FLARE is a novel framework that combines behavioral theories with large language models to predict evacuation decisions during wildfires, enabling more effective emergency management.

• Integrates Protection Motivation Theory (PMT) and Social Attachment Theory with LLMs
• Outperforms traditional statistical methods by understanding complex individual decision logic
• Provides explainable predictions through chain-of-thought reasoning
• Helps emergency managers anticipate traffic patterns and optimize resource allocation

This research advances public safety applications of AI by enabling more accurate prediction of human behavior during crisis situations, potentially saving lives through improved evacuation planning and emergency response.

From Perceptions to Decisions: Wildfire Evacuation Decision Prediction with Behavioral Theory-informed LLMs