Real-Time Mental Condition Monitoring and Early Suicide Warning assisted by UAV

Real-time mental health monitoring and suicide prevention require accurate multimodal behavioral analysis and ultra-low response latency, yet existing cloud-centric systems are constrained by communication delays, limited sensor coverage, and the lack of unified cross-layer optimization. In this paper, we propose the UAV-assisted Cloud-Edge Collaborative Framework (UAV-ACE), which leverages unmanned aerial vehicles as mobile edge nodes to support large-scale, low-latency psychological state assessment. To ensure high accuracy and low delay, the framework integrates a Transformer-TCN model for stable mental-state prediction, and formulates UAV scheduling as a latency-aware joint optimization problem involving visiting order, hovering positions, and time allocation. A genetic algorithm and successive convex programming are employed to achieve efficient trajectory planning and communication delay reduction. Experiments demonstrate that UAV-ACE significantly reduces end-to-end latency while improving prediction accuracy, showing its potential for real-time mental health monitoring and early suicide intervention.