EpiMix: A novel method to estimate effective reproduction number

Abstract

Transmission potential of a pathogen, often quantified by the time-varying reproduction number $R_t$ , provides the current pace of infection for a disease and indicates whether an emerging epidemic is under control. In this study, we proposed a novel method, EpiMix, for $R_t$ estimation, wherein we incorporated the impacts of exogenous factors and random effects under a Bayesian regression framework. Using Integrated Nested Laplace Approximation, EpiMix is able to efficiently generate reliable, deterministic $R_t$ estimates. In the simulations and case studies performed, we further demonstrated the method's robustness in low-incidence scenarios, together with other merits, including its flexibility in selecting variables and tolerance of varying reporting rates. All these make EpiMix a potentially useful tool for real-time $R_t$ estimation provided that the serial interval distribution, time series of case counts and external influencing factors are available.