key: cord-0833134-vnbffy0v
authors: Chen, Xingru; Fu, Feng
title: Highly coordinated nationwide massive travel restrictions are central to effective mitigation and control of COVID-19 outbreaks in China
date: 2022-01-07
journal: ArXiv
DOI: nan
sha: e9233d452a06715b83625050a1a602c72813ccb4
doc_id: 833134
cord_uid: vnbffy0v

The COVID-19, the disease caused by the novel coronavirus 2019 (SARS-CoV-2), has caused graving woes across the globe since first reported in the epicenter Wuhan, Hubei, China, December 2019. The spread of COVID-19 in China has been successfully curtailed by massive travel restrictions that put more than 900 million people housebound for more than two months since the lockdown of Wuhan on 23 January 2020 when other provinces in China followed suit. Here, we assess the impact of China’s massive lockdowns and travel restrictions reflected by the changes in mobility patterns before and during the lockdown period. We quantify the synchrony of mobility patterns across provinces and within provinces. Using these mobility data, we calibrate movement flow between provinces in combination with an epidemiological compartment model to quantify the effectiveness of lockdowns and reductions in disease transmission. Our analysis demonstrates that the onset and phase of local community transmission in other provinces depends on the cumulative population outflow received from the epicenter Hubei. As such, infections can propagate further into other interconnected places both near and far, thereby necessitating synchronous lockdowns. Moreover, our data-driven modeling analysis shows that lockdowns and consequently reduced mobility lag a certain time to elicit an actual impact on slowing down the spreading and ultimately putting the epidemic under check. In spite of the vastly heterogeneous demographics and epidemiological characteristics across China, mobility data shows that massive travel restrictions have been applied consistently via a top-down approach along with high levels of compliance from the bottom up. Our results show that such highly coordinated nationwide lockdowns have been able to effectively suppress the average number of secondary infections and thus central to mitigate and control early-stage outbreaks and avert a massive health crisis otherwise. Our work sheds light on that large-scale coordination of collective actions, albeit costly in the short term such as complying with lockdown orders, is urgently needed to fight this pandemic and get our life back to normal.

The COVID-19, the disease caused by the novel coronavirus 2019 (SARS-CoV-2), has caused graving woes across the globe since first reported in the epicenter Wuhan, Hubei, China, December 2019. The spread of COVID-19 in China has been successfully curtailed by massive travel restrictions that put more than 900 million people housebound for more than two months since the lockdown of Wuhan on 23 January 2020 when other provinces in China followed suit. Here, we assess the impact of China's massive lockdowns and travel restrictions reflected by the changes in mobility patterns before and during the lockdown period. We quantify the synchrony of mobility patterns across provinces and within provinces. Using these mobility data, we calibrate movement flow between provinces in combination with an epidemiological compartment model to quantify the effectiveness of lockdowns and reductions in disease transmission. Our analysis demonstrates that the onset and phase of local community transmission in other provinces depends on the cumulative population outflow received from the epicenter Hubei. As such, infections can propagate further into other interconnected places both near and far, thereby necessitating synchronous lockdowns. Moreover, our data-driven modeling analysis shows that lockdowns and consequently reduced mobility lag a certain time to elicit an actual impact on slowing down the spreading and ultimately putting the epidemic under check. In spite of the vastly heterogeneous demographics and epidemiological characteristics across China, mobility data shows that massive travel restrictions have been applied consistently via a top-down approach along with high levels of compliance from the bottom up. Our results show that such highly coordinated nationwide lockdowns have been able to effectively suppress the average number of secondary infections and thus central to mitigate and control early-stage outbreaks and avert a massive health crisis otherwise. Our work sheds light on that large-scale coordination of collective actions, albeit costly in the short term such as complying with lockdown orders, is urgently needed to fight this pandemic and get our life back to normal.

