Classification of Fall Detection Based on Daily Living Activities using Machine Learning Approach

Population aging is a worldwide phenomenon, with agricultural regions housing the greatest proportion of aged individuals in the labor force per unit of population. Nonetheless, in the area of agricultural machinery, very few comprehensive investigations on farmer falls have been carried out. The implementation of classification techniques for monitoring devices to identify farmworkers' falls and non-fall movements is the main topic of this study. Agricultural biomechanical parameters are taken into account while identifying everyday living activities. In this study, 40 healthy participants who conducted a activities of daily living (ADLs) and range of falls had their initial acquisition datasets of signals containing collected and processed signals from one gyroscope and two accelerometers. Using spatial features, the machine-learning classifiers were trained to discriminate between fall and non-fall events. The efficacy of the suggested strategy was assessed by supervised machine learning experiments. The k-nearest-neighbors (KNN) and support vector machine (SVM) algorithms were able to distinguish between falls and ADLs (binary-class classification) with ROC auc-scores of 0.999.