Machine Learning- and Deep Learning-Based Myoelectric Control System for Upper Limb Rehabilitation Utilizing EEG and EMG Signals: A Systematic Review

Tala Zaim, Sara Abdel-Hadi, Rana Mahmoud, Amith Khandakar, Seyed Mehdi Rakhtala, Muhammad E. H. Chowdhury

Upper limb disabilities, often caused by conditions such as stroke or neurological disorders, severely limit an individual’s ability to perform essential daily tasks, leading to a significant reduction in quality of life. The development of effective rehabilitation technologies is crucial to restoring motor function and improving patient outcomes. This systematic review examines the application of machine learning and deep learning techniques in myoelectric-controlled systems for upper limb rehabilitation, focusing on the use of electroencephalography and electromyography signals. By integrating non-invasive signal acquisition methods with advanced computational models, the review highlights how these technologies can enhance the accuracy and efficiency of rehabilitation devices. A comprehensive search of literature published between January 2015 and July 2024 led to the selection of fourteen studies that met the inclusion criteria. These studies showcase various approaches in decoding motor intentions and controlling assistive devices, with models such as Long Short-Term Memory Networks, Support Vector Machines, and Convolutional Neural Networks showing notable improvements in control precision. However, challenges remain in terms of model robustness, computational complexity, and real-time applicability. This systematic review aims to provide researchers with a deeper understanding of the current advancements and challenges in this field, guiding future research efforts to overcome these barriers and facilitate the transition of these technologies from experimental settings to practical, real-world applications.