Pilot Evaluation of Stage-Based VR Training with EMG Control in Upper-Limb Amputees

Gyounghyun Kim¹Heejun Youn²Hyojin Kim³Joonpio Hong³Changsik Pak³Hyunsuk Suh³Soonchul KWON^2*

• ¹Kwangwoon University, Seoul, Republic of Korea
• ²Kwangwoon University, Nowon-gu, Republic of Korea
• ³Asan Medical Center, Songpa-gu, Republic of Korea

Surface electromyography (sEMG) has emerged as a promising approach for controlling prosthetic devices in upper-limb amputees, yet effective training methods that can strengthen residual muscle activity remain limited. This study developed a virtual reality training system that interprets motor intent using sEMG signals and deep learning classification to provide stage-based rehabilitation tasks for upper-limb amputees. Three progressive training stages were designed to explore functional adaptation through intention-driven interaction. In this pilot study, a six-week intervention involving three amputees and three controls suggested that distinguishable motion-specific muscle activation patterns were preserved despite reduced muscle mass from amputation, enabling reliable motion classification. Repeated VR training was associated with progressive increases in residual muscle activation among amputee participants, indicating preliminary evidence of neuromuscular adaptation. These findings provide a feasibility assessment of EMG-based VR rehabilitation as an engaging, stage-based approach, while highlighting the need for larger-scale studies to validate its clinical impact.