Date/Time Date(s) - 04/03/20253:30 pm - 4:30 pm
Speaker: Dr. Justin Slater (University of Guelph) Research area: Bayesian methods in statistical epidemiology
Location: BSB 121
Title: A statistical framework for reconstructing epidemic curves
Abstract: Estimating the number of individuals who have had an infectious disease is essential for understanding disease burden, yet this remains challenging as all sources of surveillance data come with their own biases. A comprehensive approach must integrate reported cases, wastewater surveillance, and serosurvey data while addressing biases in each source. In this talk, I present a flexible Bayesian framework that (i) models under-reporting using approximations of count-valued state-space models, (ii) accounts for noisy wastewater signals with differentiable Gaussian processes, and (iii) leverages serosurvey data both for informative priors and model validation. I demonstrate this approach by reconstructing epidemic curves in Toronto and New Zealand, highlighting insights gained and challenges encountered.