Foundation models for time series forecasting and policy evaluation in infectious disease epidemics: a modelling study

Suprabhath Kalahasti, Benjamin Faucher, boxuan wang, Claudio Ascione, Federico Baldo, Eugenio Valdano

Keywords: Epidemiology, Machine Learning, Time Series, Forecasting, Foundation Models, Structured Data

TL;DR: Epidemic Forecasting using time series foundation models

Abstract: Epidemic forecasting and public health policy rely on mathematical models, but traditionally struggle in data-limited settings. We evaluated whether transformer-based foundation models can serve as a new epidemic modeling framework. We tested five models across diseases and locations, including influenza, respiratory syncytial virus (RSV), chickenpox, dengue. Foundation models demonstrated strong accuracy in short-term forecasts and predicted multiple epidemic waves. They outperformed established models on limited and irregular data. We showed foundation models can generate scenarios for policy evaluation, estimating the effect of tighter restrictions on COVID-19 cases during the Alpha variant surge in Italy in 2021. We also used them to estimate the effectiveness of the 2023 RSV immunization campaign in Paris, France. Our findings suggest foundation models can complement existing modeling approaches. Their ability to generate forecasts and counterfactual analyses with minimal data highlights their potential for public health, particularly in emergent and resource-constrained settings.

Submission Number: 43