Wearable Smart Phone Sensor Fall Detection System

• Published in: International journal of interactive mobile technologies Vol. 16; no. 12; pp. 72 - 93
• Main Authors: Mohamed Hadi Habaebi, Siti Hajar Yusoff, Anis Nadia Ishak, Md Rafiqul Islam, Jalel Chebil, Ahmed Basahel
• Format: Journal Article
• Language: English
• Published: 21.06.2022
• ISSN: 1865-7923; 1865-7923
• DOI: 10.3991/ijim.v16i12.30105

One of the most important measures that developed countries need along with economic success is the provision of contemporary health services for the elderly. Inadequate care for the elderly can put them in danger of falls with serious injury or even death. When an older person falls due to illness, immediate lack of care may lead to death. In this paper, a wearable smart phone sensor fall event detection system is introduced. The proposed system utilizes real-time raw sensory accelerometer, gyroscope and gravity data collected from the AndroSensor App and is then processed and applied to a machine learning classifier. The smart phone is positioned at the waistline to yield the most accurate fall detection results. The system is trained and can detect common activities, like sleeping, walking, sitting, jogging, and events, such as falling in any direction. An full accuracy is almost accomplished using Support Vector Machine classifier in comparison with other classifiers as addressed by other previous research works. Nevertheless, the system requires more validation with elderly people under the supervision of caregivers in a controlled environment.