overcome the aforementioned barriers. what this study added to our knowledge: - free and high-rated anxiety apps were analysed: android apps, and ios apps. -anxiety apps addressed other health issues, such as: stress, depression, sleep issues, and eating disorders. -anxiety apps adopted various management, treatment and coping approaches such as, meditation, breathing exercises, mindfulness and cognitive behavioral therapy. cal support, interpretation of data, drafting of the manuscript, critical revision. maji and mg: critical revision. all authors read and approved this manuscript. the authors have no conflict of interest. this article does not contain any studies with human participants or animals. j o u r n a l p r e -p r o o f what to know about anxiety everything you need to know about anxiety share of the population worldwide who suffered from anxiety disorders from what are anxiety disorders? anxiety causes immediate psychological responses and associated factors during the initial stage of the coronavirus disease (covid- ) epidemic among the general population in china a longitudinal study on the mental health of general population during the covid- epidemic in china anxiety on rise due to coronavirus, say mental health charities mental health considerations during covid- outbreak the psychological impact of quarantine and how to reduce it: rapid review of the evidence how to manage stress and anxiety from coronavirus (covid- do psychiatric patients experience more psychiatric symptoms during covid- pandemic and lockdown? a case-control study with service and research implications for immunopsychiatry how the covid- pandemic affects people with social anxiety is returning to work during the covid- pandemic stressful? a study on immediate mental health status and psychoneuroimmunity prevention measures of chinese workforce a multinational, multicentre study on the psychological outcomes and associated physical symptoms amongst healthcare workers during covid- outbreak barriers to mental health care: perceived delivery system differences barriers to mental health treatment: results from the who world mental health surveys mobile apps for post traumatic stress disorder gamification-based apps for ptsd: an analysis of functionality and characteristics gamification in stress management apps: a critical app review a systematic review of cognitive behavioral therapy and behavioral activation apps for depression the alcohol tracker application: an initial evaluation of user preferences online and smartphone based cognitive behavioral therapy for bariatric surgery patients: initial pilot study receptiveness and preferences of health-related smartphone applications among vietnamese youth and young adults enabling psychiatrists to be mobile phone app developers: insights into app development methodologies adding a smartphone app to internet-based self-help for social anxiety: a randomized controlled trial effectiveness of an app for reducing preoperative anxiety in children: a randomized clinical trial adoption of mobile apps for depression and f anxiety: cross-sectional survey study on patient interest and barriers to engagement statista. free and paid app distribution for android and ios the prisma statement for reporting systematic reviews and metaanalyses of studies that evaluate health care interventions: explanation and elaboration number of mhealth apps available in the apple app store from st quarter to rd quarter how to conduct a heuristic evaluation effect of perceived value and social influences on mobile app stickiness and in-app purchase intention how the most successful apps monetize their user bade novel coronavirus ( -ncov) situation report- coronavirus disease (covid- ) situation report- coronavirus disease (covid- ) advice for the public effectiveness of a meditation-based stress reduction program in the treatment of anxiety disorders effects of meditation on psychological and physiological measures of anxiety meditation therapy for anxiety disorders a randomized controlled trial of the effects of brief mindfulness meditation on anxiety symptoms and systolic blood pressure in chinese nursing students effects of meditation on anxiety, depression, fatigue, and quality of life of women undergoing radiation therapy for breast cancer meditative therapies for reducing anxiety: a systematic review and meta-analysis of randomized controlled trials stress management: breathing exercises for relaxation effect of short-term practice of pranayamic breathing exercises on cognition, anxiety, general well being and heart rate variability impact of jacobson progressive muscle relaxation (jpmr) and deep breathing exercises on anxiety, psychological distress and quality of sleep of hospitalized older adults sudarshan kriya yogic breathing in the treatment of stress, anxiety, and depression: part i-neurophysiologic model effects of a relaxation breathing exercise on anxiety, depression, and leukocyte in hemopoietic stem cell transplantation patients effectiveness of controlled breathing techniques on anxiety and depression in hospitalized patients with copd: a randomized clinical trial self-regulation of breathing as a primary treatment for anxiety breathing exercises for asthma: a randomised controlled trial effect of music on anxiety, stress, and depression levels in patients undergoing coronary angiography buela-casal, acute stress recovery through listening to melomics relaxing music: a randomized controlled trial relaxing music prevents stress-induced increases in subjective anxiety, systolic blood pressure, and heart rate in healthy males and females effects of relaxing music on cardiac autonomic balance and anxiety after acute myocardial infarction relaxing music for anxiety control bridging differences: effective intergroup communication mindfulness and meditation wherever you go, there you are: mindfulness meditation in everyday life skills training manual for treating borderline personality disorder mindfulness training as a clinical intervention: a conceptual and empirical review the effect of mindfulness-based therapy on anxiety and depression: a meta-analytic review mindfulness meditation, anxiety reduction, and heart disease: a pilot study mindfulness and anxiety disorders: developing a wise relationship with the inner experience of fear how does mindfulness reduce anxiety, depression, and stress? an exploratory examination of change processes in wait-list controlled mindfulness meditation training affective reactivity mediates an inverse relation between mindfulness and anxiety mindfulness and emotion regulation in depression and anxiety: common and distinct mechanisms of action what is cognitive behavioral therapy? long-term effectiveness of cbt for anxiety disorders in an adult outpatient clinic sample: a follow-up study cognitive behavioral therapy in anxiety disorders: current state of the evidence treating the trauma of rape: cognitivebehavioral therapy for ptsd cognitive-behavioral therapy for ocd mental health strategies to combat the psychological impact of coronavirus disease (covid- ) beyond paranoia and panic hypnosis and relaxation therapies a hypnotherapy intervention for the treatment of anxiety in patients with cancer receiving palliative care use of hypnotherapy in anxiety management in the terminally ill: a preliminary study the clinical use of hypnosis in cognitive behavior therapy: a practitioner's casebook gamification. using game-design elements in non-gaming contexts a virtual reality exposure therapy application for iraq war military personnel with post traumatic stress disorder: from training to toy to treatment cognitive processing therapy for veterans with military-related posttraumatic stress disorder virtual reality and cognitivebehavioral therapy for driving anxiety and aggression in veterans: a pilot study social support and mental health in community samples effects of social support and personal coping resources on depressive symptoms: different for various chronic diseases? social network mediation of anxiety behavior change techniques in top-ranked mobile apps for physical activity popular science. a panic attack can mimic the symptoms of covid- . here's what to do about it e-health internationalization requirements for audit purposes coverage of health information by different sources in communities: implication for covid- epidemic response deep key: cord- -ccmnkl e authors: altmann, s.; milsom, l.; zillessen, h.; blasone, r.; gerdon, f.; bach, r.; kreuter, f.; nosenzo, d.; toussaert, s.; abeler, j. title: acceptability of app-based contact tracing for covid- : cross-country survey evidence date: - - journal: nan doi: . / . . . sha: doc_id: cord_uid: ccmnkl e background: the covid- pandemic is the greatest public health crisis of the last years. countries have responded with various levels of lockdown to save lives and stop health systems from being overwhelmed. at the same time, lockdowns entail large socio-economic costs. one exit strategy under consideration is a mobile phone app that traces close contacts of those infected with covid- . recent research has demonstrated the theoretical effectiveness of this solution in different disease settings. however, concerns have been raised about such apps because of the potential privacy implications. this could limit the acceptability of app-based contact tracing among the general population. as the effectiveness of this approach increases strongly with app take-up, it is crucial to understand public support for this intervention. objectives: the objective of this study is to investigate user acceptability of a contact-tracing app in five countries hit by the pandemic. methods we conducted a multi-country, large-scale (n = ) study to measure public support for digital contact tracing of covid- infections. we ran anonymous online surveys in france, germany, italy, the uk and the us. we measured intentions to use a contact-tracing app across different installation regimes (voluntary installation vs. automatic installation by mobile phone providers), and studied how these intentions vary across individuals and countries. results: we found strong support for the app under both regimes, in all countries, across all sub-groups of the population, and irrespective of regional-level covid- mortality rates. we inves- tigated the main factors that may hinder or facilitate take-up and found that concerns about cyber security and privacy, together with lack of trust in government, are the main barriers to adoption. conclusions: epidemiological evidence shows that app-based contact-tracing can suppress the spread of covid- if a high enough proportion of the population uses the app and that it can still reduce the number of infections if take-up is moderate. our findings show that the willingness to install the app is very high. the available evidence suggests that app-based contact tracing may be a viable approach to control the diffusion of covid- . the covid- pandemic is the greatest public health threat of the last years. in the absence of effective treatment or vaccination (as of april ), the public health response has so far relied on non-pharmaceutical measures to limit the spread of the epidemic, such as physical distancing, case isolation, and manual contact tracing [ ] . these measures have not been sufficient to stop the epidemic. many countries have therefore resorted to partial or full "lockdown" measures to control the epidemic, severely limiting social and economic interactions among their citizens. although lockdowns may help countries to keep the number of infections under control [ ] , they come at a great social and economic cost [ ] [ ] [ ] [ ] . covid- is difficult to trace by traditional methods as covid- cases are infectious - days before experiencing symptoms and contacts on average become infectious - days after exposure. the window to achieve containment by manual contact tracing is thus extremely short. ferretti, wymant and colleagues [ ] have proposed digital (app-based) contact tracing as an alternative measure to contain the epidemic without the large economic costs of lockdowns. the idea is to use low-energy bluetooth connections between phones to record the interactions users have with others, particularly those interactions that may pose a higher risk of infection (e.g., spending more than minutes within two metres of another person). if a user is diagnosed with covid- , they can use the app to declare the diagnosis, which leads to a notification of all other users who have come in close contact with the infected person, asking them to isolate at home for days or until they have been tested by the public health authority. the main advantage over traditional (manual) forms of contact tracing is that the app allows instantaneous notification of contacts, which is a key determinant of the effectiveness of case isolation and contact tracing strategies for covid- [ ] . other advantages are that the automatic recording of contacts scales up easily, and avoids the loss of information due to patients' recall bias and/or imperfect knowledge of the people they have been in contact with. in recent weeks, several countries have announced plans to develop various types of contact-tracing apps [ , ] , and a few countries have already launched one, e.g., singapore [ ] . the success of appbased contact tracing, however, critically depends on people's willingness to use the app. hinch and colleagues [ ] simulate the epidemic in the uk and show that the app reduces infections at all levels of take-up but that it is only sufficient to stop the epidemic if approximately % of the population use it. it is therefore important to gauge the strength of public support for this approach and to understand the factors that may hinder or facilitate take-up. for instance, since the app would need to trace individuals' interactions with others, privacy concerns may undermine support and adoption. [ ] it is also possible that such a technological solution may not work as well for the less digitally literate share of the population, further increasing the unequal impact of the covid- pandemic within and across countries [ ] . in this sense, an "opt-out" installation policy, where mobile phone providers or apple and google [ ] would automatically install the app on phones, could maximize take-up. it is unclear, however, whether the public would be willing to support this more intrusive solution. in light of the many open questions surrounding the viability of app-based contact tracing, we designed a survey to measure public support for this approach in five countries that are currently hit by the covid- pandemic: france, germany, italy, the uk and the us. the specific objectives of our study are to (i) assess the overall acceptability among the public of app-based contact tracing under different installation policies (e.g., voluntary installation or automatic installation by the government); (ii) uncover country-level and individual-level variation in support for the app; and (iii) understand the main mechanisms that may facilitate or impede app usage across various subgroups of countries and individuals. survey design. we conducted large online surveys in five countries (france, germany, italy, uk, and the us) to measure acceptability of app-based contact tracing for covid- . a complete description of the survey can be found in the multimedia appendix; here we provide an overview. at the beginning, after collecting respondents' informed consent, we described the app, explaining how it would function as well as its purpose. respondents had to pass a comprehension check to proceed further. we then asked respondents how likely they would be to install the app on their phone, if it became available to download voluntarily ("opt-in" installation policy). respondents were then asked about their main reasons for and against installing the app as well as their compliance with self-isolation requests. next, we assessed to what degree respondents would be open to an "opt-out" policy, where mobile phone providers would automatically install the app on all phones, but users would be able to uninstall the app at any time. we then collected demographic information and concluded the survey with questions about respondents' attitudes towards the government under different installation regimes. we kept the survey design as similar as possible across all five countries, with a few exceptions to accommodate country differences with regard to lockdown measures in place at the time of taking the survey. the us survey (deployed last) contained a few additional questions, including robustness checks. see section a in the multimedia appendix for more details. ethics approval was obtained from the university of oxford (reference number econcia - - ). no personal information was collected as part of the study. target population, sample size and attrition. the surveys were administered between th march and th april . we recruited respondents through lucid, an online panel provider. we targeted a sample size of respondents in each of the four european countries, and in the us -with quotas set for the samples to be representative of the overall