NeuroX: A Hybrid EEG–EMG Digital Twin Framework for Upper-Limb Rehabilitation

Abstract

Upper-limb impairment caused by stroke and other neurological disorders can significantly reduce motor function and quality of life. Conventional rehabilitation approaches are predominantly therapist-guided and provide limited objective feedback regarding patient motor intention and functional recovery. This study presents NeuroX, a unified hybrid EEG–EMG digital twin framework that integrates electroencephalography (EEG)-based motor intention decoding, electromyography (EMG)-based movement verification, three-dimensional digital twin visualization, and a clinician-oriented dashboard within a single rehabilitation framework for objective patient assessment and longitudinal monitoring. Experimental validation demonstrated a mean trial-level EEG motor imagery classification accuracy of 87.5 ± 4.42% under five-fold offline subject-specific cross-validation, while real-time multi-subject deployment achieved an average live classification success rate of 74.6% across four participants, a mean EEG–EMG synchronization error of 27.9 ± 18.6 ms, and an end-to-end feedback delay of 128.4 ± 55.2 ms. EMG analysis further confirmed consistent muscle activation patterns during prismatic pinch and open-hand movements, supporting reliable verification of intended motor actions throughout rehabilitation sessions. These findings demonstrate the feasibility of integrating neural decoding, muscle activity verification, digital twin visualization, and clinician-oriented performance monitoring within a single rehabilitation framework. The proposed NeuroX framework establishes a foundation for future personalized neurorehabilitation systems by enabling objective, comprehensive, and clinically meaningful assessment of both motor intention and movement execution.