An Elderly Fall Detection System Using Enhanced Random Forest in Machine Learning

Fall detection systems play a key role in addressing the health risks faced by elderly individuals. This work implements a fall detection system using machine learning techniques. This work used a fall detection dataset and preprocessed it by encoding categorical variables using one-hot encoding and handling missing data about ADLs, and associated data were culled from the particular database used. Classifiers such as Support Vector Machine, Logistic Regression, Random Forest, AdaBoost, and Gradient Boosting (GB) were trained and evaluated using the dataset. Training the classifier on a split dataset allows for the evaluation of its performance using a variety of metrics. The components that comprise it are the confusion matrix, the F1 score, recall, accuracy, and precision. Furthermore, in order to determine the major elements that contribute to fall detection, the system displays the importance of certain features. Bar charts showing the relative importance of features, a heatmap showing the confusion matrix, and feature-specific box plots showing the distribution of data are all part of the visualizations included. The ERF model emerged victorious in a comparison of models, achieving the highest level of accuracy. The purpose of this fall detection system is to improve the well-being of the elderly by accurately detecting and reporting instances of falls.