MindSpeak: Clinical Breakthrough in Brain-Computer Interface for Intention-Based Chinese Speech Attracts Global Academic Attention

Author |Yicai Global, Qian Tongxin

Reporters from the First Financial learned that the Shanghai Yanxu Neocortex Artificial Intelligence Research Institute (hereinafter referred to as "Yanxu Neocortex") has made a breakthrough in the field of brain-computer interface in collaboration with Huashan Hospital Affiliated to Fudan University.

After 10 subjects had electrodes implanted in their brains, they were able to make a computer connected to the brain electrodes display in real time the Chinese sentences they wanted to say by decoding their brain's electrical activity after a short period of training, achieving "mind-to-speech." This research achievement, which has entered the clinical trial stage, will bring good news to patients who have lost the ability to speak due to conditions such as ALS and stroke.

This breakthrough represents the latest development achievements in China's brain-computer interface language decoding technology. Not long ago, the international authoritative academic journal *Nature* provided an in-depth report on the breakthrough progress in China’s invasive brain-computer interface field, mentioning two clinical trial results from NeuroXess, including real-time motion decoding and real-time Chinese language decoding technology.

Brain-computer interface (BCI) is a cutting-edge interdisciplinary technology that enables information interaction between the brain and external devices, offering broad application prospects in fields such as healthcare, rehabilitation, and education. In December last year, a patient with language dysfunction successfully communicated in Chinese for the first time after being implanted with NeuroXess's 256-channel flexible brain-computer interface.

Interpreting brain electrical signals in terms of language is an important function of large-scale brain electrical models. According to reports, interpreting Chinese brain electrical signals is much more difficult than English because English has only about 50 phonemes, while the permutations and combinations of initials, finals, and four tones in Chinese result in over 400 phonemes.

NeuroXess Relies on Multi-Brain Region Stereo EEG Collaborative Decoding Technology and Intracranial EEG Dataset to Solve the Complex Language System Problem with Its Self-Developed Large EEG Model. Currently, the large EEG model's recognition accuracy for Chinese initials exceeds 83%, and the recognition accuracy for finals exceeds 84%, reaching a leading level in the industry.

"With the support of the world's largest human intracranial EEG database, our team has developed a large EEG model, which is similar to the underlying pre-trained model of ChatGPT and can accurately read the intent of EEG signals," said Professor Meng Li, Chief Scientist of NeuroXess.

Li Meng believes that this breakthrough in brain-computer interface technology means that human thoughts can be accurately and efficiently converted into text. In the future, it will not only allow aphasic patients to regain their ability to express themselves verbally but also enable them to gain other abilities through decoding neural activity in the brain to generate text, such as controlling physical smart devices with their minds, interacting with the metaverse, or using multimodal large models to paint, transforming the language in their minds into artwork that expresses their dreams.

In terms of the implantation method for brain-computer interfaces, Tao Hu, founder and chief scientist of NeuroXess, believes that achieving a balance between utilizing brain functions and avoiding damage is the core challenge of invasive brain-computer interface technology. NeuroXess has chosen to implant flexible electrodes beneath the dura mater. Compared to inserting electrodes into brain tissue, this approach significantly reduces brain damage. At the same time, compared to implantation on the dura mater, placement below the dura mater is closer to the brain tissue, enabling the acquisition of higher quality and more precise EEG signals.

WeChat Editor| Summer Wood

Source: Yicai

For academic sharing only. If there is any infringement, please leave a message and we will delete it immediately!

Welcome to the Brain-Computer Interface AI Planet to access more knowledge and resources in the fields of brain-computer interface + AI.

For cooperation inquiries, please add WeChat: RoseBCI [Note: Name + Industry/Major].

   Welcome to submit articles   

1. Welcome to submit articles. For submission inquiries, please contact WeChat: RoseBCI

Click to Submit:Submission Guidelines for Academic News in the Brain-Computer Interface Community

2. To join the community as a part-time creator, please contact WeChat: RoseBCI

One-click triple action: "Share," "Like," and "Read"

Don't miss any cutting-edge developments in brain-computer interface