Accidental Fall Prediction and Detection in Elderly Persons using Ensemble Techniques

Accidental falls in the elderly have gradually become a major health concern requiring reliable prediction and timely detection.This study has adopted the use of artificial intelligence ensemble learning techniques to assist in tackling this critical issue.Specifically, bagging techniques were deployed with Random Forest (RF), Logistic Regression (LR), Support Vector Classifiers (SVC), and Decision Tree (DT).The dataset used in the study comprised both physiological and environmental-related data that serve as indicators for falls.Results from the study yielded the best performance with the bagging technique applied on the Random Forest, Logistic Regression, and Support Vector Classifiers, which yielded an accuracy of 96%.The bagged Decision Tree model also performed significantly with an accuracy of 93%.The model was deployed using Flask, with the integration of SMS alerts and a dashboard notification feature.The deployed system demonstrates potential as a valuable tool in ensuring early fall detection in the elderly by reducing the risks of sustaining injuries, enhancing safety, and improving the overall wellbeing and quality of life of elderly individuals.