key: cord-0621723-281zogrr authors: Talagala, Thiyanga S.; Shashikala, Randi title: Interactive Dashboard to Monitor the COVID-19 Outbreak and Vaccine Administration date: 2022-05-15 journal: nan DOI: nan sha: 6dc05826f585d889abafc1a783b0856f0f59f55b doc_id: 621723 cord_uid: 281zogrr Dashboards are the most common visualization method for displaying COVID-19 data and informing the public. We examined 15 different dashboards to see how various visualization techniques were used. This paper describes the creation and implementation of a dashboard for COVID-19 epidemic and vaccination administration data in Sri Lanka. COVID-19 has expanded over the globe, having a significant impact on our daily lives and work. Early responses and timely decisions and actions are critical to saving communities and economies worldwide. Data is essential in order to make effective decisions. Data-driven information guides the decision-making process and also evaluates the effectiveness of strategies taken. Massive amounts of data are being generated in the response to the COVID-19 pandemic. Given this available data, it is critical to create tools for exploratory analysis for policy-makers, health officials, and the general public. Dashboards are one of the greatest visual interpretation methods for tracking the COVID-19 pandemics spread and vaccine administration. Dashboards allow users to quickly interact with a combination of exploratory visualizations and gain a quick overview of the data. This paper describes the development and implementation of a dashboard for the COVID-19 outbreak and vaccine administration data in Sri Lanka. There are a plethora of COVID-19 visualization dashboards that have been designed to visualize the pandemics global and local status. Different software can be used to generate dashboards. We explored 15 dashboards designed to visualize COVID-19 data at the global and country levels. First, dashboards were compared to identify the various features, visualization approaches, and enhancements that should be implemented. Next, we developed an interactive dashboard to visualize the COVID-19 outbreak and vaccination information in Sri Lanka. This dashboard provides front-line health officers a situational awareness of the spread of COVID-19 and the status of the vaccination program. The rest of the paper is organized as follows: Section 2 of dashboards created using data related to the COVID-19 pandemic. Section 3 presents the methodology and basic design concept; Section 4 presents the results; and Section 5 concludes. Dashboards are one of the best visual interpretation methods for tracking the spread and communication of the COVID-19 pandemic. The 15 dashboards we used in the literature survey are listed in Table 1 . We compared dashboards to identify data types, plotting techniques, colour themes, and other features such as interactivity on plots and panel numbers. Table 02 , all dashboards which are considered in this paper represent the data related to COVID-19 confirmed cases, recovered cases, and deaths. There were 8 dashboards out of 15 dashboards that contained vaccination details. Table 03 highlights the dashboard visualization techniques. Value boxes have been utilized to display total figures on practically every dashboard. The most common ways of visualizing confirmed cases, recovered cases, deaths, and immunization details are bar charts and line charts (trend lines). The majority of dashboards displayed data on a daily or weekly basis. The spatial distribution of COVID-19 cases by country, province, regional, and other factors is tracked using choropleth maps. When visualizing the data by the map colour code system, circles with respect to the size of the cases have been used to visualize the variation in size. Several dashboards use doughnut-shaped pie charts to indicate total COVID-19 confirmed cases, recovered cases, active cases, and deaths as a proportion. Furthermore, region, gender, age group, and ethnicity can be identified as common breakdowns of COVID-19 cases. Data tables for representing cases' distribution by province/region have been added to some dashboards. Very few dashboards have been visualized in the COVID-19 test details. Only 6 dashboards have been compared to global situations. In addition, the fatality rate, incidence rate, ICU beds, stage of the patients, and hospitalized details have been contained in the several dashboards. Before developing a dashboard, it is necessary to think about which visualization tools and features should be contained in the dashboard. What are the most suitable plots, how many panels in the dashboard, what data should be included, how to fit the dashboard on a screen, colours, and is it real time updated or not are the common things that should be considered before developing the dashboards. Table 4 summarizes information under the following categories: i. Number of panels -How many panels are included in the dashboard. ii. Visualization tools -what are the graphical representations of data. iii. Fitted on a single screen -whether the dashboard fits on a single screen or not (users can see the whole dashboard on a single screen without adjusting through grid overlay or not). iv. Colour theme -is there a unique colour used for one data type in the whole dashboard (i.e.: one colour scale for one data type everywhere on the dashboard). v. Dark background -the background colour of the dashboard is dark or light. vi. Data available -whether users can downloaded or whether data is available to reproduce the results. vii. Real time updated -whether the dashboard is updated daily/ specific time (live dashboard) or not. We obtained data from COVID-19 situation reports published by the Epidemiology Unit, Ministry of Health Sri Lanka. The data includes the number of death cases, number of hospitalized cases, number of recovered cases, and COVID-19 