Some Select COVID-19 Modeling Resources

by Joseph Rickert

There is an incredible amount of COVID-19 related material available online. While many dashboards, data sets, shiny apps and models represent significant contributions towards fighting the pandemic, we seem to have reached a point where we should be thinking about standards of quality, and should be exploring avenues for cooperation before launching more individual efforts. Below are a few examples of what I believe are exemplars of different types of COVID-19 related contributions. Please feel free to criticize my choices, or suggest other examples of quality contributions in the comments following this post.

A Shiny App

At the top of my list is the Shiny based model: Modeling COVID-19 Spread vs Healthcare Capacity developed by Alison Hill of Harvard’s Program for Evolutionary Dynamics with contributions from several researchers at the University of Pennsylvania, Harvard and Iowa State. Some things I really like about this app are: (1) that it provides enough background information to be self-documenting and self-contained. (2) The interactive visualizations are without unnecessary adornment and clearly labeled. (3) Default settings for sliders and other parameters are mostly based on documented data. (4) The differential equation theory underlying the model is clearly presented with the intention of helping model users understand the math. The content behind Model and Tutorial tabs sets a standard of excellence for concise explanations within the limits of the form factor. (5) The extensive documentation under the Sources tab inspires confidence and opens wide a window to this field of health care planning. (6) The code is straightforward and well organized. I believe that anyone developing a Shiny app for use by professionals would do well to try and meet or exceed the standard set by this app.

A Curious Number

To me, no single number in the catalog of COVID-19 models is more alluring, or cloaked in deeper mystery than R0, the reproduction number. So I was delighted to discover James Holland Jones’ 2007 Notes on R0, a very readable account of the assumptions and mathematics that underlie this parameter that provides links for those who want an even deeper understanding.

Some Prominent R Packages

The R packages from the R Epidemics Consortium are tools for professional epidemiologists. See the Tim Churches 3/19 post for an example of using EpiModel and his 3/5 post for and example using earlyR and EpiEstim.

A Site for Researchers

The comprehensive Coronavirus Disease (COVID-19) Statistics and Research site by Max Roser, Hannah Ritchie and Esteban Ortiz-Ospina offers a global view of the pandemic that is updated daily.

A Few Dashboards

Two High Impact Blog Posts

  • Tomas Pueyo’s 3/10 post which raised the alarm in American Statistical Association circles is still worth reading. The alarming predictions of human suffering backed up by data visualizations make an emotional impact. This post also displays the the masterful graph from the by Wu and McGoogan’s JAMA paper showing the early history of the outbreak in Wuhan.
  • Terry Tao’s March 25th post which announces the Christopher Strohmeier COVID-19 Polymath Proposal initiated a valuable discussion on relevant data sets. See the Covid-19 dataset clearinghouse wiki.

Some Data

In addition to the data behind some of the above links, you may find the following data sources useful. The Johns Hopkins COVID-19 data seems to have become the defacto standard for COVID-19 modeling. You can directly download: time_series_covid19_confirmed_global.csv and time_series_covid19_deaths_global.csv. Additionally, both esri and data.world (See The Corononavirus (COVID-19) Data Resource Hub) provide a number of curated COVID-19 data sets.

Share Comments · · · · ·

You may leave a comment below or discuss the post in the forum community.rstudio.com.