Brain-Computer Interfaces Enter Serious Medical Applications: AI Drives a Quantum Leap in EEG Signal Decoding

Source: Shanghai Securities News

Xinhua Finance Shanghai, May 25th, by reporter Emily Yi and Reporter He Yi. In May, China's brain-computer interface (BCI) industry has ushered in another wave of development. Policies and capital continue to flow into the BCI sector, with domestically invasive devices being approved for medical insurance and non-invasive rehabilitation equipment continuously being implemented. The industry is accelerating into the commercialization phase of serious medical applications. Academician of the Chinese Academy of Sciences, Executive Vice President of Nanjing University, and Director of the National Innovation Center for High-Performance Medical Devices, Zheng Hairong, recently stated: "Brain-computer interfaces have become an important engine for developing new productive forces and building a modern industrial system."

Recently, reporters visited frontline enterprises, laboratories, and clinical institutions, and truly felt that 2026 is a critical year for China's brain-computer interface industry, transitioning from a sci-fi concept to practical implementation. The once far-fetched technological imagination is gradually opening the door to commercial application.

Serious Medical Scenarios Accelerate Implementation

InSunnyouIn front of the multi-dimensional upper limb medical rehabilitation training instrument, the reporter experienced an interesting "whack-a-mole" rehabilitation training session. Wearing a brain-computer interface device, with the arm fixed, and eyes on the moles on the screen, the training could be carried out with the help of...RobotThe armrest drives the arm to complete upper limb rehabilitation training.

"This EEG acquisition device captures specific EEG signals generated by visual focus, and the computer system quickly decodes them into specific motion direction commands," said He Yongzheng, Chairman, General Manager, and Technical Director of Sunnyou Medical. This training method utilizes the principle of neural plasticity in the brain, effectively rebuilding and strengthening the neural connections between the damaged brain and limb muscles, providing patients with a high-quality active rehabilitation experience.

In the 10,000-level sterile GMP workshop of Shenzhen Noel Medical, staff are encapsulating deep-brain electrodes. "High-quality, low-cost, and large-scale acquisition of high-resolution brain spatiotemporal data from within the skull, and attempting to train and apply large AI models for EEG, is an important application direction for brain-computer interface technology in the field of brain diseases. This enables precise diagnosis and closed-loop treatment of neurological disorders such as epilepsy, depression, schizophrenia, Parkinson's disease, and aphasia," Yang Huan, Chairman of Noel Medical (Shenzhen) Co., Ltd., told reporters.

This is a microcosm of the brain-computer interface industry accelerating from basic signal decoding to the implementation of cutting-edge technologies. The brainwave signals captured by humans are moving from laboratories to the front lines of clinical applications. "With the advancement of technology, we will progress from understanding brain diseases to discovering brain functions, and ultimately enter a new stage of bidirectional inspiration between the brain and brain-like systems," said Yang Huan.

In China, brain-computer interfaces are making progress in multiple areas and gradually entering the serious medical field: "Beijing Brain No.1" implanted a "Chinese chip" into the brains of paraplegic patients at Beijing Tiantan Hospital to restore hand motor functions; Neuracle Technology (Shanghai) Co., Ltd. ("Neuracle") received approval for its innovative product—an implantable brain-computer interface system for compensating hand motor functions—and it has been included in Shanghai's medical insurance coverage. Chen Jun, senior analyst of the pharmaceutical team at Open Source Securities, told reporters: "This is a milestone event in the brain-computer interface industry, marking that the product has completed technological and clinical validation and is officially moving towards the commercial application stage."

In the United States, brain-computer interfaces have entered mass production. Neuralink, under Musk, announced the launch of large-scale mass production of brain-machine devices in 2026 and introduced an automated implantation surgery process, significantly lowering the threshold for clinical applications.

"The United States has an obvious first-mover advantage in brain-computer interface technology, favoring high-risk, high-reward invasive technologies to take the lead, then spilling over into non-medical and human enhancement fields; whereas China adopts a differentiated approach by focusing on non-invasive and semi-invasive technologies, combined with rehabilitation and research scenarios to gain initial traction, while following up on invasive technologies in the medium to long term, precisely aligning with domestic clinical needs and the rhythm of industrial development."Southwest SecuritiesDu Xiangyang, chief analyst of the pharmaceutical industry, said.

Notably, despite the accelerating pace of brain-computer interface applications in serious medical scenarios and the active deployment by companies both in China and abroad, as a cutting-edge technology that directly interacts with the human brain, the issue of safety in its clinical application has become increasingly prominent, posing a critical challenge that must be overcome for the industry to achieve large-scale development.

Clinical safety remains the primary hurdle for the industry

Unlike other organs and tissues in the human body, the brain is soft in texture and fragile in structure. Invasive brain-computer interfaces require implanting electrodes into brain tissue, which inevitably causes damage. Therefore, breakthroughs in and assurance of safety are core challenges that must be addressed for the current development of the industry.

