Advances in mechanisms of neuroplasticity induced by multimodal closed-loop brain-computer interfaces after stroke

Post-stroke motor dysfunction is one of the leading causes of acquired disability worldwide. The induction and maintenance of neuroplasticity constitute the core mechanisms underlying motor function recovery. Conventional open-loop brain-computer interfaces (BCIs) lack real-time closed-loop feedback and are therefore unable to reliably activate the "temporal contingency" principle required by Hebbian synaptic remodeling, resulting in limited rehabilitation efficacy. Multimodal closed-loop BCIs integrate motor intent decoding, functional electrical stimulation (FES), virtual reality (VR), and exoskeleton-mediated proprioceptive feedback to construct a complete sensorimotor closed-loop circuit. These systems can precisely induce activity-dependent synaptic plasticity, facilitate cortical reorganization, and ameliorate interhemispheric inhibitory imbalance. The present review systematically examines the theoretical foundations of neuroplasticity induction by multimodal closed-loop BCIs following stroke, the constituent system components, electrophysiological and neuroimaging evidence, and the key factors modulating neuroplasticity induction efficacy. Future directions toward personalized adaptive closed-loop systems and long-term home-based rehabilitation are discussed. This review integrates converging evidence from electroencephalography, functional magnetic resonance imaging, transcranial magnetic stimulation, and randomized controlled trials to establish a comprehensive mechanistic framework for multimodal BCI-mediated neuroplasticity, and provides reference for both basic research and clinical translation in this field.