Recent Advances in Fall Detection and Prevention Using Machine Learning and IoT: A Comprehensive Review
J. Akilandeswari, A Velusamy · 2025
Traditional falls among elderly adults generate major health problems which result in severe injuries as well as independence loss while driving up healthcare expenses. Threshold-based sensor systems used traditionally exhibit insufficient accuracy together with slow response times which explain their inability to help during important scenarios. The use of machine learning (ML) and Internet of Things (IoT) technology to enable sophisticated proactive monitoring, realtime data analysis, and personalized intervention systems has resulted in significant advancements in fall detection and prevention. A synthesis of ML and IoT progress for fall detection analyzes ten core studies which appeared from 2018 to 2023. Because they provide noticeably higher detection accuracy, deep learning models—in particular, Convolutional neural networks (CNNs) and Recurrent neural networks (RNNs)—progressively gain traction. Because edge computing and wearable sensors, such as smart watches and inertial measurement units (IMUs), can guarantee low-latency processing, their application has grown widespread. The detection system faces ongoing obstacles since it requires solutions to address privacy concerns and scalability problems and require customized detection protocols. Future study directions on fall detection and prevention systems require solutions around multimodal sensor fusion and edge-ML deployment and user-specific modeling for optimal performance.