Wearable Pre-Impact Fall Detection System Based on 3D Accelerometer and Subject’s Height

This study presents a low-power wearable system able to predict fall by detecting pre-impact condition, performed through simple analysis of motion data (acceleration) and height the subject. The can detect in all directions with an average consumption 5.91 mA; i.e., it monitor activity daily living (ADL), whether or not occurs. entire detection uses single tri-axis accelerometer placed on waist for comfort wearer during long-term application. algorithm is based following hypothesis: “A region defined as balanced boundary circle, user’s height, characterized fact chance that actual happening minimal. When classified outside this acceleration determine impending condition”. Our threshold-based was validated experimentally, first 9 young healthy volunteers performing both normal ADL activities then using 10 5 falls from public SisFall dataset. results show could be detected lead-time 259 ms before impact occurs, minimal false alarms (97.7% specificity) sensitivity 92.6%. good achieved thus far detection, permitting integration our inflatable airbag hip protection.