Detection and Classification of Elbow Movements for Prosthetic Applications Using EMG Signals

Limb amputation presents significant physical and emotional challenges, impacting mobility and daily activities. To address this issue, prosthetic limbs, particularly myoelectric pros-theses, have been developed. These devices use muscle-generated electrical signals, known as electromyography (EMG) signals, to provide a more natural and precise interaction. This paper focuses on the software aspect of developing a control system for a myoelectric elbow prosthesis, evaluating various machine learning algorithms for interpreting surface electromyography (sEMG) signals. The objective is to identify the most effective algorithm for accurate sEMG signal interpretation, providing a foundation for future hardware integration. Among the evaluated models, the Random Forest algorithm achieved the highest accuracy of 90.63%, outperforming Support Vector Machine (SVM) with 86.46%, K-Nearest Neighbors (KNN) with 87.50%, and Decision Tree with 85.42%. This result demonstrates the potential of Random Forest as the optimal solution for prosthetic control systems.