Night-time falls pose a severe threat to elderly independence, yet current monitoring systems remain reactive, triggering alerts only after incidents occur. This project proposes NightGuard, a Predictive Night Risk Modeling System that transitions elderly care from emergency response to proactive risk management. Central to our approach is the Personalized Night Movement Risk Index (NMRI), which establishes a unique 7-day behavioral baseline for each user by analyzing gait stability, nocturnal frequency, and hesitation patterns. Unlike traditional binary alarms, NightGuard calculates a continuous risk score (0.0-1.0) using edge-based time-series forecasting to detect subtle behavioral drifts. This enables the system to predict fall probability up to 3 days in advance, empowering caregivers to intervene before accidents happen while providing nursing institutions with data-driven insights for dynamic care optimization. This study refers to the contactless mmWave micro-motion sensing and edge perception methods proposed by Professor Zhang Daqing’s research team to improve sensing accuracy and privacy protection.
