Statistical methods for real-time forecasts of infectious disease

Abstract

Developing methods to model and forecast the transmission of COVID-19 using tools from epidemiology, statistics, and machine learning.

Topic
Agent-based modeling
Machine learning

Project Details

This project develops methods to model and forecast the transmission of COVID-19 using tools from epidemiology, statistics, and machine learning. Pandemic forecasting is needed to provide actionable information for outbreak response; however, modeling a novel emerging disease using real-time surveillance data from around the globe is an an unprecedented challenge. We develop computationally efficient Bayesian models for forecasting and understanding COVID-19. Our “MechBayes” (Mechanistic Bayesian) forecast model is submitted weekly to the COVID-19 Forecast Hub and the CDC, is featured on the FiveThirtyEight website, and has consistently ranked as one of the top forecasting models in different evaluations

Funding: NIH R35 supplement: $314,023.

More information: preprint, code