Stride length estimation from an ankle-worn IMU in Parkinson's disease

Abstract-Text (inkl. Referenzen und Bildunterschriften)

Introduction
Parkinson's disease (PD) is a neurodegenerative disorder whose treatment is mainly based on dopaminergic medication, and stride length has shown high responsiveness to levodopa-based treatment. Inertial measurement units (IMUs) can be used to measure gait, however, the accurate extraction of stride length is an open issue. The aim of this study was to estimate stride length from IMU data that was worn laterally above the ankle joint.
Methods
Seven subjects with PD walked five meter overground (at preferred, fast and slow speed), and if possible, then in medication "off" and "on" state. Subjects wore an IMU above each ankle joint, whilst reflective markers were placed on top of the usual footwear on the heel, toe, and ankle. The reflective markers were tracked using a twelve-camera optical motion capture (OMC) setup and provided reference stride lengths.
IMU data were preprocessed and segmented into consecutive strides¹, and subsequently the ankle trajectory was estimated². Stride length was computed for each stride and compared to the reference obtained from the OMC.
Results
A Wilcoxon signed-rank test showed IMU-based stride length was on average 5 cm shorter than the reference stride length (Fig 1 "Stride-level comparison of stride length") (stride length from IMU: 1.00 m, from reference: 1.06 m ). This difference was consistent across different stride lengths (Fig 2 "Bland Altman plot of OMC- and IMU-based stride lengths").
Summary
Stride length was estimated from signals recorded with an ankle-worn IMU in people with PD. Results are promising and argue in favor of the usefulness of stride length estimation in PD in the free-living environments. The methods will be further validated on subjects from other neurological cohorts and healthy adults, and the effect of walking speed will be explored.
References
1. Salarian, A. et al. IEEE Trans. Biomed. Eng. 51, 1434–1443 (2004).
2. Hori, K. et al. Front. Physiol. 10, (2020).