Remote Actigraphy for Quantitative Assessment of Walking Speed in People with MS
Published in American Academic of Neurology (ANN), 2016, 2016
Abstract. Objective: To design a wearable actigraphy device and enable subject-specific remote calibration for accurate estimation of walking speeds across a wide disability range. Background: Remote actigraphy has become popular a source of data regarding activity, but existing systems for estimating walking speed suffer high estimation errors because of individual differences in gait biomechanics. In people with MS, these errors vary with disability. We describe an approach to address this limitation. Methods: An AX3-Axivity triaxial-accelerometer (http://axivity.com/product/ax3) protected with silicone backing and attached to the trunk with surgical tape allows > 7 days of continuous data acquisition. Data analysis involves: 1) Feature extraction - deriving 29 features characterizing trunk acceleration with each step; 2) Initial calibration - Support-Vector-Regression subject-specific models relating features to gait speed during a Timed 25-Foot Walk and 2-Minute Walk; 3) Remote calibration - GPS positions from a smartphone app update calibration with “real-life” gait; 4) Database - patient-centre data repository in WikiHealth (http://www.wiki-health.org/about/overview.php). Results: Actigraphy data from 10 healthy (mean speed, 1.05±0.45m/s; error, 0.01±0.05m/s) and 7 MS subjects (EDSS scores 1.0-5.5, mean speed, 1.05±0.35m/s; error, -0.03±0.10m/s) was used to accurately estimate walking speeds. It also showed a potential for high estimation error if only the initial calibration is applied to “real-life” outdoor gait (mean speed, 1.09±0.49m/s; error, -0.17±0.16m/s). However, with a remote GPS calibration update, mean estimation error can be reduced to -0.01±0.06m/s. >80[percnt] of subjects tolerated chronic attachment of the device on the trunk. Conclusions: Remote actigraphy can accurately measure walking speed for both healthy volunteers and people with MS across a range of disability. This may provide a new tool for assessing treatment effectiveness based on a well accepted measure of disability. Study supported by: Progressive MS Alliance, Imperial College Healthcare Trust Biomedical Research Centre.