A two-component model for counts of infectious diseases.

Area of Science:

• Epidemiology
• Biostatistics
• Mathematical Modeling

Background:

• Infectious disease surveillance systems collect time series data on notifiable diseases.
• Analyzing these data requires models that account for both endemic and epidemic patterns.
• Existing models may not adequately capture the dynamic nature of disease transmission.

Purpose of the Study:

• To propose a flexible stochastic model for analyzing time series of disease counts.
• To incorporate both parameter-driven (endemic/seasonal) and observation-driven (epidemic) components.
• To allow for time-varying autoregressive parameters using a Bayesian changepoint model.

Main Methods:

• A Poisson or negative binomial observation model with two components.
• Parameter-driven component for endemic/seasonal patterns.
• Observation-driven component with autoregression and a Bayesian changepoint model for unknown changepoints.
• Bayesian model averaging using Markov chain Monte Carlo (MCMC) for parameter estimation.

Main Results:

• The proposed model effectively analyzes simulated and real-world infectious disease surveillance data.
• Demonstrated ability to capture endemic seasonality and epidemic dynamics.
• Identified time-varying autoregressive parameters indicating changes in transmission patterns.

Conclusions:

• The developed stochastic model provides a robust framework for analyzing infectious disease surveillance data.
• It enhances understanding of disease dynamics by separating endemic and epidemic influences.
• The model offers valuable tools for public health surveillance and policy-making.