Bayesian Inference for the Evolving Epidemiological Characteristics: A Case Study of COVID-19 in Hong Kong

Emerging infectious diseases pose significant threats to global public health. Prompt detection and response are crucial for containing their spread. Understanding the evolving epidemiological characteristics of these diseases is key to timely and effective interventions. However, it remains challenging to accurately predict epidemiological parameters due to the complexity of epidemiological dynamics. We introduce a Bayesian inference approach based on state-space modeling with a stochastic Susceptible-Exposed-Infectious-Removed (SEIR) model to fit epidemic dynamics and infer epidemiological parameters. Specifically, our approach uses the Particle Markov Chain Monte Carlo (PMCMC) method to estimate key epidemiological parameters, including the transmission rate, the latent period, the infectious period, as well as the peak time of the epidemic. To validate the proposed method, we conduct a retrospective study on four waves of COVID-19 pandemic in Hong Kong, China. The results confirm that the PMCMC algorithm provides accurate predictions of epidemiological parameters, consistent with empirical research and literature. By accurately estimating epidemiological parameters, our approach is versatile and efficient, extending its utility beyond COVID-19.