P2‐338: GAIT ANALYSIS OF ELDERLY PEOPLE WITH MILD COGNITIVE IMPAIRMENT USING SHANK MOUNTED INERTIAL SENSOR
Ahsan Shahzad, Kiseon Kim · Alzheimer s & Dementia · 2018
For early AD diagnosis, detection of MCI stage is critical. As gait is a complex cognitive task, gait abnormalities can be used to identify person's cognitive status. Currently, gait assessment is typically performed in a clinical setting using a 6m long electronic (pressure sensitive) walkway. In this study, we propose an alternative and cost effective solution for pervasive gait analysis of elderly people using inertial sensors. In addition, we evaluate a number of gait features, extracted from the inertial sensor signals, to find essential indicators for MCI diagnosis. 60 participants (HC = 30, MCI = 30), evaluated and checked by medical doctors at Chosun hospital, Gwangju, South Korea, performed the single- and dual-task walking experiments twice. For each experiment, subjects walked 10 m straight, turned around and came back 10 m. Two different dual tasks i.e., down counting decrement by one and saying animal names were considered. During the experiments, 9DoF inertial signals from Shimmer-3 sensors placed on mid-shank of each leg, were acquired wirelessly, and logged in nearby laptop. Out of 10 m, the central stable 6m walking signals were used for post processing. At first, the gait signals were segmented into gait cycles and then gait events (Toe-off, Heel-strikes) were found. After that, many gait features were extracted from each step signals. Gait features such as spatio-temporal, coefficient of variations (CV), kinematics, and non-linear features were analysed. Analysis of variance (ANOVA) test was performed to find the significantly different (p-value < 0.05) features. The ANOVA results (Table) shows that there exists many significantly different features both under single- and dual-task walking scenarios. However, dual–task scenario provides better discrimination between MCI and HC subjects. The MCI people presented significantly higher swing time, stance time, stride time, CV of swing time, CV of stride time while having lower cadence and gait speed in comparison with the HCs. The inertial sensor based pervasive gait analysis method provides many assistive markers or indicators to help MCI diagnosis and facilitates the early detection of AD.