population in terms of gender, age and region of residence. a total of individuals started the survey and consented to participate (a participation rate of %). out of the people who consented to participate, passed the comprehension check and started the main questionnaire. after removing incomplete responses and duplicates, we have a sample of complete and unique responses (a completion rate of %). finally, we removed respondents who either did not own a mobile phone or did not disclose their gender, leaving us with a final sample of respondents. to control for the potential effect of our recruitment method, we repeated the german survey with a probability-based sample in an offline recruited online panel. see section b in the multimedia appendix for further details on recruitment, filtering and attrition, and the final sample. statistical analysis. our main outcome variables measure respondents' intention to have the app installed on their phone under the two installation regimes (opt-in vs. opt-out). the outcomes were measured on a -point ordinal scale (opt-in: from definitely install to definitely won't install ; optout: from definitely keep to definitely uninstall ). in our regression analysis, we dichotomize these outcome measures (= if definitely or probably install / keep the app, and = otherwise). we use multivariate regression analysis (linear probability models; probit and ordered logit in additional analyses presented in the multimedia appendix) to examine the relationship between intention to install and a number of covariates: age, gender, country, presence of comorbidities (diabetes, high blood pressure, heart or breathing problems), usage of mobile phone outside the house, frequency of social interactions, ability to work from home during the lockdown, ability to obtain sick pay while working from home, trust in national government, and incidence of covid- deaths in a respondent's region of residence (see section c. in the multimedia appendix for more details). table b . in the multimedia appendix presents a summary of these covariates. result : we find broad support for app-based contact tracing. support is high in all countries, across all subgroups of the population, and under both installation regimes (opt-in and opt-out). panel a of figure shows that, under the voluntary (opt-in) installation regime, . % of respondents across all countries would probably or definitely download the contact-tracing app, if it was available. panel b shows that . % of respondents would also probably or definitely keep the app installed on their phone under the automatic (opt-out) installation regime. in both regimes, the share of respondents who would not have the app installed on their phone is very small (red portion of the bars in figure ). support is high in all five countries where we implemented the survey: in each country, at least % of respondents say that they would install or keep the app. moreover, figures , , and in the multimedia appendix show that support for the app is generally high across various subgroups of the population (e.g., across men and women, across different age groups, etc.), suggesting widespread acceptability of the app-based contact tracing solution to the covid- pandemic. although support is very high overall, there is systematic variation both within and across countries. despite the broad and widespread acceptability of the app, we find that support varies systematically across countries and individuals. for instance, figure shows that germany and the us are relatively less supportive of the app compared to the other countries. this is the case both under the opt-in and opt-out regimes. among individual characteristics, we find that those who have lower trust in their national government are more hesitant to have the app installed on their phones ( figure in multimedia appendix). we further explore this heterogeneity using multivariate regression analysis, where we examine the relationship between support for the app and a variety of individual-and country-level covariates. figure shows the impact that these covariates have on the probability of definitely or probably installing the app under the opt-in regime, using a linear probability model (see section c. in the multimedia appendix for a similar analysis of the opt-out regime). the analysis confirms that germany and the us are significantly less supportive of the app, especially compared to france and italy. taking the two most extreme cases, respondents in italy are . percentage points ( % ci . - . ) more likely to support the app than respondents in the us. surprisingly, figure shows very little correlation between regional-level covid- mortality rates and support for the app. among individual-level characteristics, we find that people who carry their phone with them more often are more likely to install the app. those who always carry their phone with them are . percentage points ( % ci . - . ) more likely to support the app than those who carry their phone only rarely. app support is also . percentage points ( % ci . - . ) larger among respondents with one or more comorbidities. moreover, the probability of installing the app increases with trust in the government. people who completely trust the government are . percentage points ( % ci . - . ) more likely to install the app than those who do not have any trust in the government. we find similar results using an ordered logit model, a linear probability model dichotomizing on just definitely install, and when using a probit model (see multimedia appendix). finally, results are also qualitatively similar when considering installation intentions under opt-out rather than opt-in (figure in multimedia appendix). interestingly, under the opt-out regime, trust in government displays an even stronger correlation with the intention to keep the app installed on one's phone. result : concerns about privacy and app security underlie some of the key variation in support for the app. we can use the data on respondents' reasons for or against installing the app to better understand the nature of the observed variation in app support across countries and individuals. a first set of reasons against the app revolved around concerns about government surveillance at the end of the . cc-by . international license it is made available under a is the author/funder, who has granted medrxiv a license to display the preprint in perpetuity. (which was not certified by peer review) the copyright holder for this preprint this version posted may , . . https://doi.org/ . / . . . doi: medrxiv preprint figure : determinants of stating definitely install or probably install note: the dependent variable is an indicator variable taking the value if a respondent chose definitely install or probably install when asked whether they would install the app or not, and otherwise. we use a linear probability model. lines represent % confidence intervals calculated with heteroskedasticity-robust standard errors. all coefficients are the result of a single regression and thus display marginal effects. a coefficient of . implies a respondent who chose this option is percentage points more likely to state they would definitely or probably install the app relative to the base category. epidemic (mentioned by % of respondents) and cyber security (fears that the app could make the phone vulnerable to hackers; %). respondents also reported that usage of the app may increase feelings of anxiety ( %), possibly reflecting aversion to feedback about a possible infection. the most frequent reasons in favor of the app were willingness to protect family and friends ( %), a sense of responsibility towards the community ( %), and a hope that the app may stop the epidemic ( %). figures and in the multimedia appendix show the relationship between the probability of selecting a particular reason and country-and individual-level characteristics. several patterns are of interest. first, we find that, compared to other countries, respondents in germany and the us are more likely to mention concerns about government surveillance as one of the reasons against installing the app. in these countries we also see a larger share of respondents expressing concerns about security of the app, especially compared to italy and the uk. thus, concerns about privacy and security seem to be an important impediment to the adoption of the app, particularly in germany and the us. among individual-level characteristics, we find that respondents who have less trust in their national government are also more likely to express concerns about government surveillance. this suggests that privacy concerns play a role in the negative relationship between trust in government and probability of installing the app found in figure . in contrast, we find that frequent usage of mobile phones is related to a stronger perception of the potential benefits of the app: respondents who more often carry their phone with them are more likely to believe that the app would benefit them, by helping them stay healthy and keeping them informed about the risks of infection. principal findings in our study, we find high support for app-based contact tracing -irrespective of age, gender, region or even country of residence. since the effectiveness of app-based contact tracing crucially depends on a sufficient level of take-up, our findings are encouraging for the prospects of this approach. although support is high in all countries and subgroups of the population, the data reveal that concerns about cyber security and privacy, coupled with trust in government, are important determinants of support. countries with stronger privacy and security concerns (germany and the us) are relatively less supportive of app-based contact tracing. individuals who have less trust in their national government are also less supportive. implications the lack of trust in government can have far-reaching implications. our analysis shows that this factor has a negative effect on people's intention to install a contact-tracing app on their phones. furthermore, supplementary analysis (see section c. in the multimedia appendix) also shows that people with lower trust in government are more in favor of an opt-in installation policy than an opt-out regime where the government asks mobile phone providers to automatically install the app on all phones. an opt-out regime is likely to translate into higher effective installation rates, for instance by reducing the negative effects of procrastination or inattention [ ] . however, our data suggests that only governments that enjoy a relatively high level of trust from their citizens may be able to resort to more paternalistic approaches. a policy implication of these findings is that governments should consider delegating the organization of app-based contact tracing to a highlyreputable and transparent public health authority at arm's length from the government. our results also point towards the need to address privacy and cyber security concerns with an app design that respects user personal data as much as possible. research on the privacy implications of app-based contact tracing, and the potential solutions to these concerns, is currently underway [ , , ] . interestingly, however, when we ask our respondents how the data collected by the app should be treated, we find that nearly % would consent to making the de-identified data available to research. limitations our study has some limitations that we tried to address in different ways. first, respondents recruited online may not be representative of the entire population. in particular, digital literacy and willingness to share data could be higher among such respondents. to ensure that our results do not hinge on our specific sample, we replicated an abridged version of the german survey with a different panel provider that randomly recruits its participants offline. our results remain almost completely unchanged (see section b. in multimedia appendix). second, our survey asked hypothetical questions about future behavior. however, high levels of intended installations may not directly translate into actual installations. nevertheless, studies often . cc-by . international license it is made available under a is the author/funder, who has granted medrxiv a license to display the preprint in perpetuity. (which was not certified by peer review) the copyright holder for this preprint this version posted may , . . find good correlation between what people declare they would do in surveys and actual behavior [ ] [ ] [ ] [ ] [ ] , even in relation to app installations [ ] [ ] [ ] [ ] . more generally, broad support for the approach is a necessary first stage to adoption, and our findings about heterogeneity in support point towards specific subgroups of the population that may need stronger encouragements to adoption. we show in section c. of the multimedia appendix that respondents who would install the app mention far more reasons for its adoption than those who would not install it (but a similar number of reasons against). we show in section c. that respondents in our replication study who did not answer the comprehension questions correctly were less willing to install. stressing the various benefits of the app, to oneself and others, and explaining the working and purpose of the app may be a particularly effective strategy to foster adoption. third, in our survey, we measured support for the general concept of app-based contact tracing, leaving out specific details regarding the implementation, which were not available to us at the time respondents took the survey. one downside of only surveying about the general idea is that it might be harder for respondents to visualize how such a system could work, which may increase hypothetical bias. however, we find that the details we gave about implementation (e.g., whether the app uses bluetooth or gps) seem to have very little impact on support. this suggests that our general measure of support for app-based contact tracing may be portable across different implementation settings. finally, our survey respondents were recruited from a specific subset of industrialized western democracies. attitudes towards app-based contact tracing may vary across countries with different levels of development and political regimes. it is nevertheless encouraging, in terms of external validity, that we observe a strong similarity in responses across the five countries we sampled, and that analogous findings have been reported in ongoing surveys conducted in australia and taiwan [ ] . in developing countries and among disadvantaged populations, the more limited access to smartphones raises both efficacy and equity issues; the development of low-cost bluetooth devices with similar functionalities could improve access to digital contact tracing. in conclusion, our study shows strong public support for app-based contact tracing to tackle covid- . this is an important result since public support is a necessary condition for the viability of the approach. further research is needed to gauge the extent to which public support for app-based contact tracing translates into actual app adoption and, more generally, to evaluate its potential for epidemic control. frederic gerdon: data collection, data analysis ruben bach: data collection, data analysis frauke kreuter: study design, data collection, data analysis daniele nosenzo: study design, data collection, data analysis, writing séverine toussaert: study design, data collection, data analysis, writing johannes abeler: study design, data collection, data analysis, writing abbreviations covid- : coronavirus disease % ci: % confidence interval none declared. survey details, additional results and survey questionnaire. . cc-by . international license it is made available under a is the author/funder, who has granted medrxiv a license to display the preprint in perpetuity. (which was not certified by peer review) the copyright holder for this preprint this version posted may , . . https://doi.org/ . / . . . doi: medrxiv preprint . cc-by . international license it is made available under a is the author/funder, who has granted medrxiv a license to display the preprint in perpetuity. (which was not certified by peer review) the copyright holder for this preprint this version posted may , . cc-by . international license it is made available under a is the author/funder, who has granted medrxiv a license to display the preprint in perpetuity. the copyright holder for this preprint this version posted may , . table gives an overview of the structure of the survey that was administered to respondents. in the following, we describe the various parts in more details. after expressing their informed consent to take part in the study, respondents were given a description of the contact tracing app. we explained that the app would be developed by a national health organization and that, once installed, it would register other users in close proximity via bluetooth (and, in the case of the us and the uk, potentially location data). users who were found to have been in close proximity to a confirmed case of covid- for at least minutes would