Are the arms and head required to accurately estimate centre of mass motion during running?

Accurate measurement of centre of mass (CoM) motion can provide valuable insight into the biomechanics of human running. However, full-body kinematic measurement protocols can be time consuming and difficult to implement. Therefore, this study was performed to understand whether CoM motion during running could be estimated from a model incorporating only lower extremity, pelvic and trunk segments. Full-body kinematic data was collected whilst (n=12) participants ran on a treadmill at two speeds (3.1 and 3.9ms^-1). CoM trajectories from a full-body model (16-segments) were compared to those estimated from a reduced model (excluding the head and arms). The data showed that, provided an offset was included, it was possible to accurately estimate CoM trajectory in both the anterior-posterior and vertical direction, with root mean square errors of 5mm in both directions and close matches in waveform similarity (r=0.975-1.000). However, in the ML direction, there was a considerable difference in the CoM trajectories of the two models (r=0.774-0.767). This finding suggests that a full-body model is required if CoM motions are to be measured in the ML direction. The mismatch between the reduced and full-body model highlights the important contribution of the arms to CoM motion in the ML direction. We suggest that this control strategy, of using the arms rather than the heavier trunk segments to generate CoM motion, may lead to less variability in CoM motion in the ML direction and subsequently less variability in step width during human running.