IWR Colloquium Funk 2026 Mechanism, Learning, and the Limits of Outbreak Forecasting

Every new infectious disease outbreak raises the same urgent, quantitative questions: how large is it really, where is it heading, how many hospital beds will be needed. These are prediction problems, and hard ones: most cases are typically not (yet) reported, the data that exist are sparse and noisy, and the processes that generate them are only partly understood. Addressing them means combining what we know about the mechanisms with what we can learn from the data, and every method draws that line in a different place.

Drawing on experiences from COVID-19 and other outbreaks, I will ask how well mathematical models can predict the course of an epidemic: how we draw on several models at once, how we know whether they are working, how far ahead forecasts stay useful for the decisions they inform, and what ultimately limits how good they can be. Some of those limits are ours to push back; others are intrinsic to the systems we try to predict, and a lot of the interesting work is in telling the two apart.