Fall Detection for Elderly People Using Machine Learning

Abstract

The state of one's health is a major source of concern, and this unavoidable uncertainty only grows with age. As a result, caring for our aging population is a duty of great importance. To improve people's quality of life, technology is being used in this way. 'Fall' is one of the leading causes of health decline and death among the elderly. In light of the aforementioned problem, a novel system has been proposed to detect falls in the elderly using machine learning. While other methods, such as recording and processing webcam images, have proven useful for detecting falls, this study is the first to use a data set that includes information from sensors actually used by the elderly. It is possible to implant the sensor in an item like a belt or watches, and then use the data recorded from the sensor's activities and changes. Further, we have attempted to construct, using the flask framework, a prediction system that can determine whether or not the recorded sensor activity represents a fall, thereby allowing caregivers to take the necessary precautions. It has been found that support vector machines (SVM) and decision trees can be used for prediction, with the latter providing a high level of accuracy compared to other algorithms.