Enhancing early outbreak response through generative AI-driven compliance modeling and context-specific policy design

PROJECT SUMMARY/ABSTRACT Emerging infectious agents pose persistent threats to public health, economies, and societies. Rapid response through non-pharmaceutical interventions (NPIs) can slow transmission, protect healthcare systems, and buy critical time for vaccine and therapeutic development. However, the effectiveness of NPIs depends heavily on human compliance, which varies across social and economic contexts. Current epidemic models rely on pre- defined compliance scenarios that fail to capture this variability, limiting their utility for real-time policymaking. While human behaviors have been extensively analyzed for the COVID-19 pandemic, predicting compliance for future unseen outbreaks remains difficult due to the absence of historical data and the incongruity of individual choices across social contexts. This project will address these challenges by combining behavioral surveys, large-scale mobility data, and generative artificial intelligence (AI). Building upon our extensive prior studies on infectious disease modeling, epidemic forecasting, human behaviors, and AI engineering, our interdisciplinary team proposes to pursue three aims. In Aim 1, we will quantify voluntary behavior changes and compliance with NPIs using surveys in diverse US cities under a variety of disease scenarios. We will benchmark results against mobile phone-derived foot-traffic data collected during the COVID-19 pandemic and future severe flu seasons. In Aim 2, we will develop interpretable large language model (LLM) agents that simulate human-like compliance behavior across different socioeconomic and policy contexts, validated against both survey and mobility data. In Aim 3, we will integrate these agents into behavior-driven epidemic models, enabling the design of optimized, context-specific policy bundles, and build an interactive decision- support dashboard for public health agencies. By combining behavioral science, generative AI, and real-world mobility data, this project will deliver open-source tools that realistically model compliance, improve policy design, and strengthen preparedness for emerging infectious diseases. The outcomes will enhance the effectiveness of outbreak response, minimize the disruption of NPIs on society, and save lives.