A smart system to detect non responsive state of elderly people.

Care of the elderly is becoming a relevant issue with the population ageing and the growth of nuclear families and the elderly being alone at home. The non-responsive or a motionless state of an elderly person is alarming, especially when the person is in unusual position and at location, such as, bathroom, passage or living room and needs to be detected to provide quick assistance. This paper presents a blend of classical machine learning and optical flow to detect the non-responsive state and the trigger positions of the elderly. The algorithm proposed in the paper is based on explainable AI. It takes in consideration 3 criterion namely pose, motion and context. The nonresponsive or motionless state for identified pose in an unusual context is identified as an alert for assistance to the elderly. The work proposes a smart system with low computational complexity that makes it easy to embed in a portable system. The results of functional and utility evaluation of the system under simulated environment are recorded. [ABSTRACT FROM AUTHOR]