In Sports biomechanics
The purpose of this study was to investigate the relative importance of net joint moments (NJM) in relation to bar kinematics during the clean. Ground reaction force and 3-D motion data were recorded as seven weightlifters performed cleans at 85% of their competition maximum, and were used to calculate hip, knee, and ankle NJM. Vertical bar kinematics were also calculated. NJM were used as inputs to a three-layer feedforward artificial neural network (ANN), which was trained to predict bar kinematics. Subject-specific ANN with 15 hidden neurons could effectively model the association between NJM and bar kinematics for each individual weightlifter (r: 0.965 ± 0.031; MSE: 0.169 ± 0.152). The relative importance (%) of hip, knee, and ankle NJM to bar velocity were 23%, 31%, and 46%, respectively, whereas the relative importance of hip, knee, and ankle NJM to bar acceleration were 23%, 39%, and 38%, respectively. Non-parametric statistics indicated that the ankle NJM exhibited the greatest relative importance in relation to bar velocity, whereas the knee and ankle NJM showed the greatest relative importance in relation to bar acceleration. These results indicate that the NJM produced at the knee and ankle joint are of great importance in contributing to bar kinematics during weightlifting.
Sports, artificial neural network, biomechanics, machine learning, weightlifting