be alerted by the app and asked to quarantine at home for days or until they could be tested for the virus. all other users would see an "all clear" message. we further explained that an early quarantine would prevent individual users from passing on the virus to their loved ones in the early (presymptomatic) stages of their potential infection, and might slow down or even stop the epidemic. we also stressed that the identity of all users would be private throughout the entire process. in order to progress to the main part of the survey, respondents had to correctly answer three comprehension questions about the functioning of the app. cc-by . international license it is made available under a is the author/funder, who has granted medrxiv a license to display the preprint in perpetuity. the copyright holder for this preprint this version posted may , . . after this introductory part, the main questionnaire began, which was designed to address the following three objectives: . assess general support for adopting the app (and complying with its requests); . identify the main drivers of adoption intentions (reasons for and against); . evaluate support for different implementation policies (opt-in vs. opt-out installation regimes). first, respondents were asked to assess their likelihood of installing (or not) the app on their phone; responses were collected on a -point scale, from definitely install to definitely won't install. respondents were then asked about their main reasons for and against installing the app; in both cases, respondents could select multiple reasons from a menu of options (the order of these options was randomized at the individual level). we subsequently asked respondents how likely they would be to comply with the request to self-isolate for days if they had been in close contact with an infected person. responses were collected on a -point scale from definitely comply to definitely won't comply. those who did not select definitely comply were then asked whether their chances of compliance would increase, decrease, or remain the same if health services committed to test them for the virus within two days from the start of their self-isolation (a negative test allowing them to stop self-isolating). respondents who did not say they would definitely install the app in response to the initial installation question were asked further questions about their willingness to install the app in three additional scenarios: (i) in case the epidemic had spread to someone in their community, (ii) to someone they knew personally, (iii) or in case an "all clear" message by the app would be associated with a relaxation in lockdown restrictions. we next assessed whether respondents would be open to an "opt-out" policy: the government would require mobile phone providers to automatically install the app on all phones, but users would be able to immediately uninstall it. respondents were asked about their willingness to keep (vs. uninstall) the app in this case (on a -point scale from definitely keep to definitely uninstall ). as a follow-up question, they were asked to rate the extent to which they agreed with the following statement: "the government should ask mobile phone providers to automatically install the app on all phones" (on a -point scale from fully agree to fully disagree). respondents who did not fully agree with the statement were then asked whether their opinion would change if someone in their community, or someone they knew personally, had been infected with the virus. finally, they were asked for their preference regarding the data policy that should be adopted once the epidemic is over: automatically delete all data, de-identify the data and make it available for research purposes, or some other option of their choice. the third block of the survey collected basic demographic information (age, gender, region of residence) as well as information about potential risk factors for contracting the virus (frequency of social interactions and health risks), use of smartphone, and ability to work from home and receive (some fraction) of payment. the final block consisted of questions about political orientation and attitudes towards the government. we first asked respondents to state their political affiliation. we then asked them whether, in general, they "trusted the government to do what is right". finally, we asked them whether their opinion of the government would improve in case of (i) an opt-in installation regime, and (ii) an opt-out regime. we kept the survey design as similar as possible across all five countries, with a few exceptions. first, the uk survey was administered before the government issued a nation-wide lockdown. on the other hand, respondents in germany, italy and france, were surveyed after such a lockdown had been implemented. in the us, most states were under a "stay at home" order by the time of the survey, but not all of them. to reflect this difference in the environment respondents faced, we slightly adjusted the phrasing of some of our questions between areas in lockdown and those that were not. second, unlike in the other countries, we asked uk and us respondents about an app that might use gps data in addition to bluetooth. third, in the us survey, which was run last, we included a question explicitly asking respondents which installation regime (opt-in or opt-out) they preferred. we did this in order to check the consistency between their actual preferences for installation regimes and their installation intentions under each regime elicited in parts and . we also added demographic questions about ethnicity, area of residence, health insurance and media use, as well as two versions of a question about willingness to install if a private company like facebook endorsed it. a detailed overview of the entire survey flow explaining the survey logic and highlighting differences between countries can be found in section d in the multimedia appendix. the full texts of the different surveys can be found here: uk, france, germany, italy and us. the survey was administered between the th and th of march in the four european countries (france, germany, italy, uk), and between the th and th of april in the us. no personal data was collected at any point during the survey and we obtained informed consent as well as checked for bots before the survey began. respondents who accepted the survey invitation were directed to our online questionnaire, programmed using the software qualtrics. the survey was pretested before fielding by generating test observations (around iterations for each country) to check data quality and consistency. moreover, we always had a soft launch before the full launch. that is, after collecting the first responses, we paused recruitment and checked the data before launching the survey widely. no adjustments had to be made in any of the countries. our survey was not password protected but distributed through an open link, so in principle respondents could take part multiple times. however, the panel provider lucid assigns a unique id to each respondent which we tracked within the survey, which allowed us to identify and exclude multiple entries from the same respondent. only a very small number of respondents completed the survey more than once; see section b. for more details. throughout the survey, respondents were allowed to navigate back and forth between screens except for when adaptive questioning was used. furthermore, we enforced responses in all questions but those about reasons for and against installation (as we did not want to force people in favor of/opposed to installation to give reasons against/for, respectively). consequently, every participant who submitted the survey also replied to every question. we did give people the option to choose "don't know" in response to all the installation questions (as well as a question about sick payment in the demographics part); however, since generally not many people chose it, we merged this option with the midpoint of the scale during data analysis. the entire survey was stretched over - screens ( - screens in the us), depending on the adaptive questioning, with - items per screen. the average completion time was . minutes in france, . minutes in germany, . minutes in italy, . minutes in the uk, and . minutes in the us. in all five countries, we recruited respondents through lucid, an online panel provider that works with a variety of sample suppliers to get a broad range of volunteers. volunteers were recruited using a multitude of methods, from double opt-in panels (the vast majority), publishing networks, social media, and other types of online communities. in some cases, participants were furthermore recruited offline e.g., via tv and radio ads, or with mail campaigns. participation was voluntary and, in the majority of cases, incentivized with most suppliers providing loyalty reward points or gift cards, and some providing cash payments. in each country, we set recruitment quotas so as to achieve a sample of respondents that was representative of the adult population of the respective country with regard to gender, age and region of residence (quota-based sampling). furthermore, we did not invite individuals who did not own a mobile phone. across the five countries, individuals started the survey, out of which consented to participate (a participation rate of %). before participants could begin the main part of the survey, we briefly described the app, and asked three comprehension questions to ensure participants were paying adequate attention and not just (randomly) clicking through the survey. only if all three comprehension question were answered correctly were respondents allowed to continue with the survey. out of the people who consented to participate, failed to answer all three comprehension questions correctly, leaving us with a sample of respondents who started the main questionnaire. of these respondents either exited the survey before submitting or took the survey more than once and were therefore excluded from the sample, which left us with complete and unique responses (a completion rate of %). finally, we dropped observations from individuals who indicated in the comments section they did not own a smartphone, and individuals who either did not identify as male or female, or who preferred not to disclose their gender. this gave us a final sample of participants for whom we have responses to all questions. a few respondents still noted that they did not have a smartphone in the comments section (as a reason for not installing the app) -we excluded those responses from our final sample. for the few people who took the survey more than once, we kept their first submission. because there were so few people who either wanted to keep their gender private or identified as non-binary, it did not seem useful to have an extra category when analyzing the results by gender. . cc-by . international license it is made available under a is the author/funder, who has granted medrxiv a license to display the preprint in perpetuity. the copyright holder for this preprint this version posted may , . . cc-by . international license it is made available under a is the author/funder, who has granted medrxiv a license to display the preprint in perpetuity. the copyright holder for this preprint this version posted may , . notes: health problems refers to diabetes, asthma, high blood pressure, heart or breathing problems. in the demographics, we only gave respondents a "don't know" option for the question about sick pay. "trust" indicates whether respondents agreed that, in general, they trust their government to do what is right. the number of cases and deaths per million people in each country refers to the number on the final day of surveying. . cc-by . international license it is made available under a is the author/funder, who has granted medrxiv a license to display the preprint in perpetuity. the copyright holder for this preprint this version posted may , . lucid recruits survey participants online through a variety of sampling partners. from these partners, lucid receives demographic information about the pool of potential participants and uses this information to target respondents in accordance with the quotas set for the survey. in our case, we targeted the gender and age composition as well as the regions of residence of the respondents to be representative of each country's overall population. after sending the survey out to an initial sample meeting these quotas, lucid subsequently focuses on particular demographic groups in the re-sampling to meet the quotas in the final sample of participants (i.e., if initial take-up among people over the age of was low, lucid would disproportionately target them or even close data collection on younger cohorts completely after a while to ensure balance across age groups). however, while we tried to ensure that our sample was representative of each country's population in terms of gender, age and region of residence, the composition could still differ from the country averages in terms of other characteristics. furthermore, respondents who are recruited online may be more tech savvy and willing to use a phone application than the average individual. this could potentially bias our estimates. to address these concerns and assess the external validity of our sample, we investigated its representativeness in two ways, using the german sample as an example. first, we re-created our key figures using sampling weights to harmonize the characteristics of our sample with the german population at large. our results remained generally consistent -see for example figure which depicts the re-weighted response likelihoods to the question "how likely would you be to install the app on your phone". however, if unobserved factors like tech savviness are not strongly correlated with demographics, then using survey weights alone is not enough to un-bias the results. as a second step, we also tested whether our results are robust to alternate recruitment methods. one week after the initial survey, we repeated our online survey twice, once again with lucid and also with forsa. although both surveys were conducted online, forsa, unlike lucid, recruits its online panel members from a probability based, randomly selected telephone sample. this offline recruitment process ensures that technical literacy and willingness to share data should play less of a role in the selection into the forsa sample. we find almost exactly the same results in the replication survey, alleviating concerns about our original sample -see, for example, figure which depicts results for the question "how likely would you be to install the app on your phone". more details about the replication can be found in the german country report. to recruit participants, forsa randomly generates phone numbers, calls the respective households and invites them to participate in the survey. due to the direct contact over the phone, the sample should contain neither bots, nor respondents with a dubious identity (due to fake identities or multiple sign-ups). . cc-by . international license it is made available under a is the author/funder, who has granted medrxiv a license to display the preprint in perpetuity. the copyright holder for this preprint this version posted may , . . figure : would you install the app? -robustness in the german sample note: lucid refers to our initial, unweighted german sample collected with lucid. lucid weighted refers to the same sample albeit now reweighted with population weights. lucid refers to our follow-up german sample that we collected a week after the initial one with lucid. finally, forsa refers to the sample we collected a week after the initial one with forsa. light/dark red bars correspond to probably/definitely won't install. for more information, please see the german country report. . cc-by . international license it is made available under a is the author/funder, who has granted medrxiv a license to display the preprint in perpetuity. the copyright holder for this preprint this version posted may , . using data from only individuals who pass all three comprehension questions may introduce selection bias in our sample. for example, more tech savvy respondents may simultaneously be more likely to pass the comprehension checks and more inclined to install any app. to address this concern, we consider a selection model where the likelihood of a respondent consenting to the survey and correctly answering all comprehension questions is a function of the demographic information we obtained from lucid for everyone who started the survey (age, gender, region, employment status, and, in the case of italy, france and germany, education). we estimate the following probit model: where Φ() gives the normal cumulative distribution function. we also estimate an ordered probit model, taking the stage of a respondent's drop out (consent, comprehension question , or ) as our ordinal outcome variable. the results of these empirical specifications are given below. respondents with less than high school education are significantly less likely to be included in our final sample. the