A1267
Title: Exact inference for stochastic epidemic models via uniformly ergodic block sampling
Authors: Raphael Morsomme - Duke University (United States) [presenting]
Jason Xu - Duke University (United States)
Abstract: Stochastic epidemic models provide an interpretable probabilistic description of the spread of a disease through a population. Yet, fitting these models to partially observed data is a notoriously difficult task due to the intractability of the likelihood for many classical models. To remedy this issue, a novel data-augmented Markov chain Monte Carlo algorithm is introduced for exact Bayesian inference under the stochastic susceptible-infectious-removed model, given only discretely observed counts of infections. In a Metropolis-Hastings step, the latent data are jointly proposed from a surrogate process carefully designed to resemble the target process closely and from which epidemics consistent with the observed data can be efficiently generated. This yields a method that efficiently explores the high - dimensional latent space and easily scales to outbreaks with thousands of infections. Further, this Markov chain Monte Carlo algorithm is proved to be uniformly ergodic, and it is observed to mix much faster than existing single-site samplers. The algorithm is applied to fit a semi-Markov susceptible-infectious-removed model to the 2013-2015 outbreak of Ebola Haemorrhagic Fever in Gueckedou, Guinea.
