Reliability of running gait variability measures calculated from inertial measurement units

Jones, BDM, Wheat, J ORCID: https://orcid.org/0000-0002-1107-6452, Middleton, K, Carey, DL and Heller, B, 2025. Reliability of running gait variability measures calculated from inertial measurement units. Journal of Biomechanics, 180: 112515. ISSN 0021-9290

Text 2395593_Wheat.pdf - Post-print Full-text access embargoed until 3 January 2026. Download (215kB)

Abstract

Changes to the variability within biomechanical signals may reflect a change in the health of the human system. However, for running gait variability measures calculated from wearable device data, it is unknown whether a between-day difference reflects a shift in system dynamics reflective of a change in human health or is a result of poor between-day reliability of the measurement device or the biomechanical signal. This study investigated the reliability of stride time and sacral acceleration variability measures calculated from inertial measurement units (IMUs). Nineteen runners completed six treadmill running trials on two occasions seven days apart. Stride time and sacral acceleration signals were obtained using IMUs. Stride time variability and complexity were calculated using the coefficient of variation (CV) and detrended fluctuation analysis (DFA), respectively. Sacral acceleration regularity was quantified using sample entropy with a range of input parameters m (vector length) and r (similarity threshold). Between-day reliability was assessed using the intraclass correlation coefficient (ICC), standard error of measurement (SEM) and minimum detectable change. Stride time CV displayed moderate relative reliability (ICC = 0.672), but with a large absolute minimum detectable change = 0.525 %, whilst stride time DFA-α displayed poor relative reliability (ICC = 0.457) and yielded large minimum detectable changes (≥ 0.208). Sample entropy displayed good relative reliability in mediolateral and resultant sacral acceleration signals for certain combinations of the parameters m and r, although again with large minimum detectable changes. Researchers should be cognisant of these reliability metrics when interpreting changes in running gait variability measures in clinical contexts.

Item Type:	Journal article
Publication Title:	Journal of Biomechanics
Creators:	Jones, B.D.M., Wheat, J., Middleton, K., Carey, D.L. and Heller, B.
Publisher:	Elsevier BV
Date:	February 2025
Volume:	180
ISSN:	0021-9290
Identifiers:	Number Type 10.1016/j.jbiomech.2025.112515 DOI 2395593 Other
Divisions:	Schools > School of Science and Technology
Record created by:	Laura Borcherds
Date Added:	07 Mar 2025 10:03
Last Modified:	07 Mar 2025 10:03
URI:	https://irep.ntu.ac.uk/id/eprint/53201

Actions (login required)

Edit View

Statistics

Views

Downloads