same is true for male respondents, younger respondents, and the self-employed. given our selection model, we re-weight our analyses with inverse probability weights reflecting the likelihood that each individual was sampled. therefore, using our selection model, weights are constructed using the predicted probability of completing the survey, Φ(βx i ). the results from re-weighting the responses to the main questions on voluntary and automatic installation are given in figure . importantly, we see that the results are broadly consistent with the results presented in the main text. . cc-by . international license it is made available under a is the author/funder, who has granted medrxiv a license to display the preprint in perpetuity. the copyright holder for this preprint this version posted may , . . note: the reference categories for the covariates are as follows: female (gender), - (age), less than high school education (education), employed full-time (employment). "don't know", and empty answers are excluded. we also control for region fixed effects. plotted points are the estimated model coefficients. lines represent % confidence intervals calculated with heteroskedasticity-robust standard errors. therefore the graph is predominantly helpful for assessing which demographic factors are statistically significant in predicting whether a respondent completes the survey. . cc-by . international license it is made available under a is the author/funder, who has granted medrxiv a license to display the preprint in perpetuity. the copyright holder for this preprint this version posted may , . . https://doi.org/ . / . . . doi: medrxiv preprint figure : likelihood of having the app in an opt-in and opt-out regime, weighted by inverse probability of staying in sample note: in addition to the covariates shown in figure , we also include region fixed effects, and country-specific education and employment coefficients. light/dark red bars correspond to probably/definitely won't install (uninstall) in panel a (b). finally, we also compared the demographic information we received from lucid to the self-reported demographic information on gender and age. this allows us to check that respondents gave accurate information in our survey. respondents ( . %) gave at least one inconsistent answer. removing these individuals has no effect on the results. . cc-by . international license it is made available under a is the author/funder, who has granted medrxiv a license to display the preprint in perpetuity. the copyright holder for this preprint this version posted may , . . https://doi.org/ . / . . . doi: medrxiv preprint c additional results in the main part, we make a couple of modelling assumptions. namely we use a linear probability model and dichotomize the outcome measure by grouping everyone who said they would probably or definitely install the app into the "install" category (assigned value ) while everyone else falls into the "non-install" category (with value ). however, our results are robust to making different modelling assumptions as well. figure displays the results of the opt-in installation question if we model the entire choice space as an ordered logit where high numbers indicate a lower probability of installing the app, so that negative coefficients here have a similar interpretation to positive coefficients in the linear probability model in the main part. we see that the results look very similar to the ones obtained with the main specification (see figure ). furthermore, our results also hold when we sort only the people who say they would definitely (rather than definitely or probably) install the app into the install category -both using a linear probability and a probit specification. for brevity, we do not show these results here. . cc-by . international license it is made available under a is the author/funder, who has granted medrxiv a license to display the preprint in perpetuity. the copyright holder for this preprint this version posted may , . . just as we have looked at the covariates determining the choice to either install or not the app voluntarily (opt-in) in figure , we can also look at the impact of different influencing factors on the decision to keep or immediately uninstall an automatically installed app (opt-out). figure displays the results if we once again dichotomize the outcome so that the dependent variable takes the value if a respondent indicated that they would probably or definitely keep the app and otherwise. we see that participants from the uk are relatively more likely to keep the app in this scenario while french respondents are less likely in comparison. as with the opt-in installation outcomes, the results are robust to using an ordered logit specification as well as using a probit or linear probability specification when the "keep category" only consists of people saying they would definitely keep the app. for brevity, these robustness checks are omitted here. figure using the answers to the opt-out installation question as dependent variable. the dependent variable is an indicator variable taking the value if a respondent chose definitely keep or probably keep when asked whether they would keep the app or not, and otherwise. we use a linear probability model. lines represent % confidence intervals calculated with heteroskedasticity-robust standard errors. all coefficients are the result of a single regression and thus display marginal effects. a coefficient of . implies a respondent who chose this option is percentage points more likely to state they would definitely or probably keep the app relative to the base category. . cc-by . international license it is made available under a is the author/funder, who has granted medrxiv a license to display the preprint in perpetuity. the copyright holder for this preprint this version posted may , . . in the main text, we note that the results presented in figure hold across a variety of different demographic groups -this is illustrated in the following. figure shows that support for the app does not vary significantly when splitting outcomes by gender, the existence of comorbidities or the availability of sick pay. figure shows that there are only very small differences with regard to age. the only dimension where we see significant differences in support for the app is trust in government. figure shows that someone's propensity to install a contact-tracing app decreases the less they generally trust the government to "do what is right". the results of the opt-out scenario show the exact same pattern as the opt-in results. for brevity, we do not show them here. . cc-by . international license it is made available under a is the author/funder, who has granted medrxiv a license to display the preprint in perpetuity. the copyright holder for this preprint this version posted may , . . . cc-by . international license it is made available under a is the author/funder, who has granted medrxiv a license to display the preprint in perpetuity. the copyright holder for this preprint this version posted may , . . https://doi.org/ . / . . . doi: medrxiv preprint c. take up and disease severity at the geographical level we collect data on the number of infections as well as deaths due to covid- at the geographic area by day level where the geographic area is at the same level of disaggregation as the respective survey asked for (e.g. state for the us). the data is publicly available on github, and the sources are santé publique france (for france), protezione civile (for italy), the robert koch institute (for germany), phe (for the uk), and the new york times (for the us). we adapted the aggregation level of the french data from département to region. as can be seen in figure in the main part, we generally find only a very weak relationship between the severity of the outbreak (measured by deaths per capita as well as the absolute number of deaths) in someone's area of residence and the probability that they will want to download or keep the app. the exception to this rule are the very badly affected areas of new york and northern italy -here, respondents are about percentage points more likely to state they would definitely install the app than the average participant. the fact that we do not find a strong effect generally could be due to a number of different factors. firstly, death rates do not vary much across geographic areas once we remove new york and northern italy. secondly, much of the rhetoric surrounding the severity of the disease is at the national, not sub-national level (again excluding new york and northern italy). lastly, it is likely that the cartography of infections does not exactly follow the cartography of regions at our disposal. . cc-by . international license it is made available under a is the author/funder, who has granted medrxiv a license to display the preprint in perpetuity. the copyright holder for this preprint this version posted may , . . https://doi.org/ . / . . . doi: medrxiv preprint c. reasons for/against installation and take up decisions figure shows that in general, far more reasons are given in favor of installing the app than against it -this is true even among those who state that they probably or definitely won't install the app. the number of reasons for installing the app decreases sharply as respondents become less likely to want to install it. this is not the case for reasons against, where most respondents select only one reason regardless of their intention to install. these findings suggest the need to better explain the various ways in which the app might benefit a user and his or her surroundings. figures and . while all reasons in favor of an installation appear to come somewhat evenly to people's minds, there is a strong clustering of reasons against the app around surveillance and security concerns as well as possible impacts on mental health. these reasons highlight the biggest concerns that would need to be addressed in the design and implementation of the app to make sure take-up would be sufficiently high. in addition to the common reasons displayed in figures and , we also gave a few answer options, which differed by country. in the uk, respondents were given "don't want the nhs to have access to my location data" as an additional reason against, which was replaced by "i don't want to activate bluetooth" in the other surveys. germans in particular, were quite hesitant to activate bluetooth, making it the third most popular reason against installation there. in the french, german, italian and us surveys, respondents were given "would allow me to return to normal life faster " as an additional reason in favor -however, it was not chosen by many people. finally in the us survey the additional reason against "no one else will use the app" was the second most popular reason, given by % of respondents. . cc-by . international license it is made available under a is the author/funder, who has granted medrxiv a license to display the preprint in perpetuity. the copyright holder for this preprint this version posted may , . figure shows the impact each reason had on the probability with which a respondent stated they would probably or definitely install the app -controlling for all the covariates displayed in figure . unsurprisingly, the first seven reasons, which are in favor of the app, increase the probability of intending to install the app. strikingly, out of the six reasons listed against installation (the latter half in the graph), only one had a large impact on installation decisions and that was "i would not benefit". this may indicate that while many people are concerned about government surveillance . cc-by . international license it is made available under a is the author/funder, who has granted medrxiv a license to display the preprint in perpetuity. the copyright holder for this preprint this version posted may , . . and cyber security, these reasons are not prohibitive of installing the app or only become so if they lead participants to conclude that the app would not be useful to them. we found similar results with an ordered logit model, a linear probability model dichotomizing on just definitely install, when using a probit model and dropping the controls. finally, the results are also qualitatively similar when we look at opt-out rather than opt-in decisions. figure : impact of reasons on installation probability note: the dependent variable is an indicator variable taking the value if a respondent chose definitely install or probably install when asked whether they would install the app or not, and otherwise. we use a linear probability model. lines represent % confidence intervals calculated with heteroskedasticity-robust standard errors. all coefficients are the result of a single regression and thus display marginal effects. a coefficient of . implies a respondent who chose this option is percentage points more likely to state they would definitely or probably install the app. . cc-by . international license it is made available under a is the author/funder, who has granted medrxiv a license to display the preprint in perpetuity. the copyright holder for this preprint this version posted may , . . https://doi.org/ . / . . . doi: medrxiv preprint c. determinants of the reasons chosen figures and display the main determinants of reasons given for or against installing the app. each column represents a separate linear probability model where the dependent variable takes the value for individuals that chose that specific reason and the value otherwise. the main takeaway from the graphs is that the reasons are largely not explained by the covariates. only people who trust the government less as well as younger people are significantly more likely to name concerns surrounding post-epidemic surveillance as (one of) their main reasons against installation. furthermore, younger people are less likely to cite peace of mind as a reason for installation. all results hold when using a probit model as well. figure : determinants of reasons given against installation note: each column represents a separate linear probability model with the dependent variable being an indicator variable taking the value one if a respondent gave the indicated reason. lines represent % confidence intervals calculated with heteroskedasticity-robust standard errors. all coefficients within a column are the result of a single regression and thus display marginal effects. a coefficient of . implies a respondent who chose this option is percentage points more likely to state the reason given. . cc-by . international license it is made available under a is the author/funder, who has granted medrxiv a license to display the preprint in perpetuity. the copyright holder for this preprint this version posted may , . . https://doi.org/ . / . . . doi: medrxiv preprint figure : determinants of reasons given for installation note: each column represents a separate linear probability model with the dependent variable being an indicator variable taking the value one if a respondent gave the indicated reason. lines represent % confidence intervals calculated with heteroskedasticity-robust standard errors. all coefficients within a column are the result of a single regression and thus display marginal effects. a coefficient of . implies a respondent who chose this option is percentage points more likely to state the reason given. understanding whether an opt-in or an opt-out regime will yield a larger number of app users is crucial for the successful implementation of app-based contact tracing. standard results from psychology suggest that many more people will keep the app in an opt-out regime than would install it voluntarily in an opt-in regime, as the former approach would reduce (mental) transaction costs and implicitly set having the app as the societal standard. however, figure shows that instead fewer people indicate they would keep the app rather than voluntarily download it. one reason for this may be that individuals, and in particular individuals who are concerned about potential government surveillance, may perceive automatic installation as an overreach by the government and thus choose to uninstall the app on principle. to further understand respondents' preferences over an opt-in vs. opt-out regime, we directly asked us respondents which regime they would prefer. % would prefer voluntary to automatic installation. this fraction is constant across gender, region, political affiliation, lockdown status and other characteristics. we can infer the preference between voluntary or automatic installation regime indirectly for the other four countries, where we did not ask directly for this preference. we use participants' intentions to install or keep the app in an op-in vs. opt-out regime to create an 'intention measure'. we use . cc-by . international license it is made available under a is the author/funder, who has granted medrxiv a license to display the preprint in perpetuity. the copyright holder for this preprint this version posted may , . participants' opinion of the national government in response to it introducing either regime, to create an 'opinion measure'. the opinion