China debuts world’s first two-way brain-computer interface with 100x efficiency

[Beijingwalker]

China debuts world’s first two-way brain-computer interface with 100x efficiency​

The dual-loop system allows for four-degree-of-freedom control using only brain signals, improving accuracy by 20 percent.​

Updated: Feb 19, 2025 08:12 AM EST

Neurotechnology, implantable brain machine, chip inserted into brain. (Representative image)

iStock
Chinese scientists have developed the world’s first two-way adaptive brain-computer interface (BCI), significantly enhancing efficiency and paving the way for practical applications. According to the new study, the system boosts efficiency 100-fold compared to traditional BCIs.

Researchers from Tianjin University and Tsinghua University claim their innovation could soon be integrated into wearable and portable devices for medical and consumer use.

This system allows both the brain and machine to learn from each other, unlike conventional BCIs that only decode brain signals. This interaction ensures stable performance over time, a key step toward making BCIs more reliable in daily applications.

“Our work is the first to introduce the concept of brain-computer co-evolution and successfully demonstrate its feasibility, marking an initial step towards mutual adaptation between biological and machine intelligence,” said Xu Minpeng, a co-author of the study from Tianjin University.

First brain-computer interface to overcome one-way limitations​

Since the 1970s, BCIs have allowed users to control machines with their thoughts by translating brain signals into commands. Initial research focused on helping people with disabilities, but today’s applications include gaming, hands-free drone control, and other interactive technologies.

Despite these advancements, traditional BCIs have a critical limitation: they function in only one direction. This means the brain does not receive feedback that could help it adjust and refine control. Over time, this lack of adaptation leads to declining performance.

“A major challenge in advancing BCI technology is achieving mutual learning between the brain and the machine,” Xu said.

The researchers discovered that changes in brain signals were not just random fluctuations caused by emotions or fatigue. Instead, these variations were influenced by how the brain interacts with a BCI.

Using this insight, they developed a dual-loop framework using a memristor chip—an energy-efficient hardware component that mimics neural networks—to create a more natural interaction between brain and machine.

The system consists of two key loops: a machine learning loop that continuously updates the decoder to adapt to the brain’s signal variations and a brain learning loop that helps the user refine control through real-time feedback.

BCI with higher efficiency and expanded capabilities​

The study demonstrated that the dual-loop system increased efficiency by over 100 times and reduced energy consumption by 1,000 times compared to conventional BCIs.

“Compared to traditional digital BCIs, our dual-loop system increased efficiency by over 100 times while reducing energy consumption by 1,000 times,” Xu said.

This technological leap also allows users to perform more complex tasks. Traditional BCIs typically offer two degrees of freedom, such as moving a drone up and down or left and right. However, the new system enables four degrees of freedom, adding forward-backward motion and rotation—all controlled solely by brain signals.

Researchers conducted six-hour tests with 10 participants and found that the adaptive system improved accuracy by approximately 20 percent compared to non-adaptive BCIs. The study demonstrated long-term stability and a learning curve for users, reinforcing the system’s practical potential.

“Our research provided a strong theoretical foundation and technical support for the development of practical BCI systems and opened up new directions for the advancement of brain-machine integrated intelligence,” Xu added.

 

[Hussain]

what could be the possible negative uses of this tech?

 

[Zsari]

what could be the possible negative uses of this tech?

Cyborg.

 

[MH.Yang]

what could be the possible negative uses of this tech?

China's Northwestern Polytechnical University has unveiled technology to control drones using brain machines. It is about to be used on a large scale.

 

[Hamartia Antidote]

what could be the possible negative uses of this tech?

Re-education camps.

 

[冖_冖]

China's Brain-Computer Interface Sector Enters First Year of Clinical Application: World's First Approval Leads to 12 Commercial Implants, Shanghai's Huibao Reimburses Up to 30% — BigGo Finance

2026 is seen by the industry as the first year of clinical application for brain-computer interfaces (BCIs) in China. In March, BCI was written into…

finance.biggo.com

China's Brain-Computer Interface Sector Enters First Year of Clinical Application: World's First Approval Leads to 12 Commercial Implants​

In March this year, brain-computer interface (BCI) was written into China's Government Work Report for the first time, listed as one of the future industries to be prioritized during the 15th Five-Year Plan period. The same month, China's National Medical Products Administration approved the world's first implantable BCI medical device for market launch. From the "first approval" to the "first prescription," China's BCI industry completed the leap from laboratory to hospital ward in 2026. Hong Bo, chief scientist at Lingang Laboratory, said at a media briefing ahead of the BCI Conference at the 2026 Pujiang Innovation Forum that this year is a milestone year for China's BCI development, one that can be summed up with two keywords: "first year of clinical application" and "first year of the data ecosystem."

The product that obtained the world's first Class III medical device registration certificate is NEO, an implantable BCI hand motor function compensation system independently developed by Shanghai NeuroXess Technology. It uses minimally invasive epidural implantation and wireless power and communication technology, and is indicated for patients with quadriplegia caused by cervical spinal cord injury. In clinical follow-up of 32 quadriplegic patients, 81.3% of subjects showed functional recovery 12 months after surgery, with unassisted grasping function reaching clinically significant levels, and cumulative total time of safe home use by patients exceeding 19,000 days.

Commercialization has advanced just as quickly. On July 13, only four months after NEO was approved for market launch, the first prescription was written at Fudan University Huashan Hospital, and the world's first commercial implant procedure was completed the same day. To date, the system has completed 12 commercial implants and has been included in Shanghai Huibao, Shanghai's inclusive supplementary medical insurance plan, under which related treatment costs can be reimbursed up to 30%. Work to connect with medical insurance and inclusive insurance programs in provinces across China is progressing steadily.

China's BCI industry has formed a landscape in which three technical routes—epidural, subdural and fully invasive—advance in parallel, with a number of Chinese local companies achieving key progress.



In terms of implant scale, China now has about 120 BCI patients who have had devices implanted for more than a month with the devices still in their bodies. Neurosurgery and rehabilitation resources at hospitals are also expanding in tandem. Beijing Tiantan Hospital, Fudan University Huashan Hospital, Tianjin Huanhu Hospital and Hubei Union Hospital have successively built BCI clinical wards, and Shenzhen and other places are also setting up clinical research platforms. Nationwide, 40 to 50 hospitals have opened BCI wards, of which about 20 have the capacity to perform invasive implant surgery.