An Approach for Detection of Walking Related Falls During Activities of Daily Living

The proposed framework at the intersection of Internet of Things (IoT), Big Data, Human Computer Interaction, Assistive Technologies and their interrelated disciplines aims to take a holistic approach towards detecting walking related falls in the constantly increasing elderly population during Activities of Daily Living (ADL). Walking is one of the common causes of falls in older adults [1] and one of the most common movements associated with ADL. Falls in elderly can have a multitude of impacts, both temporary and permanent and could even lead to death. Walking accounts for 36% of falls in elderly - making it the second-highest cause of elderly falls after transfers [1]. The main motivation of this work is addressing the global challenge of detecting a fall during walking in the context of ADL, so that necessary assistive and preventive measures can be taken to reduce the impacts of the fall. To evaluate the efficacy of the proposed approach it has been implemented on a dataset of ADL and its performance characteristics are discussed. A case study is also included which discusses the performance characteristics of different learning models to identify the learning model best suited for this approach. The results presented uphold the immense potential of this framework to address these global challenges through detection of walking related falls in elderly during ADL.