Incorporating dynamic health behaviors in spatial agent-based models of disease spread

Abstract

Agent-based models (ABMs) serve as valuable tools to simulate the characteristics and interactions of autonomous agents within their environment. Disease spread ABMs, specifically, are essential to better predict disease outcomes and respond to outbreaks. However, many existing ABMs ignore or generalize the behavior component, a key driver of disease dynamics. Many ABMs that do incorporate behavior often lack a connection to human behavioral theories and data, and implement it in an ad-hoc manner, which oversimplifies the dynamics of human behavior. Therefore, this study aims to expand upon an agent decision-making framework to include individuals' dynamic health behaviors. This framework is based on a logistic regression model using odds ratios to determine the probability of adopting a behavior. To demonstrate dynamic behaviors, mask usage was focused on. Four variables were used to predict this behavior for each agent: income, political party, race, and worry about the disease. The framework was tested in a previously developed spatial ABM that simulates COVID-19 spread among the student population on George Mason University’s campus, specifically examining the impact of individuals’ worry about COVID-19 on mask usage. The spread of worry occurs when an agent is in the same classroom as an already worried agent. Three scenarios were tested changing the probability of becoming worried: 25%, 50% and 100%. Preliminary results show the spread of agent perceptions of the disease, causing individuals to ‘worry,’ has a direct effect on disease outcomes. The more worry among the population, the less cases overall. This framework demonstrates the flexibility to incorporate individuals’ perception based on other variables informed by behavioral constructs and data, making it adaptable to model various infectious diseases and health behaviors. Improving the representation of health behaviors in ABMs of disease spread can help prevent disease outbreaks and guide policy decisions.