measure considers respondents' answers to the questions: • how much do you agree with the following statement? "my opinion of the government would improve if they introduced the app, and allowed me to decide whether to install it or not" • how much do you agree with the following statement? "my opinion of the government would improve if they asked mobile phone providers to automatically install the app, and allowed me to decide whether to keep it or not" the difference between respondents' answers to these two questions yields information about which regime they would prefer the government to introduce. therefore, we create a variable that gives the difference between these two answers, such that when participants' opinion of the government would improve more (or worsen less) under the opt-in regime, this difference is negative. in this case, we would infer that they prefer an opt-in regime. the intention measure considers respondents' answers to the "would you install this app?", and "would you keep this app if it was automatically installed" questions. the difference between respondents' answers to these questions reflects their different intentions of having the app on their phone in either regime. thus, this difference is again indicative of participants' preferences over the opt-in/opt-out regimes. as for the opinion measure, when the intention measure is negative we can infer that a respondent believes they would be more likely to have the app installed on their phone in an opt-in regime, which we take to mean that they prefer the opt-in regime. for both difference measures, we re-code the new variable into three categories: d < (prefer opt-in), d = (indifferent), and d > (prefer opt-out). the differences in preferences, across countries, are displayed in figure . the trends are broadly in-keeping with the other results. the validity of using these inferred preferences is considered in multimedia appendix subsection c. . . . cc-by . international license it is made available under a is the author/funder, who has granted medrxiv a license to display the preprint in perpetuity. the copyright holder for this preprint this version posted may , . . https://doi.org/ . / . . . doi: medrxiv preprint to investigate the factors that are associated with opt-in/opt-out preferences, we use a simple ordered logit model. the results from this analysis are presented in figure . respondents who express less trust in the government or who worry about government surveillance or their phone being hacked, are more likely to have a preference for the opt-in over the opt-out regime. figure : determinants of opt-in vs. opt-out preference note: positive coefficient means prefer opt out. ordered logit model. the points show estimated model coefficients, while the lines show % confidence intervals, calculated using heteroskedasticity-robust standard errors. in the us survey, respondents were asked directly whether they would prefer an opt-in or opt-out regime. the below cross tabulation demonstrates a strong relationship between respondents' stated choice (columns) and their inferred preference (rows), using the opinion measure. only . % of respondents' inferred preferences are inconsistent with their stated preferences -those who selected d < and automatic, or d > and voluntary. of the respondents for which the opinion measure is negative (d < ), . % stated that they prefer the voluntary regime. likewise, of the respondents for which the opinion measure is positive (d > ), . % stated that they prefer the automatic regime. . cc-by . international license it is made available under a is the author/funder, who has granted medrxiv a license to display the preprint in perpetuity. the copyright holder for this preprint this version posted may , . respondents were asked for their preferences over what should happen to the data generated by the app. respondents could state that they wanted the data immediately deleted, or that the (deidentified) data should be made available to researchers. in the final sample, . % of respondents answered that the de-identified data should be made available to researchers. figure demonstrates that respondents who were less likely to say they would download the app were also less likely to want their data to be made available to researchers. surprisingly, figure demonstrates no relationship between trust in the government and datasharing preferences. however, once we control for other covariates, this relationship does become statistically significant, as can be seen in figure . this relationship is nevertheless non-monotonic and hence difficult to interpret. figure also demonstrates that older people, people who use their phones more regularly, and residents of the uk are more likely to consent to having their data shared. figure : percentage of respondents wanting the data to be made available by intention to install by trust in government . cc-by . international license it is made available under a is the author/funder, who has granted medrxiv a license to display the preprint in perpetuity. the copyright holder for this preprint this version posted may , . . https://doi.org/ . / . . . doi: medrxiv preprint figure : determinants of wanting the data to be made available note: computed using a linear probability model. the dependent variable is a dummy variable taking the value if the respondent prefers the data be de-identified and made available, otherwise. lines represent % confidence intervals calculated with heteroskedasticity-robust standard errors. all coefficients within a column are the result of a single regression and thus display marginal effects. a coefficient of . implies a respondent who chose this option is percentage points more likely to state they prefer the data be made available. the right hand column contains answers only from respondents who answered that they would definitely or probably install the app. the reference categories for the covariates are as follows: - (age), female (gender), less than once a week (socialise), receive sick pay (sick pay), none (work from home), uk (country), and completely (trust) we asked respondents how likely they would be to comply with the request of self-isolating for days if they had been in close contact with a person who was confirmed to be infected. responses were collected on a -item scale ranging from definitely comply to definitely won't comply. as shown in figure , the vast majority of respondents in all countries said they would comply with the self-isolation request. support is again highest in italy, where % of respondents declared they would definitely or probably comply, and lowest in germany, at a still very high %. we further asked respondents who did not say they would definitely comply, whether their chances of compliance would increase, decrease, or remain the same if the health services in their own country committed to test them quickly. in all countries, a commitment to quick testing would further increase compliance rates. this implies that the vast majority of people in all five countries are not only prepared to have the app installed on their phones, but also to use it as intended, even if this means sacrificing (more of) their personal freedom for a limited amount of time. as this survey asks hypothetical questions, we should caveat that actual compliance might be considerably lower. . cc-by . international license it is made available under a is the author/funder, who has granted medrxiv a license to display the preprint in perpetuity. the copyright holder for this preprint this version posted may , . . in the german replication survey, conducted via forsa (see section b. ), we collected all answers also for those respondents who failed to answer the comprehension questions correctly. for this survey, we can thus compare the willingness to install between respondents who did (n= ) and who did not (n= ) answer all comprehension questions correctly. we find that . % of the respondents who answered the comprehension questions correctly would definitely or probably install the app. in contrast, only . % of the respondents who answered wrongly would do so. this might be because the respondents who failed the comprehension questions answered randomly. or it could be that these respondents are actually less willing to install. the latter interpretation is consistent with previous literature [ ] that showed a higher willingness to share data and information among the more technically informed users. a lack of understanding leading to a lower willingness to install would also be in line with the results of surveys that were conducted in germany shortly after our survey. some of these surveys did not explain the app in as much detail as our survey and these surveys often find lower willingness to install. for example, infratest dimap asked respondents between and march about their willingness to install a contact-tracing app, and found that only % of respondents were willing to install the app. the comparison between our surveys and these shorter surveys suggests that a detailed explanation of how the app works and how it could mitigate or stop the epidemic increases the willingness to install. the exact question (translated from german) was: "assume that there is an app for your mobile phone that would allow you to track your covid- symptoms as well as document your geolocation data. the app could thus show all users whether they had been close to a person infected with the coronavirus. would you use such an app or not?" . cc-by . international license it is made available under a is the author/funder, who has granted medrxiv a license to display the preprint in perpetuity. the copyright holder for this preprint this version posted may , . . https://doi.org/ . / . . . doi: medrxiv preprint . cc-by . international license it is made available under a is the author/funder, who has granted medrxiv a license to display the preprint in perpetuity. the copyright holder for this preprint this version posted may , . . https://doi.org/ . / . . . doi: medrxiv preprint d survey questionnaire in the following, the uk version of the survey is presented with comments (in italics) explaining all the ways in which the italian, french, german or us surveys differed from it. overall, we tried to keep the surveys as similar as possible and only changed the phrasing if necessary (for example due to different political circumstances as some countries already were in lockdown at the time of the survey while others were not). in the us, we added a few more questions, which appear teal-colored in the overview below. finally, horizontal lines indicate page breaks in the survey. the original survey texts for each country can be found here: uk, france, germany, italy and us. in this study, we will ask you about an app that could help reduce the spread of the covid- epidemic. you may ask any questions before deciding to take part by contacting the researchers (details below). the survey is about minutes long. no background knowledge is required. do i have to take part? please note that your participation is voluntary. if you do decide to take part, you may withdraw at any point during the survey for any reason before submitting your answers by closing the browser. how will my data be used? your answers will be completely anonymous. your data will be stored in a password-protected file and may be used in academic publications. research data will be stored for a minimum of three years after publication or public release. who will have access to my data? lucid is the data controller with respect to your personal data and, as such, will determine how your personal data is used. please see their privacy notice here: https://luc.id/privacy-policy/. lucid will share only fully anonymised data with the university of oxford, for the purposes of research. responsible members of the university of oxford and funders may be given access to data for monitoring and/or audit of the study to ensure we are complying with guidelines, or as otherwise required by law. this project has been reviewed by, and received ethics clearance through, the university of oxford central university research ethics committee (reference number econcia - - ). who do i contact if i have a concern about the study or i wish to complain? if you have a concern about any aspect of this study, please contact johannes abeler at johannes.abeler@ economics.ox.ac.uk and we will do our best to answer your query. we will acknowledge your concern within working days and give you an indication of how it will be dealt with. if you remain unhappy or wish to make a formal complaint, please contact the chair of the research ethics committee at the university of oxford who will seek to resolve the matter as soon as possible: economics departmental research ethics committee at ethics@economics.ox.ac.uk please note that you may only participate in this survey if you are years of age or over. if you have read the information above and agree to participate with the understanding that the data (including any personal data) you submit will be processed accordingly and that you need to be years of age or over to participate, please confirm below. • i confirm • i do not confirm . cc-by . international license it is made available under a is the author/funder, who has granted medrxiv a license to display the preprint in perpetuity. the copyright holder for this preprint this version posted may , . . https://doi.org / . / . / in the us version, we asked the following filtering question on a separate screen: is your area of residence currently under a stay-at-home order where you are no longer allowed to leave your home for non-essential reasons? depending on the answer, participants continued with slightly different survey versions -one making reference to current restrictions and one referring to possible future restrictions. below we will point out the few instances where this leads to minor differences in formulations. the first screen was the same across all surveys, the only difference being that in the us survey we said "if enough people used the app, it could automatically alert you.". the current coronavirus epidemic ("covid- ") is all over the news. people can get infected if they are in close contact with someone who has the virus. people do not notice when they get infected. they only notice when they start having a fever or a cough, perhaps a week later. imagine there was an app that you could install on your mobile phone. this app would automatically alert you if you had been in close contact for at least minutes with someone who was infected with the coronavirus. such an app does not exist yet in the uk. but we, researchers from the university of oxford, are interested in understanding what you would think about such an app. the next pages explain how such an app could work and will ask comprehension questions. you can only continue the survey if you answer all questions correctly. in the german, italian and french versions, we said the app would only use bluetooth and not location data. in every other country, we therefore listed "activate bluetooth" as the first (and correct) answer option in the comprehension question. furthermore, we adapted the health services responsible for the app depending on the country, referring to the center for disease control and prevention (cdc) in the us, the robert koch-institut (rki) in germany and more broadly to the "health services" in italy and france. the app would be developed by the nhs. you would need to install the app by simply clicking a link. once installed, the app would register which other users are close to you. the app would do this by using bluetooth and your location. the app would not access your contacts, photos, or other data held on your phone. only the nhs would have access to the data collected. . cc-by . international license it is made available under a is the author/funder, who has granted medrxiv a license to display the preprint in perpetuity. the copyright holder for this preprint this version posted may , . participants were only allowed to continue to the next screen if they selected option ( ). if they chose one of the latter two options, the survey was terminated. in the us, germany, italy and france, we said the app would request people found to have been in contact with a confirmed case of covid- to go into "quarantine at home" rather than "self-isolate" -terminology we stuck to throughout the respective surveys. we also explained the difference between the restrictions imposed on people in the existing lockdowns and the restrictions associated with being alerted by the app. in the uk, we did not consider this to be necessary as, at the time of the survey, the government had not issued a lockdown order yet. finally, in the us survey, the "lockdown version" referenced a current stay-at-home-order while the "no lockdown version" mentioned a hypothetical one. if the nhs diagnoses the coronavirus in somebody you have been in close contact with, the app would notify you automatically. the app would give you targeted advice on what to do. it will ask you to self-isolate at home for days or until you have been tested for the