Resumen de Validity and Reliability of Thoracic-Mounted Inertial Measurement Units to Derive Gait Characteristics During Running

Benjamin J. Horsley, Paul J. Tofari, Shona L. Halson, Justin G. Kemp, Daniel Chalkley, Michael H. Cole, Rich D. Johnston, Stuart J. Cormack

• Inertial measurement units (IMUs) attached to the tibia or lumbar spine can be used to analyze running gait but, with team-sports, are often contained in global navigation satellite system (GNSS) units worn on the thoracic spine. We assessed the validity and reliability of thoracic-mounted IMUs to derive gait characteristics, including peak vertical ground reaction force (vGRFpeak) and vertical stiffness (Kvert). Sixteen recreationally active subjects performed 40 m run throughs at 3-4, 5-6, and 7-8 m[middle dot]s-1. Inertial measurement units were attached to the tibia, lumbar, and thoracic spine, whereas 2 GNSS units were also worn on the thoracic spine. Initial contact (IC) from a validated algorithm was evaluated with F1 score and agreement (mean difference +/- SD) of gait data with the tibia and lumbar spine using nonparametric limits of agreement (LoA). Test-retest error {coefficient of variation, CV (95% confidence interval [CI])} established reliability. Thoracic IMUs detected a nearly perfect proportion (F1 >= 0.95) of IC events compared with tibia and lumbar sites. Step length had the strongest agreement (0 +/- 0.04 m) at 3-4 m[middle dot]s-1, whereas contact time improved from 3 to 4 (-0.028 +/- 0.018 second) to 7-8 m[middle dot]s-1 (-0.004 +/- 0.013 second). All values for Kvert fell within the LoA at 7-8 m[middle dot]s-1. Test-retest error was <=12.8% for all gait characteristics obtained from GNSS units, where Kvert was most reliable at 3-4 m[middle dot]s-1 (6.8% [5.2, 9.6]) and vGRFpeak at 7-8 m[middle dot]s-1 (3.7% [2.5, 5.2]). The thoracic-spine site is suitable to derive gait characteristics, including Kvert, from IMUs within GNSS units, eliminating the need for additional sensors to analyze running gait.