Global health and humanity has been constantly threatened by emerging novel zoonotic diseases [1, 2] , such as Zika [3] , Ebola [4] , and more recently the COVID-19 [5, 6] . The relentless siege of SARS-CoV-2, the pathogen causing COVID-19 infections [7, 8] , has upended everyone's normal life and caused health crises, lockdowns, and economic percussions at an unprecedented pace and scale. The world has resorted to mandatory non-pharmaceutical interventions (NPI), including lockdowns, face-covering, and social distancing, so as to mitigate disease impact before effective pharmaceutical interventions (e.g., vaccines) become available [9] [10] [11] . Such top-down approaches consider the society as a whole and attempt to optimize intervention measures from the perspective of central planners. On the other hand, adopting personal intervention measures such as complying with lockdown measures incurs a cost to oneself, but collectively protects the community especially these vulnerable. To address these issues, infectious disease dynamics has been an important research area in relevant mathematical and biological fields [12] [13] [14] . Over the years, researchers have proposed behavioral epidemiology as a means of integrating the study of epidemiology with an understanding of health decisions made by individual actors responding to infection risks [15] [16] [17] [18] [19] [20] .

Models of spatial epidemiology have been extensively studied using mathematical approaches combined with real data, with a focus on revealing the spatio-temporal pattern of epidemic spreading [21] [22] [23] . In particular, it is shown that the persistence and resurgence of local community transmissions can be driven by movement between interconnected populations. To understand the persistence and cycles of measles outbreaks, prior work has found that the ubiquitous community structure and their intrinsic heterogeneity can hamper public health efforts to control and prevent childhood diseases [22] . Moreover, previous studies take into account the network topology of communities along with their interconnected mobility in metapopulation models [24, 25] . These prior results provide novel insights into understanding the impact of individual movement on disease dynamics and implications for interventions. In recent years, with the increasing availability and particularly unprecedented dataset about human mobility data [26] [27] [28] , it becomes feasible to study infectious disease and spatial epidemiology with more realistic considerations [29, 30] .

Of particular interest, previous research has demonstrated the effectiveness of travel restrictions to mitigate respiratory virus transmission, yet there are also significant limitations of this approach [31] [32] [33] [34] . As aforementioned, epidemics in interconnected regions, partly due to the het-erogeneity of underlying epidemiological characteristics, can exhibit complicated dynamics during an outbreak [21, 22, 25] . Among others, one of the important driving factors is movements, or more generally mobility patterns, that accounts for local commutes and domestic and international travels [29, 30] . It is known that COVID-19 has a relatively long incubation period [35] and can be contagious before the onset of symptoms (even asymptomatic transmission is more pervasive than previously thought [36] ). Thus, in the very early phase of COVID-19 outbreaks when infrastructures like digital contact-tracing, high throughput testing capacity, and isolation stations are still lacking or insufficient to handle overwhelming exponential outgrowth of the epidemic, the somewhat brute-force lockdowns (through travel restrictions) seem to be the last resort to mitigate the disease impact and save time for development and deployment of other alternative interventions such as vaccines and treatments. However, as time goes by, testing, contact tracing, and isolation may provide a feasible approach for controlling local outbreaks of COVID-19 [10, 11] .

In the wake of COVID-19 outbreaks, a wealth of studies focus on investigating how reductions in international travels through various forms of lockdowns and travel bans would lessen the impact of the pandemic around the world [37] [38] [39] [40] [41] . Retrospectively, these studies invariably confirm the importance of curbing imported cases so as to prevent cross seeding of infections due to the spatial and temporal heterogeneity in epidemics, particularly when local transmissions are outweighed by the risk of imported cases [37] . Noteworthy, [42] analyzes how the cumulative population outflow from the epicenter impacts the timing of onset of local community transmissions in other receiving locations. The authors find a positive correlation between the two quantities, thereby enabling a leading time for predicting the onset of local outbreaks.

