A novel kinematic detection of foot-strike and toe-off events during noninstrumented treadmill running to estimate contact time

Abstract

Contact time (tc) relies upon the accuracy of foot-strike and toe-off events, for which ground reaction force (GRF) is the gold standard. However, force plates are not always available, e.g., when running on a noninstrumented treadmill. In this situation, a kinematic algorithm (KA) – an algorithm based on motion capture data – might be used if it performs equally for all foot-strike angles across speeds. The purpose of this study was to propose a novel KA, using a combination of heel and toe kinematics (three markers per foot), to detect foot-strike and toe-off and compare it to GRF at different speeds and across foot-strike angles. One hundred runners ran at 9 km/h, 11 km/h, and 13 km/h. Force data and whole-body kinematic data were acquired by an instrumented treadmill and optoelectronic system. Foot-strike and toe-off showed small systematic biases between GRF and KA at all speeds (≤5 ms), except toe-off at 11 km/h (no bias). The root mean square error (RMSE) was ≤9 ms and was mostly constant across foot-strike angles for toe-off (7.4 ms) but not for foot-strike (4.1–11.1 ms). Small systematic biases (≤8 ms) and significant differences (P ≤ 0.01) were reported for tc at all speeds, and the RMSE was ≤14 ms (≤5%). The RMSE for tc increased with increasing foot-strike angle (3.5–5.4%). Nonetheless, this novel KA computed smaller errors than existing methods for foot-strike, toe-off, and tc. Therefore, this study supports the use of this novel KA to accurately estimate foot-strike, toe-off, and tc from kinematic data obtained during noninstrumented treadmill running independent of the foot-strike angle.