Enhancing Elderly Care with Wearable Technology: Development of a Dataset for Fall Detection and ADL Classification During Muslim Prayer Activities

Abstract

Caring for elderly individuals, particularly those residing alone, is pivotal for cultivating a compassionate and inclusive society. The ageing population grapples with various challenges, necessitating additional support. A comprehensive and culturally sensitive dataset focusing on elderly individuals within Muslim communities is developed, contributing to the field of Activity of Daily Living (ADL) and fall detection. Utilising low-cost, lightweight wearable technology, the focus centres on inertial-based data for Activity of ADL classification and fall detection as a crucial research area. A culturally diverse dataset comprising 16 classes, specifically tailored for ADLs and fall detection during Muslim prayer movements, is gathered from a self-developed wearable device equipped with dual inertial measurement units (IMUs) on the waist and thigh, ensuring dependable and synchronised information. A Convolutional Neural Network (CNN) classification model is employed and rigorously tested for its effectiveness, revealing high performance with an average accuracy of 98.974% owing to the synchronised acquisition of data from the two IMUs. The acquired CNN model is adapted for deployment on a wearable embedded system, and authentic experiments are conducted, yielding precise outcomes. The results underscore the potential of wearable technology and advanced machine learning in enhancing elderly support and fall detection systems, fostering a safer and more empathetic environment for our ageing population.