Heterogeneity Learning for SIRS model: an Application to the COVID-19

TL;DR

This paper introduces a Bayesian heterogeneity learning method for the SIRS model to identify regional differences in COVID-19 transmission, recovery, and immunity loss rates, providing both parameter estimates and clustering insights.

Contribution

It develops a hierarchical Bayesian framework with mixture priors for the SIRS model, enabling simultaneous parameter inference and clustering of regions based on COVID-19 data.

Findings

Effective in identifying regional heterogeneity in COVID-19 transmission dynamics

Provides accurate parameter estimates and cluster configurations

Demonstrates good performance through extensive simulations

Abstract

We propose a Bayesian Heterogeneity Learning approach for Susceptible-Infected-Removal-Susceptible (SIRS) model that allows underlying clustering patterns for transmission rate, recovery rate, and loss of immunity rate for the latest coronavirus (COVID-19) among different regions. Our proposed method provides simultaneously inference on parameter estimation and clustering information which contains both number of clusters and cluster configurations. Specifically, our key idea is to formulates the SIRS model into a hierarchical form and assign the Mixture of Finite mixtures priors for heterogeneity learning. The properties of the proposed models are examined and a Markov chain Monte Carlo sampling algorithm is used to sample from the posterior distribution. Extensive simulation studies are carried out to examine empirical performance of the proposed methods. We further apply the proposed methodology to analyze the state level COVID-19 data in U.S.