Falls are the most critical health problem for elderly people, which are often, cause significant injuries. To tackle a serious risk that made by the fall, we develop an automatic wearable fall detection system utilizing two devices (mobile phone and wireless sensor) based on three axes accelerometer signals. The goal of this study is to find an effective machine learning method that distinguish falls from activities of daily living (ADL) using only a single triaxial accelerometer. In addition, comparing the performance results for wearable sensor and mobile device data .The proposed model detects the fall by using seven different classifiers and the significant performance is demonstrated using accuracy, recall, precision and F-measure. Our model obtained accuracy over 99% on wearable device data and over 97% on mobile phone data.
Alzoubi, H., Ramzan, N., Shahriar, H., Alzubi, R., Gibson, R., & Amira, A. (2016). Optimization and evaluation of the human fall detection system. In Remote Sensing Technologies and Applications in Urban Environments Proceedings, International Scoiety for Optics and Photonics  SPIE. https://doi.org/10.1117/12.2242162