Prediction of Lower-Limb Kinematics Using Ground Reaction Force and Machine Learning Models

Machine learning (ML) advancements have made it possible to predict certain aspects of human locomotion. ANNs and other machine learning algorithms can be used to identify subject gait trajectories. The models that were built are the Multi-layer Perception (MLP) Neural Network, the Recurrent Neural Network (RNN), and the Long-Term Short-Term Memory (LSTM) Model. The ground reaction force (GRF) was used by the models as an input constraint when they were attempting to make predictions regarding the kinematics of the lower limbs. For this study, data are analyzed with respect to the sagittal plane. To determine how accurate the models were, a statistic called root-mean-square error (RMSE), and Coefficient of determination (R2) were utilized. It was determined that the root mean square error (RMSE) was ≤5.5°, and the (R2) value was obtained ≥0.8. On evaluating the results, the LSTM had the best performance, followed by the MLP and the RNN. The models that were produced have the potential to be used in the development of assistive equipment for those who are disabled. These technologies will make it easier for the disabled to lead regular lives and will assist them in reintegrating into society, aligned with Sustainable Development Goals (SDG-3) of good health and well-being.