World Health Organization Team Develops COVID-19 Transmission Estimation Tool
The WHO Team developed an interactive visualization tool that allows users to upload a .csv file and produce COVID-19 transmission rate estimates and epidemic curves.
Introduction
The World Health Organization Team developed and deployed a visualization tool that estimates COVID-19 transmission rates and epidemic curves using the number of cases for each country. This interface helps policymakers monitor transmissions and understand the efficacy of public health measures like social distancing interventions.
The Case Team Lead, Justin Zhu ‘21, worked with Amy Tan ‘20, Sophie Khorasani ‘21, RunLin Wang ‘23, and Theresa Nguyen ‘20 to produce a final deliverable. Carina Peng ‘23 served as the Engagement Coordinator.
Methodology & Findings
When a user uploads a csv file with the latest COVID-19 data, the webpage generates dynamic epidemic curves and estimates the reproductive number known as R with a 95% confidence interval. With this tool, policymakers can utilize comprehensible visualizations and forecasts to inform their decisions about social distancing and reopening measures. In particular, for policymakers who may not have sufficient funding or expertise to build these models themselves, these datasets allow easy access to evidence-based models. Since the interface allows users to upload any data, it is flexible and useful for individuals from different countries, levels of government, and technical backgrounds.
This ShinyApp was developed using EpiEstim, parametric and nonparametric models, and ggplot visualizations (their work can be found on Github too). Throughout the process, the team of data scientists learned how to version control and create branches in Git. They gained familiarity with data.frame, as well as ui and server components of shiny. The Team Lead organized weekly lessons on statistical methods and technologies that were later integrated in the interface. For instance, the visualization of the reproductive number involves 95% confidence intervals that were calculated using sliding weekly windows, with a parametric serial interval based on a mean of μsi = 4.8 and standard deviation σsi = 2.3.
Directions for Future Research
The team is also considering actionable next steps to further develop their interface. First, it would be helpful to analyze the subset of the population who is infected. Policymakers may utilize data such as the proportion of infected individuals who become ill, or how many of those individuals self-isolated, to inform policy decisions. To understand the effectiveness of isolation, government agencies would also benefit from examining differences — percentage reduction, in particular — in rates of transmission between consecutive days, as well as between the day of isolation and the day of incidence. The team also aims to separate the effects of isolating sick patients from the effects of social distancing on the day of isolation.
Case Team Lead Justin Zhu reflected on the process: “I enjoyed this experience volunteering with the World Health Organization and Harvard Data Analytics Group. I’m happy to see many of my teammates learn so many things in such a short amount of time and to see through the successful completion of this project as team lead. Together, we helped make a difference during a critical time for many countries.”
This article reports the work of Harvard College Data Analytics Group’s COVID-19 Crisis Response Team. Edited by Kelsey Wu.