Abstract:As an emerging neural engineering technology, the brain-computer interface (BCI) acquires and decodes neural activity to convert it into control commands, enabling direct interaction between the human brain and external devices. In the field of Parkinson''s disease (PD), BCI technology has primarily developed along two pathways: closed-loop adaptive deep brain stimulation (aDBS) systems and EEG-based neurofeedback and rehabilitation training systems. In invasive BCI, closed-loop aDBS has entered clinical application. By recording neural activity in real time to dynamically adjust stimulation parameters, it offers significant advantages over traditional open-loop stimulation in terms of therapeutic efficacy, side effects, and energy efficiency. In non-invasive BCI, neurofeedback, motor imagery training, and cognitive rehabilitation provide patients with safe, portable, and home-based rehabilitation options. This article systematically reviews the major advances in the application of BCIs in Parkinson''s disease, evaluates the efficacy and limitations of each application direction, discusses current technical, clinical, and ethical challenges, and provides perspectives on future development trends.