As an initial pandemic response when there are not any other pharmaceutical interventions at disposal, drastic lockdown imposes tremendous short-term cost but it can bring long-term positive impact if implemented with ideal timing and coordination. From the top-down management perspective, prior research has extensively investigated optimization of intervention policy using simple epidemic models [43] [44] [45] [46] [47] [48] , including vaccination [49] and isolation [50, 51] . In regard to adaptive social distancing, prior work uses evolutionary game theory [52] or differential games [53] from the perspective of individuals to understand factors of behavioral compliance. While these prior models shed light on adaptive social distancing, it remains urgently needed to examine the effectiveness of lockdowns and assess their actual impact on disease mitigation and control using a data-driven approach. For this purpose, changes in mobility patterns can be a good proxy to study the impact of non-pharmaceutical interventions (NPI) such as lockdowns and social distancing on behavior changes that help reduce community transmissions in an interconnected setting of spatial epidemiology.

In the meantime, it is not uncommon that some people protest against measures of lockdowns and NPIs [54] , together with debates about their cost-effectiveness and impacts on health and society [55, 56] (see a recent comprehensive review in [57] ). To examine and validate conditions for lockdowns to be effective, we take a second look at China's containment efforts. Despite its vast size and huge populations, China's lockdowns have been able to mitigate and control local outbreaks through massive travel restrictions. In early 2020 when knowledge about the COVID-19

is still rather limited, the decision to lockdown came after many deliberations and amid tremendous uncertainty. Even so, to put a population of more than 1.4 billion on lockdowns and in some provinces strict home quarantines for an exceedingly long period is a hard, top-down decision that comes at an astronomical economic cost. Yet it turns out to have long-term positive impacts mainly because the stringent lockdowns have been implemented effectively.

Here, we use a data-driven modeling framework to assess the effect of lockdowns on transmission reductions and improve our understanding of the necessity of uniform and highly synchronous lockdowns in light of the spatio-temporal pattern of COVID-19 outbreaks. To this end, we explore China's COVID-19 lockdowns as a concrete example, examine the level of synchrony of implementing travel restrictions across China, and quantify the impact of lockdowns on people movements.

Our data analysis based on the massive mobility data reveals that lockdowns are implemented highly synchronously and uniformly at multi-levels, that is, between provinces (30 administrative regions) and within provinces (at the level of prefecture within a given province such as Hubei). As demonstrated in theoretical research about spatial epidemiology [21, 22, 25] , highly coordinated nationwide massive travel restrictions are central to effective mitigation and control of COVID-19

in China, especially during the early stage of epidemic outbreaks. Such massive travel restrictions ultimately lead to successful control of COVID-19, saving the country from the blink of a huge health crisis. Albeit the immeasurable loss of lives and economy, it is a remarkable achievement, which is made possible by the sacrifice of each and every one of ordinary people just like everyone elsewhere who has been impacted by this pandemic.

In late December 2019, the outbreak of COVID-19 is first reported in Wuhan, Hubei, which is a central transportation hub (especially for trains). The situation is rapidly escalated to a public health emergency after local case surges and excessive hospitalizations that cause hospital overflow Due to the strict implementation of national lockdowns, the travel volume eventually approaches desired control target (Fig. 1 ). In particular, the epicenter Hubei experiences the strictest ever travel restrictions and there are barely any free people movements except for essential travels for an exceedingly long period (Fig. 1) . As a matter of fact, Wuhan residents are strictly confined in their homes for months (in total 76 days) until early April 2020. In late February 2020, a month after the lockdown, many provinces lower their response levels and lift their lockdowns.

As a consequence, travels rebound but still, the impact of lockdowns on mobility is long-lasting.

Despite lower numbers of active cases, the reopening up efforts by the government see little effect as people need time to feel comfortable about traveling again due to the fear of lockdowns and the potential risk of infections. Fig. 2 shows the pairwise mobility index between provinces; the lump sum of each column and each row, respectively, gives the total volume of 'move in' and 'move out' in Fig. 1 . The order of listed provinces is ranked according to their final epidemic size by the end of our study. The level of inter-province mobility appears to be correlated with their epidemic sizes, forming a cluster on the upper right corner (Fig. 2) .

