NeuroXess Achieves Real-Time Mandarin Decoding Breakthrough with Implantable BCI, Published in Science Advances and Highlighted by Science and Nature

Recently, NeuroXess, in collaboration with Huashan Hospital Affiliated to Fudan University and the Shanghai Institute of Microsystem and Information Technology, Chinese Academy of Sciences, achieved a milestone breakthrough in brain-computer interface technology—their research outcome on "Real-time decoding and sentence synthesis of Mandarin using a high-throughput implantable flexible brain-computer interface" was officially published in the prestigious international journal *Science Advances*. The work was subsequently featured in-depth articles on the official websites of *Nature* and *Science*, marking China's entry into the forefront of international research in Mandarin-based brain-computer interfaces and laying the foundation for the global development of tonal language brain-computer interfaces.

Tackling International Challenges: Real-Time Decoding of Chinese Achieves "From 0 to 1" Breakthrough

As the tonal language with the largest number of speakers globally, Chinese, characterized by monosyllabic words, tone-based meaning, and a high density of homophones, poses a well-recognized challenge for real-time decoding in the brain-computer interface field. Previous related research has primarily focused on alphabetic languages like English. Additionally, given that approximately 8 million patients in China suffering from major brain diseases such as stroke and ALS have lost their ability to speak, developing a Chinese brain-computer interface tailored for the Chinese population is both critically important and urgently needed.

In response, the research team made targeted breakthroughs to develop a 256-channel high-throughput implantable flexible brain-computer interface. The micro-cortical electrodes can precisely cover key brain regions to collect neural signals. They also innovated a real-time neural network decoding algorithm, extracting specific frequency bands (using a 50-millisecond sliding window to extract 70-170Hz High-γ band neural activity) and ultimately integrated a language model to complete sentence output, forming a neural encoding and decoding solution adapted to the characteristics of the Chinese language.

Impressive Clinical Results: Achieving Daily Communication Needs in 9 Days of Training

This study selected a tumor epilepsy patient with a lesion in the language area. While helping to locate the epileptic focus and protect the language functional area, a clinical trial of Chinese language decoding was conducted. After only 9 days of training, the system demonstrated significant advantages: the average pure neural decoding accuracy for 394 commonly used Chinese syllables (covering over 98% of total Chinese syllables, with uncovered ones being rare syllables unfamiliar to the subject) reached 71.2%, the single-syllable decoding delay was just 65 milliseconds, and the real-time Chinese sentence decoding rate achieved 49.6 characters per minute, basically meeting daily communication needs.

Relying on this technology and the general-purpose brain-computer operating system independently developed by NeuroXess, the subjects achieved several breakthroughs: driving digital avatars to produce sound, engaging in real-time conversations with large AI models, and even precisely controlling dexterous hands to complete complex movements through brain electrolytic decoding commands. On the occasion of the New Year, the subject also used thoughts to output the sign language blessing "Happy 2025 New Year," intuitively showcasing the practical value and humanistic warmth of the technology.

Global Industry Recognition: Chinese Solutions Fill International Gaps

The findings gained significant attention from the international academic community as soon as they were published. The official website of *Science* released a dedicated article reporting on the study. Matthew Leonard, a neuro-linguistics expert at the University of California, San Francisco, pointed out: "This advancement opens the door to brain-computer interface applications for a large number of potential patients. There are more tonal languages in the world than non-tonal languages, and this is a major breakthrough in the field of Mandarin speech brain-computer interfaces in recent years."

Sergey Stavisky, a brain-computer interface researcher at the University of California, Davis, similarly affirmed: "There is a significant technological leap from studying speech in healthy subjects to developing medical-grade brain-computer interfaces for individuals with language disorders. The ability to generate Chinese characters 'online' during speech, rather than analyzing neural data post hoc, is itself a genuine advancement."

Notably, in July this year, the official website of the international authoritative journal *Nature* also specially reported on this research achievement, triggering widespread global attention and industry recognition.

This achievement not only fills the international gap in real-time decoding of the Chinese language but also provides a reusable technical paradigm for the development of brain-computer interfaces for tonal languages globally, laying a leading foundation for China in this frontier field of future industry.

Accelerating Clinical Translation: Benefiting 8 Million People with Speech Disorders as the Core Objective

"Results are just the starting point; the ultimate goal is to make technology benefit patients." Tao Hu, founder and chief scientist of NeuroXess, stated that in the future, the team will deepen the research and development of core technologies, optimize decoding accuracy and response speed, while accelerating clinical transformation. In collaboration with medical institutions, they will explore standardized application pathways for language disorder treatments.

The team also plans to expand the technical application boundaries, explore the deep integration of brain-computer interfaces and artificial intelligence, and help China advance from a "strong country in brain-computer interface technology" to a "strong country in application."

The research benefited from Shanghai's unique innovation ecosystem and diverse investments from the government, hospitals, institutions, etc. The relevant work received strong support from the Tianqiao and Chrissy Chen Institute (TCCI) and was supported by the Shanghai Science and Technology Commission's strategic frontier brain-computer interface special project, the Shanghai Municipal Commission of Economy and Informatization's high-quality industrial development special leading industry innovation and development project, the Chinese Academy of Sciences Youth Innovation Promotion Association outstanding member project, Fudan University's Artificial Intelligence Special Fund, and the Shanghai Magnolia Pujiang Program, among others.

Editor: Xu Lixin