ESTIMATION OF LOWER LIMB KINETICS FROM LANDMARKS DURING SIDESTEPPING VIA ARTIFICIAL NEURAL NETWORKS

The purpose of this study was to determine the validity of kinetics estimated from 3D coordinates of landmarks during sidestepping by artificial neural networks (ANN). 71 male college professional soccer athletes performed sidestepping with two directions (left and right) and two cutting angles (45° and 90°) 3times for every task, totally 12 times. Coordinates of reflective markers, ground reaction forces (GRF) and lower limb joint moments were measured. All 18 body landmarks such as joints center were obtained by reflective markers as inputs to estimate GRF and lower joint moments in the ANN whose type was multilayer perceptron. The most of kinetics estimated by ANN showed strong correlation(r>0.9) with measured results. Just few kinetic curves of ANN existed significant differences in a few time points compared to measured results. ANN could accurately estimate kinetics from the coordinates of body landmarks druing sidestepping.