Comprehensive Review and Analysis of Elderly Fall Detection System Using Machine Learning

Abstract

Immediate action is required due to the seriousness of the effects that might result from falls among the elderly. This groundbreaking open-source project uses machine learning techniques to quickly detect fall patterns by analyzing data from wearable sensors like magnetometers, gyroscopes, and accelerometers, as well as ambient data like temperature and humidity. A dataset with fourteen variables gathered from various subjects and activities is utilized in the research. These variables include age, sex, medical indications, and more. It is crucial to refine the model by adjusting the dataset, as shown by the testing results. There has to be a system in place to identify and prevent falls since the risk of falling rises as people age due to a deterioration in physical, cognitive, and sensory skills. This study examines and evaluates, using a number of criteria, the most recent fall detection and prevention systems that rely on machine learning. While it acknowledges wearable devices and support vector machines as typical instruments, it stresses the importance of doing wider investigations in varied circumstances. Future research directions to increase fall detection and prevention for the elderly include energy efficiency, sensor fusion, context awareness, and wearable design. The article also visualizes the performance metrics of ML algorithms in combination with various wearables.