|
Abstract:
|
Stochastic epidemic models such as the Susceptible-Infectious-Removed (SIR) model are widely used to model the spread of disease at the population level, but fitting these models present significant challenges when missing data or latent variables are present. In particular, the likelihood function of the partially observed data is typically considered intractable. We will discuss recent advances that enable likelihood computations without model simplifications in the presence of missing infection and recovery times via efficient data-augmented samplers. Our methods target the exact posterior without relying on model-based forward simulation, and apply to several classic stochastic compartmental models and allow for disease-dependent contact networks to evolve dynamically. We apply our methods to high-resolution mobile contact tracking data from the eX-FLU study of influenza on a college campus and observational data from the COVD-19 pandemic.
|
