Passive Tracking of Gait Biomarkers in Older Adults: Feasibility of an Acoustic Based Approach for Non-Intrusive gait Analysis

Gait parameters have been established as a health biomarker for physical and cognitive health for older adults. Extracting these parameters however requires specialist equipment such as on-body sensors or video-based systems. Identifying low-cost techniques to capture gait parameters can open the opportunity for in-home monitoring of gait related biomarkers. In this study, we demonstrate that gait can be analysed in in-door at-home settings using only the sound of footsteps. To establish a comparative baseline, we use inertial measurement units (IMUs) and video footages as reference systems alongside our proposed acoustics-based approach. Gait parameters, specifically cadence, step time, and stride time, were extracted from audio, video, and IMU data streams recorded from 10 community-dwelling older adults. Bland Altman Analysis was performed to assess the agreement between the parameter values from the three different systems. We found that the gait parameters derived from our acoustics-only system exhibit relative standard errors of 0.71% and 0.58% against IMU and video-based systems, respectively.