The heatmap plots provide us with an intuitive visual guide for understanding the degree of interconnectedness between provinces in terms of their bidirectional travel volume (Fig. 2) . After two weeks since the lockdown (Feb 4, 2020), the travel reduces to a bare minimum that is required to maintain essential living and work. The entire nation is paused at a massive scale and at a highly coordinated pace (Fig. 2) . In most places, the intra-province mobility is reduced dramatically more than 90%, and even more, the inter-province travels are cut almost at 100%, in particular for the travel from and to the epicenter and other most affected provinces (see the upper rows, those provinces that had the largest outbreaks, of the heatmaps in the middle row of Fig. 2 ). Taken together, these results demonstrate that China's lockdowns are highly synchronous and effectively stop long-range spatial spreading due to domestic travels.

As an outbreak unfolds, its emerging spatio-temporal pattern is highly dependent on the underlying multi-scale and multi-layer population structure, among others, most crucially on the mobility pattern [22, 24, 30, 42] . People make local and non-local movements during which inevitable close contact/proximity with others can seed infections near and far. Fig. 3 presents an overview of the emergent spatio-temporal pattern of the COVID-19 outbreak in China. In accordance with Fig. 2 , most affected provinces suffering the largest outbreaks by 10 March 2020 are those with the greatest levels of interconnectedness with, and thus receiving the largest population outflow from, the epicenter Hubei, including Guangdong, Henan, Zhejiang, Hunan, and Anhui (highlighted in Fig. 3a) . Fig. 3b further demonstrates how the phase and magnitude of outbreaks in each province correlate with the cumulative population outflow received from Hubei.

The greater outflow received from the epicenter, the earlier onset of the local outbreaks will be detected along with a larger number of cases (Fig. 3b) . These data-based results provide a direct and intuitive rationale for synchronous lockdowns that are required to ultimately control and possibly eliminate infections, at least in the early phase of an epidemic when limited options other than costly non-pharmaceutical interventions are possible. Altogether, our analysis based on China's COVID-19 dataset suggests such population outflow from epicenters determines the timing and scale of the outbreaks (Fig. 3) .

To further quantify the synchrony of lockdowns implemented across China, we perform com-prehensive comparative statistics of inter-province and intra-province time series of mobility data (Fig. 4 ). The correlations on the level of provinces range from 0.77 to 0.98 (with value one suggesting perfect synchrony) (Fig. 4a) . Further zooming in, we take a close look at the prefecture level within the epicenter Hubei province: the synchrony of mobility changes due to well-coordinated lockdowns between the capital Wuhan and other cities within Hubei is significant and with little variations (correlations ranging from 0.85 to 0.95) (Fig. 4b) .

We further use an SEIR compartment model with a data-driven approach to infer the timedependent transmission rate β i (t) (see Methods & Model, and also the SI). This parameter reflects how well mobility patterns (reductions and changes) translate into effective transmission rates (infections via contacts) (Fig. 4c) . We can see that the impact of lockdowns on reducing transmissions does not occur immediately, but lags a certain time (on average two weeks or so) to reach desired effective behavioral changes that eventually lead to reductions in transmissions. This result implies that lockdown measures need to last sufficiently long enough so as to see their positive mitigation impact, partly because people need some time to fully adjust to, and more importantly comply with, quarantine orders, especially strict home quarantine.

