Computing and Informatics

Brain-computer interface (BCI) technology holds immense promise in the rehabilitation of patients with movement disorders, leveraging the body's physiological mechanisms to enhance their quality of life by reshaping motor neural circuits through external devices. Nevertheless, current BCI applications for rehabilitation predominantly rely on a single physiological signal, often overlooking the synergistic impact of multiple signals. Simultaneously, while reinforcement learning shows significant potential for BCI applications, there exists a scarcity of studies exploring this intersection.This paper introduces a novel motor intention judgment model grounded in multimodal signal fusion and reinforcement feature selection. The model adeptly extracts comprehensive motor intention features by integrating pertinent information from both electroencephalogram (EEG) and electromyogram (EMG). Furthermore, reinforcement learning is employed for judicious feature selection, yielding promising outcomes in subsequent experiments. The study utilizes publicly available datasets to diagnose the motion intention of the subjects, complemented by ablation experiments to affirm the efficacy of the model components. In instances of feature-level fusion, the model demonstrated a noteworthy enhancement in the average five-classification accuracy, surpassing results obtained from isolated EEG and EMG experiments by 28.46 % and 12.68 %, respectively. The primary objective of this research is to furnish robust model support for motor rehabilitation training with exoskeletons, offering personalized solutions for the restoration of motor functions.