Active Infections:
Confirmed Infections:

*We are updating our website, and regular public estimates are temporarily suspended, but our custom modeling service is uninterrupted.

Explore global and national mitigation strength:
*These numbers are for illustration, based on one reasonable choice of R0 and serial interval. In practice some people can infect dozens of others.
Total Infected by 2021
Peak Active Infections

How the Model Works

The impact on is simulated using an epidemiological model based on global, real-world data on human population and travel patterns, displayed as an an epidemic curve.

We want to emphasise that real outcomes depend on human actions. The mitigation strength is our choice. Even given some mitigation level, the outcome is still uncertain, and depends on:

  • the seasonality of the virus and its response to changing weather conditions
  • the mitigation decisions made by other countries
  • technological progress in testing, contact tracing and vaccines

For now we are capturing the international response in a single parameter determining the rate of spread of SARS-CoV-2, but we are building a database of containment and mitigation efforts and working to integrate them into the model.

If you're a decision-maker thinking about how various measures can impact COVID-19 outcomes, please reach out here.

Active Infections Estimate

This counts number of people currently infected with COVID-19 (including those have not yet shown symptoms or are not yet infectious, but excluding those who have recovered or died).

We are trying to augment data on confirmed cases by estimating the true number of infections in selected regions. This gets around issues of testing capacity and under-reporting in official statistics. This is done using a combination of statistical modelling and human forecasting.

Mitigation Measures

We're modelling mitigation measures via their impact on effective transmission rate, since the same mitigation strength can be often achieved in multiple ways. For example, something roughly in the range of "weak mitigation" can be achieved either by closing schools and banning large gatherings (alone) or by compulsory wear of masks and widely adopted voluntary social distancing (alone).

The available mitigation levels are:

  • Do nothing
  • Decrease of effective transmission rate by 20%
  • Decrease of effective transmission rate by 50%
  • Decrease of effective transmission rate by 80%

For comparison, in Wuhan the estimated reproduction number was ~3.5 before mitigation measures, and ~0.32 after implementation of all measures. This would correspond to a mitigation strength of ~90% — stronger than our "Strong" setting.

We are tracking the effectiveness of containments measures in a public database with thousands of entries. As more data becomes available on their effectiveness we will integrate them into the simulation.

Computational modelling

The results are used as an input to the publicly available version of the epidemic modelling software GLEaM (the authors of the software are not part of our project and have not been involved in the study design).

The global development of COVID-19 is simulated using:

  • a global transport network (with multiple settings of travel reduction)
  • multiple models of SARS-CoV-2, capturing uncertainty about seasonality and transmissibility
  • multiple levels of global mitigation measures, represented by their effect on reproduction rate

The results are displayed as a pack of trajectories, representing the uncertainty inherent in our model — and the difference in trajectory caused by the response of us as a society.