Core Viewpoint - The article highlights a significant breakthrough in brain-computer interface (BCI) technology achieved by Shanghai YanSi Brain-like Artificial Intelligence Research Institute in collaboration with Huashan Hospital of Fudan University, enabling patients with speech impairments to communicate through thought [1][2]. Group 1: Technological Breakthroughs - The BCI technology allows real-time decoding of brain activity to display intended Chinese phrases, marking a major advancement in language decoding for patients with conditions like ALS and stroke [1]. - The research has been recognized in the prestigious journal "Nature," showcasing China's progress in invasive BCI technology, including real-time motion and language decoding capabilities [1]. Group 2: Technical Details - The decoding of Chinese brain signals is more complex than English due to the higher number of phonetic components, with over 400 compared to English's 50 [2]. - The YanSi Brain-like Institute utilizes a multi-region collaborative decoding technique and a large intracranial EEG dataset to achieve over 83% accuracy in recognizing Chinese consonants and over 84% in recognizing vowels, positioning it at the forefront of the industry [2]. Group 3: Future Applications - The BCI technology not only aims to restore communication abilities for speech-impaired patients but also has potential applications in controlling physical devices through thought, interacting with the metaverse, and generating artistic expressions from mental imagery [2]. Group 4: Implantation Techniques - The founder of Brain Tiger Technology emphasizes the challenge of balancing brain function utilization with minimizing damage during invasive BCI implantation, opting for flexible electrodes placed beneath the dura mater to enhance signal quality while reducing brain injury [3].

Core Insights - The article discusses significant advancements in brain-computer interface (BCI) technology, particularly in decoding Chinese language brain signals, which is notably more complex than English due to the higher number of phonemes involved [1][2]. Group 1: Technological Breakthroughs - Shanghai Yansi Brain AI Research Institute has collaborated with Huashan Hospital to achieve breakthroughs in BCI, enabling real-time display of intended Chinese phrases from brain activity [1]. - The BCI technology developed by Yansi Brain utilizes a multi-region collaborative decoding technique and a large intracranial EEG dataset, achieving over 83% accuracy in recognizing Chinese consonants and over 84% in recognizing vowels, marking a leading position in the industry [2]. Group 2: Clinical Applications - The advancements in BCI technology are expected to benefit patients with speech impairments, such as those suffering from amyotrophic lateral sclerosis (ALS) and stroke, by restoring their ability to communicate [1][2]. - The technology also holds potential for broader applications, including controlling intelligent devices through thought and interacting with the metaverse [2]. Group 3: Implantation Techniques - The founder of Brain Tiger Technology emphasizes the importance of balancing functionality and minimizing damage in invasive BCI techniques, opting for flexible electrodes implanted beneath the dura mater to reduce brain damage while obtaining higher quality brain signals [3].

记者 俞陶然 见习记者 李昂 最近，上海岩思类脑人工智能研究院与复旦大学附属华山医院合作，在脑机接口领域取得突破。10 例受试者的大脑植入电极后，经过短时间训练，通过解码其大脑神经电活动，与脑部电极相连的电脑就 会实时显示出他们想说的中文语句。这一进入临床试验阶段的科研成果，将为渐冻症、脑卒中等失语患 者带来福音。 脑机接口是实现大脑与外部设备信息交互的交叉前沿技术，在医疗、康养、教育等领域有广阔的应 用前景。近年来，上海在全国率先开展脑机接口关键技术的系统性布局，培育产学研医深度融合的创新 生态，已涌现出一批国内领先的科研成果。 "在这个全球最大的人类颅内脑电数据库支持下，我们团队已研发出脑电大模型，它类似于 ChatGPT的底层预训练模型，能精准读取脑电信号的意图。"岩思类脑首席科学家李孟教授说。 临床试验显示，脑电大模型可在54个汉字训练集基础上，解读出1951个常用汉字，外推率高达 1∶36，具有很强的泛化性。半秒钟内，大模型就能解读出一个完整的中文语句，且语句字数没有限制， 理论上可达到很高的输出率。 李孟认为，这一脑机接口技术突破意味着能把人的想法准确、高效地转化成文字。今后，它不仅能 让失语患者 ...

新京报讯（记者赵利新）3月31日，中关村论坛年会"脑机接口创新发展与应用论坛"上，脑机接口产业 联盟发布了2025脑机接口十大创新成果。 成果1："北脑一号"完成国际首批柔性高通量半侵入式无线脑机系统的人体植入 单位：北京脑科学与类脑研究所、北京芯智达神经技术有限公司 点评："北脑一号"智能脑机系统完成了128通道微型、无线、低功耗全植入系统研发，联合北京大学第 一医院、首都医科大学宣武医院、首都医科大学附属北京天坛医院，针对脊髓损伤、脑卒中等导致的运 动功能障碍开展精细运动解码，以及渐冻症等导致的失语开展中文语言解码，进行临床验证。 成果2："三合一"颅骨植入式脑脊接口微型设备 单位：复旦大学类脑智能科学与技术研究院 点评：首创"三合一"脑脊接口技术，通过4小时微创手术，经颅骨微孔植入直径仅1毫米的柔性电极，创 伤面积缩小90%。其AI算法可毫秒级解析信号，32个物理通道模拟256个虚拟通道，精准捕捉0.1μv级神 经电信号，下肢肌群控制准确率达92.7%，术后24小时患者就能恢复腿部运动。 成果3：双环路协同演进脑机接口系统 单位：天津大学 点评：首次提出了脑机协同演进概念，并基于忆阻器神经形态芯片完成了 ...