We also estimate the effective basic reproductive ratio, R t , in order to characterize the impact of interventions on controlling the epidemic over time. Fig. 5 shows the best estimated R t for each province, most of their values varying from 2 to 10. Owing to the unique demographics of each province, the heterogeneity of R t requires a distinct level of interventions. For example, the epicenter Hubei has an R t ∼ 4 prior to lockdowns, whereas its strongly interconnected province Guangdong has an R t ∼ 15. Despite such drastically different epidemiological characteristics and population densities and sizes, the universal lockdowns implemented almost synchronously have managed to contain the epidemic outbreaks in each province, which would have become too overwhelming to handle otherwise. Fig. 5 also reveals the intrinsic difference in the persistence of COVID-19 and the effectiveness of interventions across the nation. In Zhejiang and Shanghai (which are economically developed regions), the interventions are highly efficient and bring down the R t below one within days. In contrast, as for the epicenter Hubei, it takes a month to curb the level of infections below the critical threshold. Although lockdown is not a one-size-fits-all approach, there should be no question about its effectiveness, as long as implemented in synchrony across the target population, in control and mitigation of an emerging epidemic. In this work, we use a data-driven approach to estimate the effectiveness of such massive travel restrictions in the mitigation of disease impact. Our work shows that highly coordinated massive province's distinct pace and magnitude of epidemic impact mitigation. The plot shows that province-specific R t is heterogeneous and has a distinctive pattern with respect to the implementation of local lockdown measures across provinces, but R t is uniformly suppressed after two weeks of nationwide lockdown and invariably drops well below one after one month. tragedy of the commons can arise as a result of 'free-riding' in this important context [61] . Individuals may not follow disease intervention measures suggested by public health officials, especially if the epidemic curve is being bent down, but the uptick of cases, in turn, causes individuals to become more vigilant and increase their compliance levels. The feedback loop of this sort gives rise to oscillatory dynamics of disease prevalence and behavioral compliance to top-down public health measures, as seen in the current pandemic with multiple waves of infection [52] . In addition to the social dilemma aspect of disease control, tremendous uncertainty associated with early detection of local community transmission and overall pandemic forecasting makes top-down scenario planning and optimization of intervention and mitigation extremely challenging [62] . As such, top-down and bottom-up modeling approaches need to go hand in hand in order to better inform public health efforts for effective disease interventions.

One potential incomplete data issue of the present work is the aggregate Baidu mobility data based on phones we used. Admittedly, such mobility data can be underestimated due to population heterogeneity in phone usage and user privacy settings. However, as compared to a study published decade ago [63] , China is now one of the leading countries in smartphone usage and ownership for both rural and urban populations: a recent Pew survey reported extraordinarily high smartphone ownership [64] . Keep in mind our analysis is based on the levels of inter-province and intraprovince mobility which already has a large population size. While accurate calibration for such potential sampling biases in mobility data is out of the scope of the present work, our results on understanding the spatio-temporal pattern of COVID-19 spreading and quantifying the impact of lockdowns are still of relevance and interest even as a case study of massive travel restrictions.

It remains an open problem to promote bottom-up behavior and attitude changes for the greater good [65] . Compliance with public health recommendations and orders is an outstanding issue plaguing many parts of the world and greatly compromising efforts to mitigate the pandemic.

Lockdowns had been attempted across the world, yet with drastically different outcomes. Many regions tried varying degrees of enforcement but met with resistance due to privacy and civil rights concerns. In contrast, China has a unique regime and governance structure to enforce a national lockdown through a well-considered top-down approach.

Our work provides data-driven evidence for supporting highly coordinated lockdowns that should be implemented in the early onset of pandemics in order to be effective. Only when applied in concert with all regions are strict lockdowns effective. Across the world, other countries like Singapore and New Zealand have also seen successes in containing COVID-19 using well-coordinated national lockdowns. Undoubtedly, such lockdowns come at a huge cost -business shutdowns, worker layoffs, and lack of child care, just to name a few -and inflict economic repercussions. On the other hand, it is necessary to do so in the early phase of the global pandemic when effective pharmaceutical interventions (like vaccines and anti-viral treatments) are lacking or still under rapid development.

With the increasing options of interventions and especially ramping up vaccination, it becomes hopeful to get our life back to normal. Massive travel restrictions are no longer needed to contain case surges [66] , provided that responsive and targeted local lockdowns by means of high-precision contact tracing as well as testing and isolation are in place. Since the full-scale national lockdowns in January 2020, local outbreaks due to imported cases from time to time in China have been successfully controlled and eliminated using such prompt, targeted testing and isolation so as to avert serious and costly national lockdowns repeatedly.

