IHBEM: No One Lives in a Bubble: Incorporating Group Dynamics into Epidemic Models

The dynamics of human behavior play a crucial role in the spread of epidemics. While much research has focused on individual reactions to risks and policies, this project examines how groups of people, such as households, communities, or organizations, demonstrate coordinated risk-mitigating behavior and make collective decisions during an epidemic. These group-level behaviors can significantly impact the trajectory of an epidemic, beyond what can be captured by aggregating individual behaviors. By studying group behaviors, such as the formation of social bubbles and changes in risk-mitigating norms and conventions, this research aims to create better mathematical models that reflect real-world social interactions. These models will help scientists and policymakers develop more effective strategies for managing epidemics, ultimately saving lives and reducing social and economic impacts. Additionally, insights from this research could inform policies on a range of issues including gun violence, opioid abuse, disaster response, and community resilience, where group behaviors play a critical role. The research concentrates on two main questions: 1) How can mathematical models and scalable computational algorithms be created to incorporate group-level behavioral responses in epidemic models? 2) How much do group-level responses significantly influence pandemic trajectories, and what are the resulting policy implications? The team plans to jointly work on several interconnected research thrusts. They will build mathematical foundations using a three-level network model and cooperative game theory to incorporate group-level behavioral responses, such as the formation and transformation of pandemic social bubbles and localized risk-mitigating norms within pandemic models. Next, they will create computational models that enable scalable and interpretable execution of these network-based approaches, developing dynamic networks using geospatial data and designing network downscaling algorithms to improve simulation efficiency. The team will use causal identification based on various natural experiments to estimate the input parameters of the models, focusing on empirically measuring perceived risk, peer effects on interaction networks, and the formation of social bubbles. Finally, they will implement and validate the model comprehensively at the county level in the US and at a more granular level in Boston neighborhoods, examining the policy implications of group-level behavioral responses. This award is co-funded by DMS (Division of Mathematical Sciences), SBE/SES (Directorate of Social, Behavioral and Economic Sciences, Division of Social and Economic Sciences), and SBE/BCS (Directorate of Social, Behavioral and Economic Sciences, Division of Behavioral and Cognitive Sciences). This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.