An Iphone Based System For Physical Activity Recognition And Fall Detection

Population aging is a common feature of developed countries. With the developmentof society, the movement safety of elderly has become a social problem. Therefore,recognizing people’s physical activity and fall detection have a broad range of applications.Recognizing people’s physical activity could help people maintaining their energy balanceby developing health assessment and intervention tools, investigating the links betweencommon diseases and levels of physical activity, and providing feedback to motivateindividuals to exercise. In addition, fall detection could avoid the serious effects andproblems caused by fall.This paper presents a system based on iPhone. With embedded accelerometer andgyroscope, the movement information was got in real-time for off-line analysis. Thisresearch was divided into four parts. First, evaluate different algorithms to find the mostfitted one which is C4.5algorithm. Second, in order to find the best sensor placement on ahuman body with one iPhone, data of different position were compared and chest was thebest position. Third, use the C4.5algorithm for physical activity recognition by data ofaccelerometer. Finally, use the C4.5algorithm and one-class Support Vector Machine(SVM) for fall detection by data of accelerometer and gyroscope.The result indicated an overall accuracy about80%through the laboratoryenvironment for distinguishing fall and physical activity. The system is small andexquisite. And it is stable and very convenient to use without influencing user’s normallife. It will make great contribution to improving the movement safety of the elderly.