The head is an excellent proxy for the whole body center of mass when measuring running velocity in competition

Whole body digitizing used to calculate whole body center of mass (CM) variables from competitions is particularly time-consuming, and shortcut methods that substitute for it could expediate the calculation of spatiotemporal variables. The aim of this study was to measure the appropriateness of using the head as a proxy for the CM when calculating running velocity in competition. Fifty-six athletes in the IAAF World Championship marathons were recorded using two high-definition cameras (50 Hz) on two laps so that 112 running sequences were analyzed. The video files were imported into SIMI Motion and manually digitized. The horizontal running velocity during one gait cycle was obtained using four methods: horizontal velocity of the CM; horizontal velocity of the head (raw data); horizontal velocity of the head (Butterworth filtered); and horizontal displacement of the head (a single measurement using SIMI Motion 3D still image measurement) divided by time taken. In comparison with the criterion CM measurements for mean horizontal velocity, the filtered head data had the best 95% confidence interval (95% CI) for intraclass correlation coefficient (ICC) (0.999 - 1.000), the least bias (-0.006 m/s), and the lowest root mean square difference (0.024 m/s). The filtered head condition also had the best 95% CI for ICC for maximum and minimum horizontal velocities during the stride (> 0.988) and the lowest bias (-0.001 m/s and -0.003 m/s, respectively). With the application of an appropriate filter, the head is thus an excellent proxy for whole body CM velocity calculations. (c) 2021 Elsevier Ltd. All rights reserved.

Automated summary

By analyzing running sequences from the IAAF World Championship marathons, this study found that using the head as a proxy for the center of mass (CM) in calculating running velocity can produce comparable results to CM measurements, with better intraclass correlation coefficient and smaller bias values.