Brain–computer interface (BCI) is a cutting-edge interactive technology promising for restoring motor function in motor neuron diseases (MND) patients. Combining virtual reality (VR) with BCI has shown better results in rehabilitation training, yet research and practical applications remain limited, especially in Vietnam. This research aims to propose and develop a comprehensive VR-BCI system to aid stroke patients in recovering motor neuron function. The proposed system integrates a deep learning model to analyze electroencephalogram (EEG) signals, translating them into inputs for a VR environment tailored for rehabilitation, featuring avatars of the patient and a physical therapist. This approach seeks to improve neuroplasticity and motor function recovery through immersive interactions, also advancing the development of effective rehabilitation tools, particularly in developing country with limited access to advanced medical technologies such as Vietnam.