B1191
Title: A Bayesian nonparametric spiked process prior for dynamic model selection
Authors: Alberto Cassese - University of Florence (Italy) [presenting]
Weixuan Zhu - Xiamen University (China)
Michele Guindani - University of California Los Angeles (United States)
Marina Vannucci - Rice University (United States)
Abstract: In many applications, investigators monitor processes that vary in space and time, with the goal of identifying temporally persistent and spatially localized departures from a baseline or normal behavior. The monitoring of pneumonia and influenza (P\&I) mortality is considered to detect influenza outbreaks in the continental United States, and a Bayesian nonparametric model selection approach is proposed to take into account the spatio-temporal dependence of outbreaks. More specifically, a zero-inflated conditionally identically distributed species sampling is introduced, which allows borrowing information across time and assigning data to clusters associated with either a null or an alternate process. Spatial dependences are accounted for using a Markov random field prior, which allows informing the selection based on inferences conducted at nearby locations. The proposed modeling framework is shown to perform in an application to the P\&I mortality data and to a simulation study and compare with common threshold methods for detecting outbreaks over time, with more recent Markov switching-based models, and with spike-and-slab Bayesian nonparametric priors that do not take into account spatiotemporal dependence.