Advancing stroke rehabilitation: the potential and challenges of closed-loop brain-computer interface technology

Stroke is one of the leading causes of long-term disability in older worldwide. As an emerging neuromodulation intervention, closed-loop brain-computer interfaces (BCIs) aim to promote the reconstruction of the damaged cortex through real-time feedback mechanisms. This study aims to systematically review the latest clinical advancements, neural mechanisms, and challenges of closed-loop BCIs in post-stroke rehabilitation.

Following the PRISMA guidelines, this study systematically searched databases including PubMed, Web of Science, Cochrane Library, Embase, Scopus, and IEEE Xplore. Given the high methodological heterogeneity in intervention paradigms and outcome measures across different studies, a qualitative synthesis strategy was employed. The minimal clinically important difference (MCID) was introduced to evaluate the substantive clinical benefits of various interventions. Ultimately, 42 original studies meeting the strict definition of closed-loop systems were included.

Closed-loop BCI technology demonstrates multi-dimensional application potential in stroke rehabilitation. In motor rehabilitation, BCIs combined with external actuators (e.g., robotics, FES) promote interhemispheric functional rebalancing and corticospinal tract remodeling. In the cognitive domain, although neurofeedback has shown initial efficacy in improving specific executive functions and attention, the current evidence exhibits high heterogeneity and requires cautious interpretation. Regarding safety, adverse reactions to non-invasive devices primarily manifest as mild fatigue; for invasive systems, the incidence of device-related adverse events is approximately 5.6 per 1,000 device-days, indicating overall controllable safety.

Closed-loop BCIs provide a promising novel neuromodulation strategy for stroke rehabilitation. Future validation of their efficacy and acceleration of clinical translation will rely on multicenter randomized controlled trials (RCTs), standardized core outcome sets (COS), and deep integration with artificial intelligence (AI).