Huashan Hospital Reports Breakthrough in Brain-Computer Interface: Paralyzed Patient Achieves Finger-Level Control via Thought

After the rise of brain-computer interface technology, Huashan Hospital Affiliated with Fudan University has collaborated with multiple enterprises to conduct clinical trials on the implantation of medical devices under research into the brain. At the main forum of the 2025 Brain-Computer Interface Conference held today, Huashan Hospital's president Mao Ying introduced the progress of these clinical trials and the large-scale EEG model jointly developed with corporate research teams. In his view, one future trend of brain-computer interface development is its integration with artificial intelligence, using large models to decode EEG signals and translate them into language and images, helping those who have lost speech to talk again and enabling the blind to regain sight.

Restore Partial Motor Function in Paralyzed Patients

"This is the latest advancement in clinical practice." At the beginning of his speech, Professor Mao Ying revealed to the audience the progress of China's first fully implanted, fully wireless, and fully functional brain-computer interface clinical trial conducted by Huashan Hospital in collaboration with NeuroXess: On October 30, a 28-year-old male patient who had been paralyzed for eight years due to high-level paraplegia underwent surgery, during which NeuroXess's ECoG motor brain-computer interface product was implanted in both his head and chest. He returned to a general ward just one day after the surgery, and five days post-operation, during the first activation training, he was able to control a computer cursor with his thoughts. After 17 days of training, his average pass rate in the 32×32 grid Webgrid test (a standardized experimental interface used to assess the accuracy and speed of thought-controlled operations) reached 5.2 BPS (bit rate), comparable to the level achieved by Neuralink, the company founded by Elon Musk.

NeuroXess's XessOS brain-computer operating system integrates dozens of games and other software applications. Currently, this patient can skillfully control "Mario Kart" with his mind and win in multiplayer online games like "Snake." After wearing a mechanical glove, he is able to control each finger of his hand through thought and can also "brain-control" smart home devices and humanoid robots.RobotMao Ying said, "In the implant surgery of this case, we used the first fully implantable and fully wireless brain-computer interface product with a built-in battery in China." This product, developed by NeuroXess, has achieved excellent clinical results in the field of real-time motion decoding. Previously, this Shanghai-based company founded by Dr. Tao Hu also realized real-time Chinese language decoding, allowing aphasia patients to communicate using their thoughts and generating Chinese characters in real time. A related paper was published in November this year in the prestigious international journal *Science Advances*. Motor control, language decoding, visual reconstruction, and neural modulation are four key tracks in the brain-computer interface industry. In recent years, a group of brain-computer interface companies founded by researchers have emerged in Shanghai, continuously exploring these four tracks and collaborating with medical institutions such as Huashan Hospital. At the Brain-Computer Interface Conference, Neuracle introduced that in May this year, the company's NEO wireless minimally invasive brain-computer interface system initiated a multi-center registration clinical trial across China. Currently, Huashan Hospital affiliated with Fudan University and Xuanwu Hospital affiliated with Capital Medical University, in collaboration with other clinical trial centers, have completed implant surgeries on 32 patients paralyzed due to cervical spinal cord injuries. All patients successfully achieved home-based brain-controlled grasping assistance and rehabilitation training, reaching 100% success at the primary clinical endpoint without any serious device-related adverse events. This preliminarily verifies the safety and long-term feasibility of the epidural minimally invasive implantation pathway.

Large Brainwave Model Brings Hope to People with Aphasia

The large-scale EEG model jointly developed by YanSi Neuro-Inspired and Huashan Hospital is also an important achievement. With the support of the municipal Science and Technology Commission, Huashan Hospital has established a high-quality intracranial EEG database using years of accumulated clinical resources and formed the iBrain EEG Data Alliance with several leading neurosurgical hospitals in China. Member units that have joined the alliance, such as YanSi Neuro-Inspired, can access high-quality intracranial EEG datasets for the development of brain-computer interface technologies and products. Mao Ying believes that one of the main bottlenecks in the development of brain-computer interfaces is the ability to decode EEG signals, which urgently requires the development of a generalizable EEG encoding-decoding framework. In this research direction, artificial intelligence can play a significant role. As a product of the intersection of brain-computer interface and artificial intelligence technologies, the large-scale EEG model will become a powerful assistant for researchers and clinicians, with its EEG decoding capabilities continuously improving. Similar to large language models, the development of large-scale EEG models also requires big data pre-training. In recent years, the substantial growth of brain data in China has provided excellent conditions for the development of large-scale EEG models. Interpreting brain signals related to language is an important function of the large-scale EEG model. According to reports, interpreting Chinese brain signals is much more difficult than English because English has only about 50 phonemes, while the combinations of initials, finals, and tones in Chinese result in over 400 phonemes. Relying on advanced multi-brain region stereotactic EEG collaborative decoding technology and supported by high-quality intracranial EEG datasets, YanSi Neuro-Inspired has used the large-scale EEG model to solve this complex language system problem. Currently, this Shanghai-developed large-scale EEG model achieves an accuracy rate of over 83% in recognizing Chinese initials and over 84% in recognizing finals, reaching a leading level in the industry.

In July this year, after 10 subjects had electrodes implanted at Huashan Hospital, they were able to make a computer connected to the brain electrodes display the Chinese sentences they wanted to say in real time through decoding their brain neural electrical activity after a short period of training. This clinical trial result will bring good news to patients with speech impairments such as ALS and stroke.

Mao Ying predicted: "On the road to the widespread application of brain-computer interfaces, large-scale EEG models will exhibit excellent performance in fields such as language decoding, image decoding, motion decoding, and smart device applications."

Original Title: Huashan Hospital President Reveals Clinical Progress in Brain-Computer Interface, Paralyzed Patients Can "Brain-Control" Each Finger