In conclusion, the pandemic has fundamentally shaped the whole world and reminds us of the importance of pandemic preparedness and global health management. The worst of all are discrimination and hate crimes around the globe [67] . Large-scale cooperation is urgently needed to solve many challenging issues facing our common humanity. Fighting this pandemic is yet another wake-up call for that.

a. Model description. Our modeling framework builds on multi-scale behavioral epidemiological spreading processes that incorporate mobility patterns [29, 30, 37] of inter-province migrations (which affect the spatial spreading among provinces). Specifically, we consider a Susceptible-Exposed-Infected-Recovered (SEIR) model in a metapopulation structure with migration. The systems of ODEs describe the dynamics in continuous time t, that is, days since the disease outbreak:

Here, the subscript i refers to the ith compartment on the provincial level. N i (t) = S i (t) + E i (t) + I i (t) + R i (t) is the total population size of compartment i at time t. j, j i α i j (t) [S i (t) + E i (t)] is the total outflow from compartment i to other compartments, and j, j i α ji (t) [S i (t) + E i (t)] is the total inflow to compartment i from other compartments. To parameterize migration flows between compartments, we use the real provincial level mobility data from the Baidu Map service, which provides aggregate level tracking of domestic travels on a daily basis. The impact of lockdowns and travel restrictions on aggregate behavioral responses/changes that lead to transmission reductions is characterized by the province-specific transmission rate β i (t). More detailed modeling and data analysis can be found in the SI.

• unit of time t: day.

• β i (t): transmissibility rate, which can be time dependent, due to lockdown efforts (home quarantine and travel restrictions) [68] .

• 1/σ i (t): incubation period, which ranges from 1 ∼ 14 days [35, 36] .

• 1/γ i (t): number of days remain infectious, which ranges from 1 ∼ 14 days [69] .

• R 0 = β i (0)/γ i (t): the initial values of R 0 are bounded within 1.4 ∼ 4 [70, 71] .

Uncertainty quantification and sensitive analysis. We apply the dual annealing algorithm to perform a nonlinear least square fitting procedure for estimating time-varying epidemiological parameters, denoted by a vectorθ, combined with real mobility data. We further calculate the covariance matrix cov(θ) = s 2 (F F) −1 , with F = ∂ f (θ) ∂θ θ=θ , and hence standard errors of the estimated parameters (the diagonal elements ofθ). The covariance matrix contains complete information about the uncertainty of parameter estimations. Following a Student's t-distribution, the confidence interval at (1 − 2α) significance is given byθ 1−2α =θ ± t α n−p diag cov(θ) (see, e.g., the uncertainty quantification for our predicted curve in Fig. S11 in the SI, where the uncertainty of prediction, denoted by the shaded area, propagates as a function of the mean behavior of the spreading dynamics).

b. Datasets. Mobility data is obtained from Baidu https://qianxi.baidu.com.

COVID-19 data is collected and curated by DXY and archived at https://github.com/ BlankerL/DXY-COVID-19-Data.

c. Open code. Source code is available at the GitHub repository https://github.com/ fudab/China-COVID-19-mobility. An interactive website can be found at https://fudab. github.io/covid-19/china.

Predictions of virus spillover across species

Mathematically modeling spillovers of an emerging infectious zoonosis with an intermediate host

Zika virus

Fruit bats as reservoirs of Ebola virus

A novel coronavirus from patients with pneumonia in China

Estimating clinical severity of COVID-19 from the transmission dynamics in Wuhan, China

