Brain–computer interface-assisted pneumatic hand training improves upper limb function after stroke: a randomized controlled trial and EEG characterization

Abstract Background The clinical efficacy of brain–computer interface (BCI)–based rehabilitation after stroke remains uncertain, particularly in the subacute phase, where evidence is limited and clinically meaningful improvement is seldom evaluated. This study aimed to investigate the therapeutic efficacy and electrophysiological mechanisms of BCI-assisted pneumatic hand training in subacute stroke. Methods In this prospective randomized controlled trial, 129 patients with stroke were assigned to a BCI group ( n = 72) or a control group ( n = 57). Upper limb function was assessed using FMA-UE, ARAT, WMFT, and MBI. Clinically meaningful improvement was evaluated based on minimal clinically important difference (MCID). Resting-state EEG was analyzed for spectral power and topographic distribution. Results The BCI group demonstrated greater improvement in FMA-UE scores than the control group (11.57 ± 7.98 vs 5.94 ± 6.13, p = 0.01). A higher proportion of patients in the BCI group achieved MCID in FMA-UE, ARAT, and WMFT. EEG analysis revealed increased beta-band power following BCI intervention, predominantly over central and parietal regions. Between-group comparisons showed greater beta-band changes in the BCI group. Post-treatment, no significant differences were observed between the BCI group and healthy controls, suggesting a normalization trend. Conclusion BCI-assisted pneumatic hand training enhances upper limb recovery in subacute stroke and is associated with increased beta-band activity, indicating potential neurophysiological mechanisms of motor recovery.