Enhancing Safe Mobility: Development of a Fall Detection System for Smart Walkers

The World Health Organization estimates 684,000 annual deaths and 37.3 million severe falls, emphasizing vulnerability in individuals aged 60 and above. Falls not only cause physical injuries but also contribute to psychological and social challenges, impacting independence and increasing social isolation. Cognitive and neurological diseases increase the risk of falls in the elderly, necessitating effective fall detection systems. Near falls, precursors to actual falls, are crucial indicators for predicting risks. Improper use of walking aids, like canes and walkers, adds to fall risks among older adults. Smart walkers, incorporating advanced technologies like sensors and actuators, offer a potential solution. This thesis presents a study integrating an efficient fall detection system into a smart walker, using a multi-sensor setup with a wireless Inertial Measurement Unit sensor, Force-sensitive handlebars, and a Laser Range Finder sensor. The thesis objectives aim to analyse fall risk factors, assessment, detection, prevention, and their application for people with dementia. Specifically, a fall detection system based on thresholding algorithms and inertial sensors was integrated into a smart walker and a dedicated safe experimental protocol was developed. A real-time gait analysis, facilitated by sensor-specific algorithms, was implemented to provide clinical insights and tested with 6 able-bodied participants. Several state-of-the-art fall detection algorithms were compared to evaluate the overall system, achieving an accuracy rate of up to 73.9%. The development of the multi-sensor setup and real-time algorithm aimed to enhance daily fall detection, addressing the urgent need for effective fall detection systems in an aging population. The study outcomes significantly fluctuate depending on the case study, highlighting the need for personalization of the system for the specific user.