Neuro-Motor Index for Upper Limb Exoskeleton Control: A Machine Learning Approach
Résumé
This paper presents a hybrid machine learning model for controlling an upper limb exoskeleton to help children with cerebral palsy. The model integrates Support Vector Regression (SVR) and Long Short-Term Memory (LSTM) networks to predict joint angles based on electromyography (EMG) signals. An important innovation is the introduction of the Neuro-Motor Index (NMI), which combines the Co-Contraction Index (CCI) and kinematic joint data to improve the accuracy of movement prediction. The model was evaluated using cross-validation and showed significant improvements in mean absolute error (MAE), root mean square error (RMSE) and coefficient of determination R2 compared to single models. These results emphasize the potential of the model for improving exoskeleton-assisted rehabilitation. Future work will focus on clinical validation and real-time optimization.