EpiCovDA: Forecasting COVID-19 using a mechanistic model and data assimilation
When
In this talk, we present a framework for forecasting the on-going COVID-19 pandemic in the United States. We use a mechanistic model that formulates a compartmental model in terms of incidence vs. cumulative-cases (ICC) curves. We adapt a variational data assimilation approach in order to perform parameter estimation for the mechanistic model using recent case incidence data. While not modeled as an explicit component in the mechanistic modeling framework, our approach allows for incorporation of changes in disease dynamics due to, for example, stay-at-home orders. Our results include real-time forecasts for case and death counts for the US and each of its 50 states.