The proximal origin of SARS-CoV-2

A novel coronavirus emerging in China? key questions for impact assessment

Compassionate Use of Remdesivir for Patients with Severe Covid-19

Feasibility of controlling COVID-19 outbreaks by isolation of cases and contacts

2020 Quantifying SARS-CoV-2 transmission suggests epidemic control with digital contact tracing

Infectious diseases of humans: dynamics and control

The mathematics of infectious diseases

Population biology, evolution, and infectious disease: convergence and synthesis

The effect of opinion clustering on disease outbreaks

Social factors in epidemiology

Addressing vaccine hesitancy

The vaccine wars

Dueling biological and social contagions

Imperfect vaccine and hysteresis

Measles periodicity and community size

Space, persistence and dynamics of measles epidemics

Spatial epidemiology: current approaches and future challenges. Environmental health perspectives

Reaction-diffusion processes and metapopulation models in heterogeneous networks

Epidemic modeling in metapopulation systems with heterogeneous coupling pattern: Theory and simulations

Modelling the scaling properties of human mobility

A universal model for mobility and migration patterns

Exploring the mobility of mobile phone users

Multiscale mobility networks and the spatial spreading of infectious diseases

Quantifying the impact of human mobility on malaria

Effectiveness of travel restrictions in the rapid containment of human influenza: a systematic review

The effect of travel restrictions on the spread of a moderately contagious disease

Modeling the impact of air, sea, and land travel restrictions supplemented by other interventions on the emergence of a new influenza pandemic virus

A conceptual model for the coronavirus disease 2019 (COVID-19) outbreak in Wuhan, China with individual reaction and governmental action

Early transmission dynamics in Wuhan, China, of novel coronavirus-infected pneumonia

Epidemiology and Transmission of COVID-19 in Shenzhen China: Analysis of 391 cases and 1,286 of their close contacts

The effect of travel restrictions on the spread of the 2019 novel coronavirus (COVID-19) outbreak

2020 A data-driven assessment of early travel restrictions related to the spreading of the novel COVID-19 within mainland China

Association of public health interventions with the epidemiology of the COVID-19 outbreak in Wuhan

Changes in contact patterns shape the dynamics of the COVID-19 outbreak in China

Effect of non-pharmaceutical interventions to contain COVID-19 in China

2020 Population flow drives spatio-temporal distribution of COVID-19 in China

Optimal control of some simple deterministic epidemic models

Optimizing infectious disease interventions during an emerging epidemic

Controlling epidemic spread by social distancing: Do it well or not at all

Time-optimal control strategies in SIR epidemic models

Optimal timing of interventions during an epidemic. Available at SSRN 3607048

2021 Optimal, near-optimal, and robust epidemic control

Optimal immunisation policies for epidemics

An optimal isolation policy for an epidemic

Optimal isolation policies for deterministic and stochastic epidemics

Oscillatory dynamics in the dilemma of social distancing

Game theory of social distancing in response to an epidemic

Protest Against COVID-19 Containment Policies in European Countries. Available at SSRN 3844989

2020 Do lockdowns work? A counterfactual for Sweden

2021 COVID-19, lockdowns and well-being: Evidence from Google Trends

Non-pharmaceutical interventions during the COVID-19 pandemic: A review

Optimal control of an epidemic through social distancing

Behaviour fuels, and fights, pandemics. Nature human behaviour 4

The tragedy of the commons: the population problem has no technical solution; it requires a fundamental extension in morality

Defining the epidemiology of Covid-19? studies needed

Digital divide between urban and rural regions in China

China outpaces India in internet access, smartphone ownership. Pew Research Center

Using social and behavioural science to support COVID-19 pandemic response

Local lockdowns outperform global lockdown on the far side of the COVID-19 epidemic curve

2020 Xenophobia in the time of pandemic: othering, anti-Asian attitudes, and COVID-19

Estimates of the severity of COVID-19 disease

Temporal dynamics in viral shedding and transmissibility of COVID-19

Temporal variation in transmission during the COVID-19 outbreak

2020 The novel coronavirus, 2019-nCoV, is highly contagious and more infectious than initially estimated

We thank Timmy Ma, Xin Wang, and Daniel Escudero for helpful discussions when initializing