virus. this would be useful since people can infect others even before they have a fever or a cough. selfisolating would thus protect your family, friends and colleagues from being infected by you. at the same time, only people who were in contact with an infected person would need to self-isolate. if you had not been in close contact with a confirmed case, then the app would show you an "all clear" message. comprehension check: what would the app do if you were found to have been in contact with someone diagnosed with coronavirus? • ask me to self-isolate • give me an "all clear" message • tell me the name of the person who was diagnosed participants were only allowed to continue to the next screen if they selected option ( ). if they chose one of the latter two options, the survey was terminated. in the us, italy, france and germany, we stressed that, if enough people used it, the app could also shorten the duration of existing lockdowns (or school closures in the case of the "no-lockdown" us version). if you are diagnosed with coronavirus, the app would notify all people you have been in close contact with, without identifying you to them, and advise them to self-isolate. this would increase the chance of finding all the people you might have infected and help make sure they can keep their loved ones safe as well. if enough people use the app, it will slow down the epidemic and might even stop it entirely. comprehension check: what would the app do if you were diagnosed with the coronavirus? • give my name and address to all people i have been in close contact with . cc-by . international license it is made available under a is the author/funder, who has granted medrxiv a license to display the preprint in perpetuity. the copyright holder for this preprint this version posted may , . . • advise all people i have been in close contact with to self-isolate • shut down my phone participants were only allowed to continue to the next screen if they selected option ( ) . if they choose the first or the last option, the survey was terminated. the general installation questions were the same in all countries. however, in the us, half of the participants were randomly assigned to see all the answer options in the survey in ascending rather than descending order i.e., the answer options were ordered from "definitely won't install" to "definitely install" (with "don't know" remaining the last option). whether or not a participant saw the ascending or descending order remained the same throughout the survey. for the following questions, please imagine that an app like the one described before exists. how likely would you be to install, or not install, the app on your phone? • definitely install • probably install • may or may not install • probably won't install • definitely won't install • don't know the ordering of the different reasons (both for and against installation) was randomized and people could choose multiple answers. in the non-uk versions, we gave one more reason for installing the app: "it would allow me to return more quickly to a normal life". this is because in every country but the uk, the government had already issued a (partial) lockdown order. what would be your main reasons for installing the app (you may click up to five)? it would help me stay healthy it would let me know my risk of being infected seeing the "all clear" message would give me peace of mind it would protect my family and friends it would help reduce the number of deaths among older people a sense of responsibility to the wider community it might stop the epidemic other (please indicate in the field below): in italy, we gave "i worry the government would use this as an excuse for greater surveillance during the epidemic" (not just after) as an additional reason against, while in the us, we gave "i don't believe other people will install it" as an additional reason against. plus, in every non-uk country, we used "i don't want to activate bluetooth" rather than "i don't want the [health services] to have access to my location data" as a reason against. what would be your main reasons against installing the app (you may click up to five)? . cc-by . international license it is made available under a is the author/funder, who has granted medrxiv a license to display the preprint in perpetuity. the copyright holder for this preprint this version posted may , . . https://doi.org/ . / . . . doi: medrxiv preprint . cc-by . international license it is made available under a is the author/funder, who has granted medrxiv a license to display the preprint in perpetuity. (which was not certified by peer review) the copyright holder for this preprint this version posted may , . . https://doi.org/ . / . . . doi: medrxiv preprint participants were only shown this entire section if they did not select "definitely install" in response to the question "how likely would you be to install, or not install, the app on your phone?" at the beginning of the main questionnaire. many people in the uk worry about the effect of the virus on their community and on their family and friends. suppose someone in your community had been infected with the virus. how likely would you then be to install, or not install, the app on your phone? • definitely install • probably install • may or may not install • probably won't install • definitely won't install • don't know only participants who did not select "definitely install" in response to the previous question were shown this screen. now suppose someone you personally know had been infected with the virus. how likely would you then be to install, or not install, the app on your phone? • definitely install • probably install • may or may not install • probably won't install • definitely won't install • don't know in italy, france and germany as well as areas under a stay-at-home order in the us, we dropped the "imagine the government would introduce.." intro and instead only said: "imagine the government decided to lift the restrictions of the current lockdown for those people for whom the app showed an all clear message. this means they would be able to leave their homes even without an essential reason." imagine the government decided to introduce a full lockdown as in italy to limit the spread of the coronavirus. this would mean that only essential stores, like supermarkets, would remain open, and you would only be allowed to leave your house in exceptional circumstances. imagine that this restriction would be lifted for those people for whom the app showed an "all clear" message. in this situation, how likely would you be to install, or not install, the app on your phone? • definitely install • probably install • may or may not install • probably won't install • definitely won't install • don't know . cc-by . international license it is made available under a is the author/funder, who has granted medrxiv a license to display the preprint in perpetuity. (which was not certified by peer review) the copyright holder for this preprint this version posted may , . . https://doi.org/ . / . . . doi: medrxiv preprint . cc-by . international license it is made available under a is the author/funder, who has granted medrxiv a license to display the preprint in perpetuity. (which was not certified by peer review) the copyright holder for this preprint this version posted may , . at the end of the epidemic, a decision would need to be made about what to do with the data collected. which of the following policies would you prefer? • all data will be automatically deleted at the end of the epidemic and not used for any other purpose • all data will be de-identified and made available to university researchers to prepare for future epidemics • other (please indicate in the field below): how old are you? • in all countries except the uk, we asked about state or region of residence, while in the uk, we used a higher level of aggregation. where in the uk do you currently reside? . cc-by . international license it is made available under a is the author/funder, who has granted medrxiv a license to display the preprint in perpetuity. (which was not certified by peer review) the copyright holder for this preprint this version posted may , . . https://doi.org/ . / . . . doi: medrxiv preprint . cc-by . international license it is made available under a is the author/funder, who has granted medrxiv a license to display the preprint in perpetuity. (which was not certified by peer review) the copyright holder for this preprint this version posted may , . . https://doi.org/ . / . . . doi: medrxiv preprint us participants only saw this screen if they selected "newspaper" as their main source of information in the question about news. which newspaper do you read most frequently (either in print or online)? • the new york times to what extent do you agree with the following statement: "i generally trust the government to do what is right."? • fully agree • somewhat agree • neither agree nor disagree • somewhat disagree • fully disagree to what extent do you agree, or not, with the following statement: "my opinion about the british government would improve if they introduced such an app and allowed me to decide myself whether to install it or not."? • fully agree • somewhat agree • neither agree nor disagree • somewhat disagree • fully disagree to what extent do you agree, or not, with the following statement: "my opinion about the british government would improve if they asked mobile phone providers to automatically install such an app on all phones to maximise the chance of stopping the epidemic."? . cc-by . international license it is made available under a is the author/funder, who has granted medrxiv a license to display the preprint in perpetuity. (which was not certified by peer review) the copyright holder for this preprint this version posted may , . in the us version, we asked an additional, direct question about which installation regime respondents would prefer. imagine that the federal government introduced such an app -which of the two installation regimes described above would you prefer? • voluntary installation • automatic installation (with an option to uninstall) in each country, we referred participants to the coronavirus information website of the relevant government agency. thank you very much! if you have any questions about the app or any feedback, please let us know by writing them into the field below. you can also email the researchers at johannes.abeler@economics.ox.ac.uk. if you want to know more about the coronavirus and how to protect you and your family, please click this link to the nhs coronavirus website: https://www.nhs.uk/coronavirus please click the button below to finish the survey. . cc-by . international license it is made available under a is the author/funder, who has granted medrxiv a license to display the preprint in perpetuity. (which was not certified by peer review) the copyright holder for this preprint this version posted may , . . https://doi.org/ . / . . . doi: medrxiv preprint impact of non-pharmaceutical interventions (npis) to reduce covid mortality and healthcare demand impact assessment of non-pharmaceutical interventions against coronavirus disease and influenza in hong kong: an observational study economic outlook sectoral effects of social distancing impact of school closures for covid- on the us healthcare workforce and net mortality: a modelling study inequality in the impact of the coronavirus shock: evidence from real time surveys quantifying sars-cov- transmission suggests epidemic control with digital contact tracing". en nhs coronavirus app to target % of smartphones". en-gb. in: bbc news library catalog: www.pepp-pt.org effective configurations of a digital contact tracing app: a report to nhsx". en covid- contact tracing and data protection can go together". en second-level digital divide: differences in people's online skills". in: first monday apple and google partner on covid- contact tracing technology. en. library catalog: www.blog.google do defaults save lives? en. ssrn scholarly paper id contact tracing mobile apps for covid- : privacy considerations and related trade-offs tracing contacts to control the covid- pandemic validating vignette and conjoint survey experiments against real-world behavior". en do intentions really predict behavior? self-generated validity effects in survey research". en willingness to pay versus willingness to vote: consumer and voter avoidance of genetically modified foods". en don't get it or don't spread it? comparing self-interested versus prosocially framed covid- prevention messaging. preprint. psyarxiv on the benefits of explaining herd immunity in vaccine advocacy". en willingness to participate in passive mobile data collection". en collecting survey and smartphone sensor data with an app: opportunities and challenges around privacy and informed consent:" en willingness to use mobile technologies for data collection in a probability household panel". en participation in a mobile app survey to collect expenditure data as part of a large-scale probability household panel: coverage and participation rates and biases the app would be too much hassle to install i would not benefit from the app i don't want the nhs to have access to my location data i won't be infected anyway i worry the government would use this as an excuse for greater surveillance after the epidemic i don't want to feel more anxious than i already feel i worry that my phone will be more likely to get hacked other how likely would you be to comply with the recommendation of the app to self-isolate at home for days if you had been in close contact with an infected person? • definitely comply • probably comply • may or may not comply • probably won only participants who did not select "definitely comply likely to comply with the advice to self-isolate for days if the nhs committed to test you for the virus within days from the start of your self-isolation? • more likely • equally likely • less likely in the us version, participants were randomly assigned to see either a question about mark zuckerberg endorsing the app in a letter to all users or facebook promoting the app on their website to install such an app, developed by the cdc, if a private company like facebook officially endorsed it, for example in a letter from their ceo mark zuckerberg? • more likely • equally likely • less likely • i don to install such an app, developed by the cdc, if a private company like facebook promoted it on their website? • more likely • equally likely • less likely • i don't have a facebook account automatic installation in every other country, we stressed that people would be able to immediately uninstall the appboth in the intro and the first question now, imagine the government asked the mobile phone providers (vodafone, ee, etc.) to automatically install the app on all phones to what extent do you agree, or not, with the above statement? • fully agree • somewhat agree • neither agree nor disagree • somewhat disagree • fully disagree only participants who did not select "fully agree" in response to the previous question saw this screen only participants who did not select "fully agree now suppose someone you personally know had been infected with the virus. to what extent do you agree, or not, with the above statement? • london • england not including london • wales • scotland • northern ireland in the us version, we asked an additional question about the environment participants resided in. how would you best describe the area you currently reside in? • urban • suburban • rural in all other countries how often are you currently in close contact with people outside of your household, for example, at work or socially? • not more than once per week • a few times per week • a few times per day • many times per day do you have any of the following health problems: diabetes, high blood pressure, heart or breathing problems? • yes • no in the us version, we asked an additional question about health insurance. do you currently have health insurance? • yes, i have private health insurance (e.g. by the employer) • yes, i have public health insurance germany and for areas under a lockdown in the us, we dropped the "imagine about half of my normal work • about three quarters of my normal work • all of my normal work • i do not work or study would you receive sick pay or continue to receive your income if you stayed and worked from home? • yes • no • don which of the following political parties do you feel closest to?". in the us, we asked which party respondents voted for in the last presidential election and displayed the question on a separate screen do you think of yourself as a supporter of any one of the following political parties? • conservatives • liberal democrats • labour • brexit party • snp • plaid cymru • other or i don't want to say in the us version, we asked participants about their main source of information us participants only saw this screen if they selected "television" as their main source of information in the question about news. which television channel do you watch most frequently for the news? we acknowledge funding from the economic and social research