NeuroAI NEUROAINEUROAI.SITE
China AI, reported clearly
1,300+ verified stories on models, robotics, BCI and the business of AI.
Browse stories
ESC

An implant the size of two coins, no battery: Tsinghua's Hong Bo is rebuilding the link between brain and hand

Tsinghua professor Hong Bo (洪波) leads NEO, a wireless, battery-free brain-computer interface that has helped paralyzed patients move a hand again.

2026-11-14 · 976 words · NeuroAI
An implant the size of two coins, no battery: Tsinghua's Hong Bo is rebuilding the link between brain and hand

A man who had not moved his own hand in fourteen years thought about closing his fingers, and a glove on his hand closed them. The signal had traveled from his brain, through a device roughly the size of two coins resting on the surface of his skull, to a machine that read his intent. No wire trailed from his head. No battery had been buried in his body. The brain itself had not been opened.

Key takeaways

  • Hong Bo (洪波) is a tenured professor at Tsinghua University's School of Medicine, Department of Biomedical Engineering, and directs the university's Wei Xian College (为先书院); he earned his PhD at Tsinghua in 2001 and was a visiting scholar at Johns Hopkins University in 2004 and at MIT's McGovern Institute in 2016.
  • He leads NEO (Neural Electronic Opportunity), a wireless minimally-invasive brain-computer interface (脑机接口): electrodes sit epidural — on the dura, outside the brain tissue — and the implant carries no battery, powered by near-field wireless coupling through the skull.
  • The first human implant took place in October 2023 at Beijing Xuanwu Hospital (宣武医院); a second followed in December 2023 at Beijing Tiantan Hospital (天坛医院) and a third in November 2024 at Shanghai Huashan Hospital (华山医院).
  • In the first patient, after about three months of training, grasp-decoding accuracy exceeded 90 percent, letting him drive an air glove to drink water on his own.
  • Nature's website listed NEO among its "events to watch in 2025," published on 17 December 2024.
  • In June 2025 the team began a multi-centre registered clinical trial across 12 hospitals in China, targeting 30 to 50 implants that year; the device was developed with Bourn (博睿康).

The problem he set out to solve

Most implantable brain-computer interfaces face a harsh trade. The highest-bandwidth versions, like those from Neuralink and the older BrainGate program, place electrodes inside the brain tissue to read clean signals — but that means cutting into the organ and risking the slow loss of signal as the body reacts. Hong's team chose a different seam: lay the electrodes on the dura, the protective layer just under the skull, so nothing pierces the brain. The cost is some signal richness; the gain is that the body does not treat the implant as an intruder.

The second design choice is the one patients feel. Instead of a battery that must be surgically replaced every few years, NEO draws power through the skull by near-field wireless coupling from an external unit. The internal device, roughly the size of two coins, can in principle stay for the life of the patient.

Three patients, and a measure of recovery

The first implant went into a man who had been quadriplegic for fourteen years after a car accident, at Beijing Xuanwu Hospital in October 2023. Three months of home training with the system let him drive an air glove to grasp and drink through brain activity alone; Tsinghua reported the grasp-decoding accuracy above 90 percent and noted improvements in the patient's clinical ASIA spinal-injury score and sensory-evoked responses. The second patient, implanted at Tiantan Hospital in December 2023, could move a cursor and a wheelchair by thought; the third, at Huashan Hospital in November 2024, showed similar gains.

Follow-up observations reported by the team and by reputable Chinese outlets suggest something beyond control: signs that the training itself may have nudged neural recovery, with one patient able to move his hand without the implant after two months. That claim is preliminary and not yet the established read of the data, but it is why clinicians are paying attention.

From one implant to a trial

Nature's listing in December 2024 put NEO alongside a short list of science events worth watching in 2025, a rare mention for a Chinese academic BCI effort. By June 2025 the work had moved from a handful of cases to a multi-centre registered trial spanning 12 hospitals, with a target of 30 to 50 implants that year. The device hardware was built with Bourn (博睿康), a Chinese neuro-technology firm, under a collaboration Tsinghua has described as covering both the system and its core algorithms.

According to Li An, chief scientist at BrainNet (脑机网), the epidural, battery-free design is worth the trade because it exchanges a little signal richness for a safety profile that lets a device remain implanted for years rather than months, which is what chronic patients actually need. Li An's comments reflect his own views and do not represent NeuroAI's editorial position or an endorsement by BrainNet (脑机网).

Honest limitations

The clinical details here come from Tsinghua University's own news and faculty pages, the China Science and Technology Association's achievement platform, and reputable Chinese outlets such as The Paper; the hard numbers — three implants, above 90 percent grasp decoding, 12-hospital trial — are Tsinghua's published figures, not independently audited clinical results, and small samples should not be read as proof of wide efficacy. I deliberately left out several colorful specifics that appeared only in Chinese self-media: a precise count of internal parts, a fixed number of safety tests, and a promised next-generation channel count. Those could not be confirmed against an authoritative primary source, so they were dropped. The neural-recovery observation is reported but early, and I have presented it as a sign worth watching rather than a confirmed outcome. NEO is still an investigational device, not an approved product, and its path to regulatory clearance is unfinished.

What readers can do now

  • Follow the 12-hospital trial's published outcomes rather than single dramatic videos; sample size and peer-reviewed data are the real signal.
  • If you or a family member faces paralysis, treat NEO as investigational and ask clinicians about eligibility for registered trials, not off-the-shelf availability.
  • Watch the epidural, battery-free approach as a design bet: if it scales safely, it changes the conversation from "how smart" to "how long can it stay."

Related coverage

Cite this article

NeuroAI (脑机网), "An implant the size of two coins, no battery: Tsinghua's Hong Bo is rebuilding the link between brain and hand," 14 November 2026, https://neuroai.site/a/na-people-hong-bo-neo-bci

Machine-readable: BibTeX · RIS · JSON · Markdown

More in “Builders” → · Back to home · Markdown version

China AI, reported clearly
1,300+ verified stories on models, robotics, BCI and the business of AI.
Browse stories