Revisiting sensor-based intelligent fall risk assessment for older people: A systematic review

Falls are a major public health concern among older people due to their high prevalence and severe consequences. Identifying individuals at high risk of falling through fall risk assessment is a fundamental step to implement effective fall prevention strategies. Recent advancements in off-the-shelf human sensing technologies have spurred a surge in sensor-based intelligent fall risk assessment. Existing reviews often overlook non-wearable technologies, free-living environments, and geriatric populations. To address these limitations and capture emerging research trends, this systematic review provides a comprehensive analysis of current literature and outlines future research prospects. Thirty-two relevant papers retrieved from major databases were critically reviewed and analyzed, presenting a variety of faller identification criteria, experimental cohorts, sensing devices, test protocols, artificial intelligence (AI) modeling techniques. Even though accumulated evidence from this review demonstrated that sensor technologies (e.g., inertial sensor, depth camera, radar, pressure sensor) combined with AI algorithms hold significant promise for objective, accurate and convenient fall risk assessment, inconsistencies in study methodologies hinder definitive conclusions about their ability to predict future falls. Moreover, the overall performance on large-scale cohorts remains relatively poor, with a mean accuracy of 63.9%. Additionally, practical standalone applications integrating motion sensing, model inference, and diagnostic reports are underdeveloped, hindering widespread deployment of fall risk assessment among older population. Future research should focus on high-risk geriatric populations, contactless and low-cost motion sensing, prospective multi-center protocol design, multifactorial test protocols, advanced AI models with explainable mechanisms, and user-centric applications to enhance fall risk assessment for older people.