Discovering Knowledge by Behavioral Analytics for Elderly Care
Bruce X. B. Yu, Keith C. C. Chan · 2017
Population aging is a phenomenon affecting many developed and developing countries in the world and many of them are expecting to face various social and economic problems as a result. In response, there have been effort to improve elderly care. Particularly, preventing dementia and taking better care of dementia patients are considered very important. In this paper, we present a system we developed to recognize indoor daily routines of elderly people so that their needs and interests can be better served. Our system uses a Kinect v2 sensor network covering the whole indoor living area of an elderly. Daily routine is the highest level of activity recognition derived based on the duration and complexity of the movement captured in a Kinect network. The relationship between movement and activities are discovered using both features extracted from sensor data and a machine learning method called AdaBoost. Based on the activities discovered, the system can be useful for assisting elderly people to better live their lives with automatic reminders, alarm services related to safety and security, the feeding of healthy information, and even in the discovering of dementia related symptoms. With the Kinect v2 sensor network, the proposed system will work even under poorly illuminated conditions and even if the elderlies do not use any wearable sensors.