ReCOVER: Accurate Predictions and Resource Management for COVID-19 Epidemic Response
Accurate forecasts of COVID-19 is central to resource management and building strategies to deal with the epidemic. We have proposed a heterogeneous infection rate model with human mobility for epidemic modeling, a preliminary version of which we have successfully used during DARPA Grand Challenge 2014. By linearizing the model and using weighted least squares, our model is able to quickly adapt to changing trends and provide extremely accurate predictions of confirmed cases at the level of countries and states of the United States. Training the model to forecast also enables learning characteristics of the epidemic. In particular, we have shown that changes in model parameters over time can help us quantify how well a state or a country has responded to the epidemic. The variations in parameters also allow us to forecast different scenarios such as what would happen if we were to disregard social distancing suggestions.
We have built a web interface that can be used to explore our predictions. We update the UI with new predictions around once in two days. More features have planned for the interface.
Currently we are working on predictions at the level of counties, cities and neighborhoods, while accounting for unreported cases as a latent variable. We have identified that with high probability, for several US states, the number of unreported cases cannot be more than 20-40 times the reported cases. We are further looking into combining mobility data with network algorithms to identify the best strategy towards reopening the economy.
Details of our initial approach can be found in our webinar.
The Github repository for this project is publicly available.
The matlab code for forecasting is also made available on File Exchange.
This work is supported by National Science Foundation Award No. 2027007 (RAPID)
- Ajitesh Srivastava and Viktor K. Prasanna, “Learning to Forecast and Forecasting to Learn from the COVID-19 Pandemic” [arXiv].
- Ajitesh Srivastava and Viktor K. Prasanna, “Data-driven Identification of Number of Unreported Cases for COVID-19: Bounds and Limitations” [arXiv].