Identifying memory mechanisms in Bayesian models of behavioural change during epidemics

Area of Science:

• Epidemiology
• Mathematical Biology
• Computational Statistics

Background:

• Accurate epidemic modeling necessitates understanding behavioral changes in response to perceived infection risk.
• Existing models often oversimplify the impact of memory on behavior, focusing primarily on recent events.

Purpose of the Study:

• To introduce a unified framework of Memory Mechanism Enhanced Behavioural Change (MEBC) models within a Bayesian SIR framework.
• To examine five distinct memory mechanisms influencing behavioral responses to epidemic risk.
• To improve the accuracy of epidemic dynamics modeling by incorporating collective memory.

Main Methods:

• Developed a Bayesian SIR epidemic modeling framework incorporating MEBC.
• Investigated memoryless, sliding window, power-law, exponential, and reciprocal memory mechanisms.
• Employed a Bayesian data-augmented Markov Chain Monte Carlo (MCMC) scheme for parameter estimation.

Main Results:

• MEBC models accurately recover parameters and demonstrate robustness even with misspecified memory structures.
• Application to COVID-19 and influenza outbreaks shows substantial improvement in model fit when memory mechanisms are included.
• Identified the critical role of collective memory in shaping behavioral adaptation and epidemic transmission.

Conclusions:

• Incorporating realistic memory mechanisms significantly enhances epidemic modeling capabilities.
• The MEBC framework provides a more nuanced understanding of behavioral feedback loops in disease dynamics.
• Collective memory is a crucial factor in predicting and managing infectious disease outbreaks.