AI Framework for Suicide Risk Assessment in Metro Stations

Researchers have developed an interpretable AI framework for suicide risk assessment (SRA) in metro stations using video surveillance. The framework, detailed in a preprint on arXiv (2605.22904), analyzes passenger behavior, spatial context, and temporal dynamics from surveillance camera footage. Unlike prior approaches that focus on isolated subtasks or direct intent inference, this method accumulates evidence over time to assess risk. The work addresses challenges in human motion perception, platform geometry understanding, and aggregation of heterogeneous behavioral cues. The framework aims to enable early identification of high-risk situations for timely intervention in suicide prevention.