Bayesian time-varying autoregressive models of COVID-19 epidemics

Area of Science:

• Epidemiology
• Biostatistics
• Statistical modeling

Background:

• The COVID-19 pandemic underscored the need for robust statistical models for accurate forecasting and policy impact analysis.
• Existing models may not fully capture the dynamic nature of disease spread and policy interventions.

Purpose of the Study:

• To propose and validate advanced Bayesian time-dependent Poisson autoregressive models.
• To estimate the impact of policy measures on disease counts using time-varying coefficients.

Main Methods:

• Development of Bayesian time-dependent Poisson autoregressive models with time-varying coefficients.
• Application of the models to analyze new positive COVID-19 case data from Italy and the United States.

Main Results:

• The proposed models effectively captured nonlinear growth patterns in disease counts.
• Significant reductions in disease counts were observed following the implementation of closure policies and vaccine distribution in both countries.

Conclusions:

• Bayesian time-dependent Poisson autoregressive models provide a reliable framework for analyzing epidemic data and evaluating policy interventions.
• Policy measures, including lockdowns and vaccination campaigns, demonstrated a significant impact on mitigating disease transmission.