LLM-enhanced agent-based modeling for mobility-based heat exposure assessment during mega events
Accepted by SIGSPATIAL 2026, this study proposes an LLM-enhanced agent-based framework linking human mobility with the thermal environment to anticipate dynamic heat exposure during mega events.
We are pleased to announce that our paper, “LLM-enhanced agent-based modeling for mobility-based heat exposure assessment during mega events,” has been accepted by SIGSPATIAL 2026.
This study proposes an LLM-enhanced agent-based framework that links human mobility with the thermal environment, providing a new approach to understanding and anticipating dynamic heat exposure during mega events. The framework combines LLM-based population forecasting, agent-based flow allocation, and mobility-based heat exposure assessment. By generating hourly crowd distributions, allocating travel flows, and connecting movement with dynamic thermal comfort conditions, it supports analysis of where and when people may experience heat exposure during large-scale events.