CAREER: AI-Driven Role-Aware Modeling of Cascading Disaster Impacts in Sociotechnical Systems

This Faculty Early Career Development Program (CAREER) project advances a new paradigm of human-centered disaster impact assessment that captures how individual adaptive choices interact with infrastructure systems to produce cascading societal impacts. Disasters disrupt the daily roles people perform that hold communities together. When infrastructure services fail, households must make difficult trade-offs across these roles. For example, an essential worker may be unable to participate in disaster recovery because the closure of a critical facility or infrastructure disruption forces them home to provide elder care or childcare. These role-based adaptations ripple through communities, slowing recovery for everyone. Yet most disaster impact assessments treat people as passive recipients rather than active decision-makers shaping disaster outcomes. The project aims to (1) introduce a role-aware adaptation model that integrates established behavioral theories with empirical disaster data to characterize how individuals adapt across multiple social roles during cascading infrastructure disruptions; (2) develop a predictive model that captures role-based adaptation at both individual and population scales, with bias auditing for vulnerable subgroups; and (3) advance an LLM-enabled generative agent-based modeling platform that bidirectionally couples human adaptation with infrastructure performance through service demand, workforce participation, and behavioral cascade feedback channels. The artificial intelligence methods at its core generalize across hazards and regions without extensive new local data, providing particular value for resource-limited and vulnerable communities. The resulting framework provides emergency managers, community planners, and policymakers with the capacity to anticipate disaster impacts across populations and identify recovery strategies transferable across hazards, regions, and time. Integrated educational activities introduce future engineers and emergency managers to responsible AI use and human-centered disaster resilience while building early STEM interest among young learners. 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.