Publication

Abstract : In the domain of upper limb prosthetics, the precise identification of electromyography (EMG) signals is crucial for intricate motor pattern recognition. This research delves deeply into Myoelectric Pattern Recognition within Hand Prosthetics, concentrating on a spectrum of Machine Learning (ML) methodologies—K-Nearest Neighbors (KNN), Random Forest, Support Vector Machine (SVM), Long Short-Term Memory (LSTM), Naïve Bayes, XG Boost, Convolutional Neural Network (CNN), and Artificial Neural Network (ANN). The study emphasizes the critical role of accurately decoding EMG patterns to augment control mechanisms for assistive devices. Employing ML algorithms, distinctive features extracted from EMG signals undergo rigorous classification. A meticulous analysis unveils KNN as exhibiting superior accuracy in discerning intricate hand gestures encompassing various hand and wrist movements—rest, clenched fist, wrist flexion, wrist extension, radial and ulnar deviations, and extended palm—outperforming its ML counterparts. This outcome underscores KNN’s efficiency in interpreting EMG patterns, highlighting its potential to refine the control of upper limb prosthetics. The investigation accentuates the prowess of ML techniques in enhancing feature recognition within EMG signals, ultimately positioning KNN as a frontrunner in elevating the precision of hand prosthetic control systems. This exploration underscores the substantial impact of KNN in advancing the capabilities of assistive technologies aimed at restoring motor functions for individuals with upper limb disabilities.