Abstract: Brain-computer interface (BCI) technology, by establishing direct communication pathways between the brain and external devices, has brought groundbreaking advancements to the diagnosis, treatment and rehabilitation of neurological disorders. A typical BCI system comprises the core steps of "signal acquisition-decoding-control-feedback", an architecture that aligns closely with closed-loop neuromodulation systems, such as responsive neurostimulation (RNS) for epilepsy, which follows a "recording-decoding-intervention" paradigm. Thus, closed-loop neuromodulation can be viewed as an integrated form of BCI designed for specific therapeutic objectives. Currently, BCI has demonstrated significant potential in replacing, restoring and enhancing neural functions across various clinical domains, including the management of drug -resistant epilepsy, motor rehabilitation post- stroke, neuromodulation for Parkinson's disease, functional compensation in spinal cord injury, and objective assessment of consciousness disorder. However, the technology still faces multiple challenges, including biocompatibility, signal stability, algorithmic generalizability, clinical standardization, and ethical-regulatory considerations. This review systematically examines the clinical progress of BCI, with the aim of outlining its technological classifications (non-invasive, partially invasive, and invasive), current applications, and key issues. Furthermore, it explores future directions, such as high-precision bidirectional closed-loop interaction, multimodal neuromodulation integration, intelligent virtual rehabilitation systems, and international collaborative standardization. Ultimately, this review seeks to contribute to the evolution of BCI from an assistive tool toward an intelligent integrated paradigm, laying the groundwork for precision neurology.

Key words: Brain-computer interfaces, Neural networks, computer, Nervous system diseases, Electric stimulation therapy, Vagus nerve stimulation, Deep brain stimulation, Review

摘要： 脑机接口通过建立大脑与外部设备间的直接通信，为神经系统疾病的诊疗与康复带来突破。典型脑机接口系统包含信号采集-解码-控制-反馈环节，与闭环神经调控（如反应性神经电刺激）的记录-解码-干预架构高度一致，可视作面向治疗目标的集成化脑机接口。脑机接口业已在药物难治性癫痫调控、脑卒中后运动功能康复、帕金森病调控、脊髓损伤功能代偿及意识障碍评估等领域展现出替代、恢复与增强神经功能的潜力。然而，该项技术仍面临生物相容性、信号稳定性、算法泛化、临床标准化及伦理规范等挑战。本文系统综述脑机接口的临床应用进展，旨在梳理其技术分类、应用现状与关键问题，并展望其向高精度双向交互、多模态调控、智能虚拟康复及国际标准合作的发展方向，为推动脑机接口向智能融合范式演进、实现精准医学提供参考。

关键词: 脑-机接口, 神经网络,计算机, 神经系统疾病, 电刺激疗法, 迷走神经刺激术, 深部脑刺激法, 综述

CLC Number: 

• R741

• TP183