Validity of an Inertial Measurement Unit System for Evaluating Gait Quality in Healthy Adults and People With Stroke

Abstract

Ambulation in the community is more challenging than walking indoors on level ground due to variable conditions including uneven terrain and slopes, but particularly so for people with hemiparesis due to stroke. Measurement of the quality of gait based on spatial-temporal parameters can provide insight to determine the capacity for community ambulation; however, instrumentation to assess gait is mostly limited to laboratory settings over flat surfaces. Wearable inertial measurement units (IMUs) have the potential to measure spatial-temporal gait parameters in community settings, but few studies have been conducted to determine their validity. The objective of this study was to determine the concurrent validity of an IMU system compared to the GAITRite system in characterizing spatial-temporal aspects of gait in healthy adults and adults with hemiparesis secondary to stroke on level, inclined, and downward sloped surfaces. A cross-sectional validation study including two groups (stroke, n=11; healthy, n=12) who completed the same three walking conditions (level ground, inclined slope, and downward slope) was conducted. The agreement of spatial-temporal outcomes associated with the IMU system relative to the GAITRite system was analyzed using a three-way mixed ANOVA, Bland-Altman plots, root mean squared error (RMSE), and coefficients of variation (CV). No significant differences between the IMU and GAITRite systems were found for measures of step length, step length symmetry, and speed but swing time was consistently overestimated, and stance time was consistently underestimated compared to the GAITRite system. The agreement between IMU and GAITRite systems was not affected by surface condition (incline or downward slope); however, examination of trends indicated that the agreement between IMU and GAITRite systems may be lower for spatial-temporal parameters measured from participants with hemiparesis compared to healthy adults. Further, the variability (RMSE) associated with IMU-derived measurements tended to be higher than those from the GAITRite for the stroke group, and there was higher variation (CV) associated with the IMU system, relative to the GAITRite, among stroke participants. The major findings indicate that the IMU system requires improvements in measurement accuracy before it can be used as a valid measurement system for spatial-temporal gait parameters in healthy and hemiparetic populations.