council (grant es/r / ), the university of oxford and volkswagen foundation (grant "consequences of artificial intelligence for urban societies"). key: cord- -u z rir authors: ranisch, robert; nijsingh, niels; ballantyne, angela; van bergen, anne; buyx, alena; friedrich, orsolya; hendl, tereza; marckmann, georg; munthe, christian; wild, verina title: digital contact tracing and exposure notification: ethical guidance for trustworthy pandemic management date: - - journal: ethics inf technol doi: . /s - - - sha: doc_id: cord_uid: u z rir there is growing interest in contact tracing apps (ct apps) for pandemic management. it is crucial to consider ethical requirements before, while, and after implementing such apps. in this paper, we illustrate the complexity and multiplicity of the ethical considerations by presenting an ethical framework for a responsible design and implementation of ct apps. using this framework as a starting point, we briefly highlight the interconnection of social and political contexts, available measures of pandemic management, and a multi-layer assessment of ct apps. we will discuss some trade-offs that arise from this perspective. we then suggest that public trust is of major importance for population uptake of contact tracing apps. hasty, ill-prepared or badly communicated implementations of ct apps will likely undermine public trust, and as such, risk impeding general effectiveness. digital technologies are increasingly being discussed and implemented for covid- pandemic management and as tools for easing restrictive measures, such as lockdowns (mello and wang ; ting et al. ) . due to the high penetration rate of smartphones, there has been a huge interest in mobile phone data as a source for public health research and measures (oliver et al. ) . to track the spread of the virus, in europe and elsewhere, network operators share (anonymized and aggregated) phone location data. apple and google, two leading providers of smartphone operating systems, release data to show mobility trends in countries and selected regions google ) . in addition, a range of new mobile phone based applications ("apps"), sometimes lumped together under the term "covid- apps", have been rolled out recently or are being under development by private as well as public actors (sharma and bashir ; privacy international ; woodhams ; gdprhub ) . these apps may serve a variety of functions: provide users with covid- -related information, monitor people in quarantine, trace movements, or give users rapid warning of potential exposure to sars-cov- (gdprhub ; rimpiläinen et al. ) . frequently, mobile phone apps are designed to fulfil more than just one purpose, e.g. symptom checkers could generate data which might also be used for epidemiological modelling, monitoring the virus spread or to evaluate public health measures. available apps differ widely regarding data use (e.g. self-reported, geolocation data, proximity tracing), data sources (e.g. gps, bluetooth), data handling (decentralized or centralized), as well as data protection (anonymization or pseudonymization) (woodhams ). proximity or contact tracing apps (ct apps) have gained notable attention so far. ct apps notify users if they have been in proximity to confirmed infected people and propose next steps (e.g. self-isolation, testing). a vital distinction to be made here is between apps that collect data on-in principle-identifiable individuals in a centralised database ('centralised' variants) and those that function by use of encrypted identifiers that connect individual users to each other ('decentralised' variants) . only the first allows 'contact tracing' in the stricter sense when individuals and encounters are retrospectively identified by a third party. the second variant warns users in the case of contact with infected individuals (i.e. exposure notification), but does not allow a centralized tracing of possible infection chains. both variants, contact tracing and exposure notification, can play an important role in a digitally supported pandemic management strategy. since analogous contact tracing is comparatively slow, resource intense and lacks reliability, digital proximity tracing has been proposed as a complementary tool to indicate possible transmission chains that analogous contact tracing might miss or take a longer time to identify. one study suggests that ct apps could, in theory, effectively decrease virus transmission by enabling targeted testing or quarantine, and thus avoid mass confinements or lockdowns (ferretti et al. ) . informing or identifying potential spreaders earlier could reduce pre-symptomatic transmission, i.e. before an infected person shows symptoms. this might also support micromanagement after lifting restrictive pandemic control measures or during future infection waves. until today, however, little is known about the effectiveness and efficiency of ct apps in real-world settings, and whether or not they could also have negative effects on pandemic management, or expose individuals to ethical downsides, such as lack of data protection. as of august , a wide range of ct apps is being used or under development globally, from algeria to vietnam . singapore pioneered a bluetooth based open-source technology named bluetrace, which underpins the tracetogether app (tracetogether ). in europe, after a joint attempt to establish a pan-european "privacy-preserving approach" for ct (pepp-pt) has seemingly failed, various countries have rolled out their own proximity tracing apps. initially, pepp-pt and a centralized database were considered as the preferred framework for ct apps, but massive criticism (e.g. joint statement ) has led some policy makers to switch to a decentralised approach (busvine and rinke ) . however, european countries are divided on the question whether to rely on centralized (e.g. france) or decentralized (e.g. germany) data management, making interoperability between different frameworks difficult. by now, the authorities of many european countries like austria, belgium, denmark, germany, italy, ireland and switzerland opted for a decentralised approach based on a joint api from apple and google ). there are plans from the european commission to build a gateway to allow cross-border exchange of information between these national ct apps. development of ct apps is not only promoted by public agencies but often relies on public-private partnerships with relevant corporate actors. notably, apple and google have collaborated to develop a joint contact tracing framework which is also founded on decentralized data management (apple and google ) . such efforts are important to guarantee interoperability between different smartphone systems and allow building efficient ct apps. however, due to this dependency, commercial companies are gaining a wide-ranging influence on the national strategies for digital contract tracing; e.g. the apple and google framework is only of limited use for countries which have opted for a centralized architecture for ct apps like france (scott et al. ). furthermore, the use of the framework is restricted to only one tracing app per country (gurman and de vynck ) . ct apps may prove to be valuable public health tools, but they also raise significant concerns (gasser et al. ; lucivero et al. ) . as part of the covid- pandemic response, advisory bodies, ngos, and expert initiatives have interrogated the ethical aspects of digital surveillance technologies, including ct apps (e.g. algorithmwatch ; amnesty international et al. ; chaos computer club ; human rights watch ; swiss national advisory commission on biomedical ethics ; who ). the first ethical frameworks for digital tools in the context of covid- have been proposed (mello and wang ; gasser et al. ; lucivero et al. ; kahn et al. ; morley et al. ; parker et al. ) , and the european commission ( ) has drafted various recommendations and guidelines for digital contact tracing in the eu. this paper focuses on ethical considerations for responsible development, design and implementation of effective and justifiable ct apps in pandemic management strategies. it considers legal and digital ethical concerns in a broader framework of public health ethics as well as related pragmatic and procedural considerations. it provides a framework for ethical analysis of concrete proposals, and suggests that to strengthen trustworthiness, policy makers need to be sensitive to the multi-faceted complexities of public health decision making. the viability of ct apps as a useful pandemic-response measure, depends on a complex interplay of criteria, such as pragmatic assumptions about effectiveness, the likelihood of public health benefit, technological specifications, legal requirements etc. to minimise the risk of adverse outcomes, ethical standards should guide and complement the process of development (ethics by design), implementation, use, and evaluation of ct apps. rather than asking general questions on the moral acceptability of ct apps, the crucial question is: "what specific interventions, if any, may be justified under what conditions?" inspired by ethical frameworks for big data in health and research, developed by the shapes initiative (xafis et al. ) , and other normative frameworks for digital health technologies (marckmann ) and pandemic management (thompson et al. ) , we propose relevant substantive values (which evaluate the outcome of measures) and procedural values (which guide decisionmaking) as well as corresponding questions, which should be considered in response to these requirements (table ) . the list of considerations provides a sketch of the complex set of criteria relevant to assessing ct apps as ethically justifiable public health tools. we neither claim that the list is complete, nor do we think that a responsible policy-making process should necessarily address all of them. on the contrary, it is highly unlikely that a solution would satisfy all these demands. not only is there a significant lack of available data and real-world experience regarding ct technologies, all pandemic management strategies will involve several trade-offs. but acknowledging the ethical values and specific questions can help during development, implementation and evaluation of ct apps in order to find ethically appropriate solutions. in what follows, we will describe some of the complexities in implementing ct apps (cf. nijsingh et al. ). considering the wide variety of mobile applications being developed in the context of the covid- pandemic, it is crucial to distinguish between different apps, their functions, purposes, and performance. the value of mobile applications being developed in the context of the covid pandemic essentially depends on specific pandemic contexts and factors such as the social and political environment, how ct apps are integrated into a comprehensive strategy of pandemic management, as well as possible and available alternatives (fig. ) . notably, and as we will demonstrate, the implementation of digital contact tracing may involve moral costs. in some countries, apps and other mobile based surveillance measures are imposed on people, leading to an infringement of privacy rights (human rights watch ). even without compulsion, ct apps can have severe consequences for social values: worries range from issues of data protection, to possible stigmatization of patients, social justice concerns, or function creep (woodhams ; hart et al. ) . nevertheless, risks that cannot be easily mitigated or avoided could still be acceptable, considering the severity of a pandemic situation, the importance of effective contact tracing to manage it, and the scope of established measures to stop virus transmission. to assess whether a certain ct app is justified, its use needs to be compared to available alternative strategies. from this perspective, infringements associated with a possible loss of privacy and risks related to an effective ct app may appear justifiable in light of the enormous costs in terms of welfare, liberty and health outcomes of either letting the virus run its course or maintaining comprehensive restrictions or lockdowns . to make a case in favour of a ct app, however, several conditions must be met. sufficient societal need and potential effectiveness need to be demonstrated, and ethical risks sufficiently mitigated in order to demonstrate proportionality. in addition, such evaluation and decision-making needs to demonstrate procedural fairness, with transparency and opportunity for potentially concerned parties to voice concerns. finally, the balance of reasons for and against needs to be superior to alternative solutions or strategies. here, again, context matters. for a ct app scheme to be worth its costs and risks, a society needs to be in a pandemic stage, in which contact tracing is a priority. this may depend both on the pattern of (community) transmission, and the healthcare capacity of this country relative to the transmission pattern. public health benefit is the pandemic situation such that contact tracing activity is motivated from a public health standpoint? is the general use of the ct app likely to enhance the effectiveness of contact tracing measures? is the technological make-up of the app such that it can actually produce public health benefit? is the pool of potential users who are willing to use a ct app large enough for epidemiological effectiveness? harm minimisation are ct apps the least harmful way of obtaining the desired benefits? are ct apps easy to use and do they minimise confusion or stress by design? has the risk of self-and social stigma effects, implicated by an elevated focus on one's or others' health status been considered and mitigated? are safeguards in place to mitigate the vulnerability of and harm to marginalized groups from ct apps and related public health and security measures? are potential, harmful social effects related to the app (widespread anxiety, ineffective quarantines etc.) adequately considered? privacy are measures in place for data protection and against data loss or misuse?are data security authorities involved? is data parsimony guaranteed and access to non-essential personal data minimised? are the most privacy-preserving solutions (e.g. no real-time data, anonymization) prioritised? is collection of the tracing-data temporary (e.g. will it be deleted after a certain, specified amount of time are social, and moral costs of ct apps proportionate to the pandemic threat and the expected effectiveness of using the app? is the cost-effectiveness of the ct app positive compared to alternative pandemic management strategies? are financial costs proportionate to the expected public health benefits? general trustworthiness are democratic procedures in place to guide decision making? can population uptake be assumed? do stated objectives of ct apps align with proposed measures? if a society is not in such a state, no app will be able to promote better contact tracing. in addition, the utility of ct apps largely depends on broader public health measures beyond digital technologies. for ct apps to contribute to an effective public health strategy, sufficient staffing of public health services as well as reliable infrastructures (e.g. for testing and for quarantine) are needed. to avoid false positive self-reports, health departments or other institutions need to confirm infection status of users. for 'centralised' ct apps, the data generated by the app needs to be collected and analysed in a meaningful and cost-effective way (from a public health perspective) in relation to a set of justified effective tracing actions that are thereby being facilitated (i.e. eased or made possible by the app data). for 'decentralised' apps, additional efforts of analogous contact tracing are necessary, because possible transmission chains are not tracked in a way to be accessible for health authorities. all ct apps require well-organised institutional efforts. little is known about the effectiveness of contact tracing apps in the real-world setting (anderson ) . even for countries with a high penetration rate of proximity tracing technologies such as iceland, the contribution of ct apps to suppressing the pandemic has been questioned (johnson ) . besides the risks of false positives (which can impose burden on unaffected individuals) and false negatives (which may lead to a false sense of security), the implementation of an ineffective app has opportunity costs: wasting time and resources, undercutting other solutions and leading to wrong political decisions. this may result in a sub-optimal approach to pandemic control, leading to higher morbidity and mortality and greater economic damage. it is also crucial to view the value of a ct app in regards to the quality of information produced by it: mobile phones are not well equipped for contact tracing of individuals. bluetooth signals, which are central for the now widely used approach supported by apple and google, only allow a rough estimation between devices (leith and farrell ) . the same is true if location data (gps) is used. apps that would rely on user-generated subjective information are also likely to produce false