Development of Real Time Step Detection Towards Continuous Frailty Analysis

Abstract

Frailty is a geriatric syndrome characterized by decreased physical strength, unintended weight loss, fatigue, and low physical activity, requiring dependence on health services. Developing a wearable device for early diagnosis is crucial for timely intervention, which improves quality of life and reduces healthcare costs for aging populations. Our approach eliminates the necessity of routine clinical visits, enabling patients to be monitored accurately and consistently in the comfort of their homes. In this work, we develop software designed to identify step movements collected by gyroscope data and define a data window that is suitable for training and analysis through machine learning algorithms to recognize distinct characteristics in the gait data of individuals with frailty. This firmware development is part of a fully autonomous wearable Biosymbiotic device that tracks gait data of patients who are at risk of frailty to provide at home automated diagnosis. Continuous data collection powered by far field wireless power casting enables the device to operate seamlessly during daily activities without the need for user interaction, enabling a seamless patient experience without barrier even for subjects with cognitive and physical limitations.