vaccinated counts in Sri Lanka. The data is made available through an open-source R package covid19srilanka (Talagala 2021). R software was used for data cleaning and analysis. The flexdashboard (Iannone, Allaire, and Borges 2020) package was used to build the data visualization dashboard. The initial layout for the dashboard was prepared based on Krispin (2021) . Data visualizations are generated using the ggplot2 (ggplot2?) and plotly (Sievert 2020 ) packages in R. We used colour-blind friendly colour palettes for the graphics. A diverging colour palette was used to represent qualitative data, and to represent numeric variables, a sequential colour theme was used. Table 5 provides an overview of methods that have been used to visualize data. We now describe the novel visualization approaches we included in our dashboard. To effectively distribute the vaccine and to support situational awareness and inform policy-makers' decision-making, it is important to know the district-wise spread of COVID-19 cases. We have daily COVID-19 data related to confirmed cases in all 25 districts in Sri Lanka. This structure generates a multiple time series collection. Visualizing this time series data is useful to identify similarities and dissimilarities between districts and their general trends. There are two approaches to visualizing these time series: (i) creating individual time series plots for each district (as shown in Figure 1 -A), and (ii) plotting all time series on a single panel at the same time (as shown in Figure 1 -B). Plotting all time series simultaneously is also not possible due to overlapping time series and scale differences. Plotting separate panels for each district is not effective. The reason is that it is hard to compare across 25 different panels at once. In order to overcome these problems in multiple time series visualization, we use heat maps (Peng (2008) ) to visualize global and local similarities and dissimilarities across districts. The associated results are shown in Figure 2 . Here, two heat maps are used to show the global variations (Figure 2 : A) and local variations (Figure 2 : B) in the time series collection. Figure 2 -A cell colours represent the actual counts of the COVID-19 confirmed cases. This is useful to get an idea of the differences in absolute values. Figure 2 -B cell colours represent the normalized values created by applying the min-max transformation. A min-max transformation is applied to each district's time series by using the corresponding district's minimum and maximum value. This helps us to get an idea about patterns within districts. For example, according to Figure 2A , we can see that in Colombo, Gampaha, and Kalutara districts, COVID-19 cases are significantly higher than in other districts. According to Figure 2B , all districts show an increasing trend pattern as the right-hand side of the cells are lighter than the left-hand side cells in the heat map. Furthermore, according to Figure 2B , all districts reported a high number of cases on August 19, 24, and 29, 2021. Figure 3B is useful for identifying these local outlying behaviours. As shown in Figure 4 , we also use a Choropleth map and Dorling cartogram to visualize the spatial distribution of COVID-19 cases. The vaccination information is visualized through interactive time series plots and bar charts. The "Sri Lanka COVID-19 Dashboard" provides an overview of the COVID-19 pandemic and administration of vaccine information in Sri Lanka. This dashboard has eight panels as listed in Table 6 . Bar charts and line charts are the most frequently used tools for the visualization of total cases, daily cases, and weekly cases, and comparisons with respect to time. Some dashboards contained doughnut-shaped pie charts to summarize the total figures. In almost every dashboard, value boxes have been used to represent total figures. Some dashboards contained interactive maps and data tables to visualize the distribution of cases by country, province, region, or state. All dashboards are updated daily in real time. Gender, age groups, and ethnicity can be identified as common breakdowns. Most data sets and related links are available. We used a colour-blind-friendly theme when creating our dashboard. The dashboard includes both static and interactive charts that can be used for explanatory (telling you something), exploratory (give user the chance to interact with the plots), and exhibitory (showing data visually) purposes. The data speaks to us, but it is not always easy to understand in what language they are doing it. Hence, we examine the same data under different angles using different types of plots. We do not have district-wise vaccination details. This is a potential future direction to think about. Our dashboard is completely reproducible. The source code for reproducing the results is accessible in a public GitHub repository at https://github.com/thiyangt/covid19srilanka. COVID 19 Dashboard -Belgium COVID 19 Dashboard India -ZOHO Analytics -ZOHO COVID 19 Dashboard -NHS providers COVID-19 integrated surveillance data in Italy -EpiCentro COVID-19 Jamaica -Ministry of Health and Wellness COVID-19: Live situational Analysis Dashboard of Sri Lanka COVID-19 Map -Johns Hopkins Coronavirus Resource Center COVID-19 Surveillance Dashboard -NSSAC Research COVID-19 ZA Dashboard -Data Studio Flexdashboard: R Markdown Format for Flexible Dashboards The Coronavirus Dashboard Pakistan's Official COVID-19 Dashboard -Shifa International Hospitals Ltd A Method for Visualizing Multivariate Time Series Data Interactive Web-Based Data Visualization with r, Plotly, and Shiny Covid19srilanka: The 2019 Novel Coronavirus COVID-19 (2019-nCoV) Data in Sri Lanka The Global COVId-19 Index (GCI) -Russia Dashboard -PEMANDU Associates Track COVID-19 Across Canada Using Our Interactive Dashboards WHO Coronavirus (COVID-19) Dashboard