Development and Validation of a New Method to Measure Walking Speed in Free-Living Environments Using the Actibelt® Platform

Abstract

Walking speed is a fundamental indicator for human well-being. In a clinical setting, walking speed is typically measured by means of walking tests using different protocols. However, walking speed obtained in this way is unlikely to be representative of the conditions in a free-living environment. Recently, mobile accelerometry has opened up the possibility to extract walking speed from long-time observations in free-living individuals, but the validity of these measurements needs to be determined. In this investigation, we have developed algorithms for walking speed prediction based on 3D accelerometry data (actibelt®) and created a framework using a standardized data set with gold standard annotations to facilitate the validation and comparison of these algorithms. For this purpose 17 healthy subjects operated a newly developed mobile gold standard while walking/running on an indoor track. Subsequently, the validity of 12 candidate algorithms for walking speed prediction ranging from well-known simple approaches like combining step length with frequency to more sophisticated algorithms such as linear and non-linear models was assessed using statistical measures. As a result, a novel algorithm employing support vector regression was found to perform best with a concordance correlation coefficient of 0.93 (95%CI 0.92–0.94) and a coverage probability CP1 of 0.46 (95%CI 0.12–0.70) for a deviation of 0.1 m/s (CP2 0.78, CP3 0.94) when compared to the mobile gold standard while walking indoors. A smaller outdoor experiment confirmed those results with even better coverage probability. We conclude that walking speed thus obtained has the potential to help establish walking speed in free-living environments as a patient-oriented outcome measure.

A feasibility study for the integration of 3D accelerometry in fall risk assessment

“…The correlation between gait speed and age adds to the construct validity of our algorithms [9] and the results obtained with the acceleration stabilogram confirm the expectations and provide more information compared to the traditional methods used to assess postural stability. The outcome obtained with the traditional methods are commonly based on a "yes/no" score depending on the patient's ability to maintain balance over a time thresholdexamples of this are the usual outcomes for clinical balance tests as the Romberg, Tandem/Semi-Tandem or Onelegged test.…”

“…2464 rapid clinical functional tests (Table 1) performed by 224 elderly women with osteoporosis (mean age 68.3 +/-7.7 years old) were collected in 4 European centers in France, Germany, Italy and Switzerland using a high precision 3D accelerometer [8][9][10] developed at the Sylvia Lawry Center for Multiple Sclerosis Research (SLCMSR), Munich. The data collection started in July 2010 and the SLCMSR database was internally freeze for exploratory analysis in December 2011; however, collection and transfer of new actibelt ® data continued in parallel.…”

“…In our study we used a custom-built 3D accelerometer with a sample frequency of 100 Hz, the actibelt ® [8][9][10]. The device is integrated in a belt buckle and placed near the center of mass of the human body ( Fig.…”

“…All the data from the rapid test was pre-processed by noise and gravity filtering using a pass-band Butterworth filter (n = 4, fmin = 1/6 Hz, fmax = 5 Hz for test 1-6 and fmax = 20 Hz for test [7][8][9][10][11]. The parameters extracted from the actibelt ® signal are described in Table 1.…”

Accuracy of the actibelt® accelerometer for measuring walking speed in a controlled environment among persons with multiple sclerosis

“…Data were downloaded to a local PC and uploaded to the central actibelt 1 web-server for off-line post-processing and feature extraction. An algorithm written in R [6] was used to calculate the time dependent walking speed in m/s for every single 6MW measurement sequence from the actibelt 1 raw data. There was no individualized calibration of the actibelt 1 nor were patient characteristics passed along with the actibelt 1 data.…”