Time-Frequency Analysis of Daily Activities for Fall Detection

Fall events in elderly populations often result serious injury and may lead to long-term disability and/or death. While many fall detection systems have been developed using wearable sensors distinguish falls from other daily activities, sensitivity specificity decreases when exposed more rigorous activities such as running jumping. This research uses time-frequency analysis of accelerometer-only activity data develop a strategy for improving accuracy. In this study, wireless sensor system (WSS) consisting three-axis accelerometer, microprocessor communications module is used collect performed following script the laboratory setting. Experiments were conducted on 36 healthy human subjects performing four types (i.e., forward, backward, left/right sideway falls) well normal movements standing, walking, stand-to-sit, sit-to-stand, stepping, total, 1227 different collected analyzed. The algorithm computes magnitude accelerometer detect if critical threshold passed, then analyzes power spectral density within duration window (500 ms) differentiate activities. observed with high energy 2–3.5 Hz range distinct (3.5–5.5 Hz) jumping (1–2 Hz). Preliminary results indicate (PSD)-based can (98.4%) (98.6%) lab-based data. proposed has benefit improved accuracy over existing time-domain only strategies multisensor strategies.