Reporters reviewed the evaluation report of Neuracle's implantable brain-computer interface system for hand motor function compensation and found that, although the response rate of participants' brain-computer interface-assisted ARAT grip was 100% at both 3 months and 6 months, it was still accompanied by 81.25% of clinical trial adverse events unrelated to the experimental device and 6.25% of serious adverse events. In other words, the long-term clinical safety of implantable brain-computer interfaces still needs breakthroughs.

Interviewed industry insiders generally believe that the large-scale development of brain-computer interfaces needs to break through the "impossible triangle" of signal quality, safe minimally invasive procedures, and long-term stability. Among these, hard barriers are concentrated at the hardware level, including insufficient long-term stability of electrode materials, lack of reproducibility in implantation processes, inadequate functionality and power consumption alignment of chips, and incomplete packaging and reliability verification. Soft barriers are reflected in limitations in data scale and model capabilities, an immature industrial ecosystem and distribution system, and the need for improved compliance standards and market trust systems.

"As the public attention on brain-computer interfaces grows, non-professionals tend to perceive it as an already mature industrial technology. Brain-computer interfaces cannot achieve the extreme efficacy seen in science fiction, nor are they pseudoscience as some might believe. The true value and application potential of this technology have yet to be fully unleashed," said Guo Yuzhu, Associate Professor at the School of Automation Science and Electrical Engineering, Beihang University.

Wei Erjia, a functional neurosurgeon at the People's Hospital of Jieyang City, Guangdong Province, who has worked in clinical practice for many years, told reporters that some self-media platforms have biased interpretations, which can easily mislead public understanding. "The decoding of brain electrical signals and the technology to convert these signals into real sensory feedback for the human body have not yet been fully breakthroughs. Although they hold great application potential, they are still in the early stages of commercialization."

Technical risks and cognitive biases are inevitable stages for many emerging technologies. Therefore, in July last year, the Ministry of Industry and Information Technology and six other departments issued the "Implementation Opinions on Promoting the Innovative Development of Brain-Computer Interface Industries," proposing to establish a full-chain technical standard system, strengthen ethical governance, enhance registration guidance for key products such as implantable brain-computer interface medical devices, control risks related to clinical surgeries and long-term implant biocompatibility, and promote the steady advancement of brain-computer interface industries under the premise of safety and controllability.

AI Helps Break Through the Decoding Dilemma of Brain-Computer Interfaces

In the age of artificial intelligence, computing is reshaping humanity's future, and brain-computer interfaces are no exception.

Shengyong Xu, a professor at the School of Electronics at Peking University, explained to reporters that when the brain works, it converts information such as auditory, visual, thought, and movement into neural activity coding. How to interpret and process EEG signals and provide effective feedback to the human brain is a key focus in the development of brain-computer interfaces.

Nuo'an Fund believes that the biggest challenge in the development of brain-computer interfaces lies in the highly noisy, non-stationary nature of brain signals and significant individual differences. AI technology can enhance raw signals into "actionable intent" and significantly reduce training costs.

"The large models currently are essentially still language models. A very important direction for the future is how to transform brainwave data, which originally could not be 'tokenized,' into data that the model can understand," said Wang Jian, academician of the Chinese Academy of Engineering, director of Zhejiang Lab, and founder of Alibaba Cloud. Guo Yuzhu also believes that once the barrier between human neural coding and machine algorithm coding is broken, brain-computer interfaces will experience the industry's 'GPT moment,' achieving a leap in capability.

Nowadays, some research institutions have begun to lay out research related to brain-computer interface signal decoding and intelligent computing. In August 2025, the National Key Laboratory of Brain-Computer Intelligence at Zhejiang University released the neuromorphic brain-like computer Darwin Monkey ("Wukong"). This brain-like computer is equipped with 960 self-developed Darwin Generation 3 neuromorphic computing chips, supporting over 2 billion spiking neurons and more than 100 billion synaptic connections, with a number of neurons approaching the scale of a macaque monkey's brain.

In December of the same year, Zhejiang Lab publicly showcased the research and development progress of the 021 scientific foundation model and a series of domain-specific scientific models. Unlike large language models, the 021 model is built on three cornerstones: interdisciplinary knowledge, cross-domain reasoning, and multilingual understanding. It possesses scientific reasoning capabilities, enabling in-depth analysis, deduction, and validation of various scientific problems… The empowerment of AI is driving brain-computer interfaces toward a "qualitative leap."

Entering 2026, positive signals for the brain-computer interface industry continue to emerge: it was written into the government work report for the first time, setting a clear development direction; multiple clinical achievements have been successively implemented, boosting market confidence; industry financing in the first quarter exceeded 3 billion yuan, surpassing the total scale of last year. Driven by policy, capital, and technology, brain-computer interfaces are accelerating from cutting-edge exploration to mass application, marking the beginning of a revolution in human-computer interaction.

Editor: Luo Hao