Validation of a single wearable sensor to monitor performance during clinical measures of gait and balance
M.D. Hidalgo Araya (TU Delft - Mechanical Engineering)
Heike Vallery – Mentor
Arun Jayaraman – Mentor
Megan O'Brien – Mentor
Manon Kok – Graduation committee member
P.M. Baines – Graduation committee member
More Info
expand_more
Other than for strictly personal use, it is not permitted to download, forward or distribute the text or part of it, without the consent of the author(s) and/or copyright holder(s), unless the work is under an open content license such as Creative Commons.
Abstract
Deterioration of gait and balance, whether from aging, disease, or injury, has been linked to reduced mobility and increased risk of falling. Wearable sensing technologies, such as inertial measurement units (IMUs), may augment clinical assessments by providing continuous gait and balance data at an increased resolution. The objective of this work was to validate spatiotemporal gait features with a single IMU sensor and to examine changes in sensor-derived features with age during the common clinical tests of gait and balance. We tested the use of an IMU place in the lower back (L5) on age-ranged, healthy individuals (N=34, 20-70 years) during the 10-meter walk test (10MWT), Timed Up and Go (TUG), and Berg Balance Scale (BBS). A total of 49 features were derived from the sensors based on a novel selection of algorithms from previous works. Six spatiotemporal gait features were validated against gold standard measures to assess accuracy and bias. There was an excellent agreement for step time, stance time, swing time, and step count (ICCs 0.90–0.99), and good agreement for gait velocity and step length (ICCs 0.84–0.88). There were 33 linear correlations between age and the sensor-derived features, including a negative correlation between age and vertical displacement of the center of mass during gait. The strongest correlation with age was found for the first slope of the second turn in the TUG (r=-0.545, p≤0.001). For the features that showed moderate correlations (|r|>30, p<0.05), a hierarchical multivariate regression model showed that age was the most important predictor independent of weight, height, and gender. Furthermore, when looking at gender-specific differences after correcting for the contribution of weight and height, women exhibited 5-fold more correlations compared to men. In conclusion, sensor-derived features demonstrated greater sensitivity to individual differences in gait and balance, which may be of a particular interest for future implementation in a clinical setting in impaired populations.
The structure of this thesis is as follows: The first chapter contains the project overview, future vision, and specific aims. The second chapter contains a manuscript that will be submitted for peer-reviewed journal publication. Finally, the third chapter is an appendix with more detailed explanations of the methods and findings of this work.