Identifying the infectious period distribution for stochastic epidemic models using the posterior predictive check

Abstract

Under the Bayesian framework, we develop a novel method for assessing the goodness of fit for the SIR (susceptible-infective-removed) stochastic epidemic model. This method seeks to determine whether or not one can identify the infectious period distribution based only on a set of partially observed data using a posterior predictive distribution approach. Our criterion for assessing the model’s goodness of fit is based on the notion of Bayesian residuals.