predictions that could affect particular tracing policies. this concerns both false positives and false negatives. as such, incorrect information will rather compromise than support particular public health measures, as well as health care systems more generally, and scarce resources may be wasted, or used suboptimally. by contrast, ct apps that appear to be effective in tracing individuals, may raise more severe privacy concerns (baumgärtner et al. ). it has been reported from south korea, where multi-source tracing and tracking technologies are being used (gdprhub ), that information was so detailed as to allow re-identification of individuals (zastrow ) . hence, the values of effectiveness and privacy need to be carefully balanced in digital public health measures. for example, while infringements on individual rights or liberties could be justified to secure health benefits, measures always need to be proportionate and aim for careful balance between competing values and considerations. effectiveness does not only presuppose a favourable context in terms of a suitable pandemic stage and accompanying interventions, but also sufficient uptake. for ct apps to offer a meaningful contribution to pandemic management, a large part of the population needs access to compatible mobile technologies (e.g. newer smartphones or beacons), install and set up the app, and be willing and able to use tools correctly. a study from the uk has estimated that to stop the pandemic on its own, around % of smartphone users (more than % of population overall) would have to use a ct app (hinch et al. ) , i.e. a user rate comparable to whatsapp or facebook messenger in some european countries. as mentioned, so far the highest penetration rate of ct apps in the world has been reported from iceland, where almost % of the overall population downloaded a ct app. for singapore's much heralded ct app, less than a quarter of the population are using this tool (tracetogether ). at the point of writing in early august germany had introduced a ct app less than two months ago, and download numbers had reached more than million, approximately % of the overall population (robert koch institut ). a lower adoption rate still has some positive effect for targeted testing and quarantine hinch et al. ) . nevertheless, population uptake is a bottleneck for success of these digital technologies. predicting future uptake of ct apps is difficult and depends on various factors, such as the penetration range rate of digital technologies in a society, the possibility to download and use the app on different types of smartphones, the credibility of institutions offering these solutions, and viable solutions for ethical concerns such as data security. recent surveys have been inconclusive about the possible uptake in different countries. a study showed a high level of support (around %) for ct apps in countries such as the uk, germany, france and the us (milsom et al. ) , while other surveys from the us and germany came to a less optimistic conclusion (anderson and auxier ; covid- snapshot monitoring ). the available data also show that some aspects could reduce the acceptability of ct apps: these include concerns about further continuation of surveillance after the pandemic and data security (anderson and auxier ) . one way to increase uptake is, of course, to force people to download and use ct apps. mandatory use of disease surveillance tools and possible moral obligations to comply with them are being discussed (lucivero et al. ; parker et al. ; schaefer and ballantyne ) . coercion, however, adds ethical downsides of liberty restrictions that are seen as substantial in a liberal democratic context, and thereby complicates the justification of a ct app policy. moreover, compulsory measures may undermine public trust and create incentives for cheating (floridi ) , necessitating even more forceful steps to secure the benefits of the policy. as a consequence, then these benefits need to be even more pronounced and certified in order to create a potential for the policy to be proportional. for this reason, ct app programs based on voluntary use with a good uptake appear preferable. but this assumes strong public trust in the apps and the program (ienca and vayena ; parker et al. ) . trust, however, must build on trustworthiness, and thus needs to be backed up by responsible design and corresponding policies. such "well founded" credence (parker et al. ) also remains a strong indicator that choices are self-determined and, thus, in line with democratic values. meanwhile, reports from china and other nations have already shown that digital measures utilised in the covid- pandemic response have been used for mass surveillance (woodhams ; human rights watch ) and that there might be plans to massively extend the use of newly established apps even after pandemic (davidson ) . in some countries such as sweden or the netherlands, the launch of ct apps has been postponed or even cancelled due to weak data security and doubts about effectiveness and concerns on the legality of apps that process sensitive personal information (wassens ; hagberg ) . such evidence might have already fuelled public mistrust in ct apps in other nations, especially in societies, in which trust in science and governance is limited. for countries like germany, public outreach by the political representation regarding the introduction of different apps has created confusion (barker ) . internationally, ct apps have already become the subject of conspiracy theories, fake news, and scams. from the perspective of liberal values, citizens should ideally support ct apps because they have (justified) faith in public health measures and, thus, freely choose to utilise disease surveillance technologies. this, however, does not rule out some measures to increase population uptake (floridi ) : encouragement, campaigning, nudges and even some stronger forms of incentives could be justified to increase adoption rates. possible benefits should be equally accessible for most citizens without disproportionate burdens, and negative incentives must not be so severe as to render ct apps de facto compulsory, for example by limiting access to essential infrastructures (lucivero et al. ; morley et al. ) . incentives can also create new risks, e.g. owing to users' psychological responses to the information regarding user-surroundings and related health risks disclosed by a particular app. a privacy infringing, unfair or burdensome app may trigger negative responses, particularly if it is perceived as being imposed upon the public. uptake depends in part on the level of trust in agencies responsible for development, marketing, and distribution of ct apps, on solving issues e.g. of data protection or stigmatization, but also on the usefulness and performance of digital proximity tracing itself. since using ct apps could have adverse consequences for individuals, for example by requiring tests and imposing isolation measures, demonstrated effectiveness and validity of ct apps will be a major factor for population uptake. trust, however, cannot be quickly established, or specifically for just one public health intervention (ward ) . it is a long-term endeavour and requires constant efforts to uphold it, e.g. through transparent communication and participatory elements in health care planning. this raises a pragmatic dilemma regarding the factor of trust: on the one hand, the effectiveness of ct apps is uncertain. on the other hand, digital proximity tracing essentially depends on population uptake and user adherence. broad scepticism about the effectiveness of digital contract tracing could eventually become a self-fulfilling prophecy. this pragmatic dilemma must therefore also be incorporated into ethical considerations. for if the probability of uptake and thus of effective pandemic control with the app is too small, the risks and moral costs of the app could be too high. for an ethically appropriate introduction of an app, that also maintains or increases well-founded trust, the functions, goals, possible chances, and risks associated with specific ct apps must be communicated clearly, as well as the measures taken to mitigate the risks. the same goes for disclosure of conflicts of interest and the procedural management of state-business relationships linked to commissions of technological development and procurement of technical products. this last aspect becomes especially important if the decision is to adopt one particular national ct app solution and policy, meaning that private developers will be in serious competition to win the race for a state contract. to increase app uptake, focusing efforts on one single ct app with just one (or a limited number of) clearly defined purpose(s) and broad support from political and health institutions may be crucial. to prevent confusion and loss of trustworthiness, there may then be good reasons to restrict privately offered ct apps, or to institute mandatory quality assurance authorisation in order to ensure that pandemic management is not undermined by business ventures. the importance of trustworthiness of technologies and policies for earning sustainable public trust also means that it is important to prevent false expectations. for instance, simplistic "solutionism", i.e. the belief that pandemic challenges could be managed by technological fixes alone, must be avoided. public decision-making on pandemic policies including decision making on ct apps, requires a structured framework to work through these ethical considerations. such a framework can play a vital role in increasing transparency of made decisions, as well as the trustworthiness of (and trust in) policies and technical solutions. based on our analysis, we conclude the following points for consideration: • the covid- pandemic cannot be solved by technological means alone. digital proximity tracing is not a panacea in the covid- pandemic response, but could become a valuable component in a comprehensive strategy. thus, it is imperative to have appropriate public health measures and infrastructures in place before and while implementing ct apps. • to ensure effectiveness and user-friendliness, there should only be a limited number of ct apps or, ideally, only one platform. reducing the functionality of apps, i.e. only one clear objective per app, seems advisable. while a joint, pan-european platform, allowing interop-erability between different ct apps is warranted, diverging requirements need to be considered. • given the inevitable risks for privacy and the potential impact on individual liberty, especially related to the centralized ct apps, there should be a reasonable expectation of population benefit of ct apps prior to their large-scale applications. effectiveness and benefits must be evaluated alongside the implementation. • the ubiquitous presence of risks necessitates a thorough and prudent approach. a particular focus on temporary measures is warranted. while science and policy have been confronted with deep uncertainty during the covid- pandemic, strategies must be carefully chosen, risks mitigated and measures reversible. uncertainties on the benefits of digital ct limit the set of legitimate pandemic response policies and actions. without sufficiently clear evidence of effectiveness, jeopardizing the rights or liberties of (some parts of) the population cannot be justified. • trust is essential in public health decision-making in general, and covid- ct apps in particular. policies, recommendations and public health measures should be part of a broader endeavour to win and maintain trust in public health measures. well-founded trust requires taking seriously the ethical complexities relating to the implementation of ct apps as well as being transparent about the inevitable trade-offs that are being made. communicating goals and functions as well as possible benefits, risks, and limitations of ct apps clearly and early can play a crucial role in preventing squandering trust and misconceptions. automated decision-making systems and the fight against covid- -our position joint civil society statement: states use of digital surveillance technologies to fight pandemic must respect human rights most americans don't think cellphone tracking will help limit covid- , are divided on whether it's acceptable contact tracing in the real world mobility trends reports contact tracing bluetooth specification preliminary-subject to modification and extension germany's angst is killing its coronavirus tracing app mind the gap: security & privacy risks of contact tracing apps germany flips to apple-google approach on smartphone contact tracing prüfsteine für die beurteilung von, contact tracing"-apps ergebnisse aus dem wiederholten querschnittlichen monitoring von wissen, risikowahrnehmung, schutzverhalten und vertrauen während des aktuellen covid- ausbruchsgeschehens commission recommendation of . . on a common union toolbox for the use of technology and data to combat and exit from the covid- crisis, in particular concerning mobile applications and the use of anonymised mobility data quantifying sars-cov- transmission suggests epidemic control with digital contact tracing mind the app-considerations on the ethical risks of covid- apps digital tools against covid- : framing the ethical challenges and how to address them projects using personal data to combat sars-cov- . updated apple, google covid- tool to be limited to one app per country folkhälsomyndigheten ratar smittspårning via app outpacing the virus: digital response to containing the spread of covid- while mitigating privacy risks report-effective configurations of a digital contact tracing app: a report nhsx no, coronavirus apps don't need % adoption to be effective a flood of coronavirus apps are tracking us. now it's time to keep track of them mobile location data and covid- : q&a on the responsible use of digital data to tackle the covid- pandemic nearly % of icelanders are using a covid appand it hasn't helped much joint statement on contact tracing: date johns hopkins project on ethics and governance of digital contact tracing technologies ( ) coronavirus contact tracing: evaluating the potential of using bluetooth received signal strength for proximity detection covid- and contact tracing apps: technological fix or social experiment? retrieved ethische fragen von digital public health ethics and governance for digital disease surveillance survey of acceptability of app-based contact tracing in the uk ethical guidelines for covid- tracing apps applying a precautionary approach to mobile contact tracing for covid- : the value of reversibility mobile phone data and covid- : missing an opportunity? arxiv preprint ethics of instantaneous contact tracing using mobile phone apps in the control of the covid- pandemic infektionsketten digital unterbrechen mit der corona-warn-app global examples of covid- surveillance technologies: flash report downloading covid- contact tracing apps is a moral obligation how google and apple outflanked governments in the race to build coronavirus apps use of apps in the covid- response and the loss of privacy protection contact tracing as an instrument for pandemic control: central considerations from an ethical perspective pandemic influenza preparedness: an ethical framework to guide decision-making digital technology and covid- tracetogether-an overview improving access to, use of, and outcomes from public health programs: the importance of building and maintaining trust with patients/clients. frontiers in public health ministerie test corona-app regionaal ethical considerations to guide the use of digital proximity tracking technologies for covid- contact tracing. interim guidance covid- digital rights tracker an ethics framework for big data in health and research south korea is reporting intimate details of covid- cases: has it helped? nature, retrieved publisher's note springer nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations acknowledgements this article builds upon a background paper published by the ethics working group within the german public health network covid- (https ://www.publi c-healt h-covid .de). the authors want to thank ansgar gerhardus, dagmar lühmann and dagmar starke for helpful feedback on an early version of this article and two anonymous reviewers for their valuable comments. we gratefully acknowledge samia hurst-majno for her valuable suggestions.funding open access funding enabled and organized by projekt deal. conflict of interest all authors declare that they have no conflict of interest.open access this article is licensed under a creative commons attribution . international license, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the creative commons licence, and indicate if changes were made. the images or other third party material in this article are included in the article's creative commons licence, unless indicated otherwise in a credit line to the material. if material is not included in the article's creative commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. to view a copy of this licence, visit http://creat iveco mmons .org/licen ses/by/ . /.