For fourteen years, a man could not move his hands. A car accident had severed the connection between his brain and the rest of his body at the neck. Then, in late 2023, surgeons placed two coin-sized devices under his skull. Three months later, he was using his own brain activity to close a pneumatic glove and bring a cup of water to his mouth.
This is not science fiction from a lab brochure. It is NEO, a wireless, minimally invasive brain-computer interface (脑机接口) built by a team at Tsinghua University and tested in real hospitals with real patients.
What NEO actually is
NEO stands for Neural Electronic Opportunity. It was designed by professor Hong Bo's (洪波) team at the Tsinghua University School of Medicine. Unlike fully invasive systems that push electrodes deep into brain tissue, NEO takes a gentler path: its electrodes sit on top of the dura (硬膜外, the membrane just beneath the skull), recording signals from the brain's sensory and motor regions without cutting into the tissue itself.
The implant is about the size of two coins and is embedded in the skull. Crucially, it contains no battery. Instead, an external unit powers it and receives its signals through near-field wireless transmission across the scalp. That design choice removes the need for a battery that could leak or require replacement surgery — a meaningful safety advantage for something meant to stay in the body for life.
The clinical timeline
The first implant went in on October 24, 2023, at Xuanwu Hospital (宣武医院) in Beijing, with the procedure led by hospital president Zhao Guoguang (赵国光) and neurosurgery chief Shan Yongzhi. A second patient received the device on December 19, 2023, at Beijing Tiantan Hospital (天坛医院) under Jia Wang (贾旺). A third implantation followed in November 2024 at Huashan Hospital (华山医院) in Shanghai.
All three were people with cervical spinal cord injuries causing total limb paralysis. The first patient was discharged just ten days after surgery and continued rehabilitation at home.
What the patients could do
After roughly three months of home-based training with the system, the first patient could drive a pneumatic glove with brain signals to perform everyday actions such as drinking water independently. The team reported grasp-decoding accuracy above 90 percent, and measured improvements in both the patient's ASIA clinical score and sensory evoked-potential responses — objective signs that the injured nervous system was responding.
The second patient, a month after implantation, could not only operate the glove but also move a cursor freely on a screen. These are early, small-number results, but they are exactly the kind of functional gain that fully invasive systems have aimed for, achieved with far less surgical trauma.
Why "minimally invasive" matters
Tsinghua describes NEO as semi-invasive (半侵入式), sitting between non-invasive headsets and the fully implanted arrays used by groups such as BrainGate and Neuralink. The trade-off is deliberate: deeper, more invasive electrodes capture cleaner signals, but they carry greater surgical risk and tissue damage. NEO's epidural placement seeks most of the benefit with a fraction of the intrusion.
NEO did not appear from nowhere. Tsinghua's BCI group had built the groundwork for years: in 2015, the team of Gao Xiaorong and Gao Shangkai reached a steady-state visual evoked potential transfer rate of 319 bits per minute, and in 2021 Hong Bo's group reported a per-electrode equivalent rate of 20 bits per minute in pre-clinical work — both above the international state of the art at the time. Those baseline metrics paved the way for the human trials.
The work drew international attention. Nature included the Tsinghua wireless implant among its list of science events to watch in 2025, noting its placement on the sensory-motor cortex to help paralyzed patients recover hand movement. The team planned to expand to between 30 and 50 clinical implants during 2025.
Key takeaways
- NEO is a battery-free, wireless, minimally invasive BCI developed at Tsinghua and implanted in three Chinese hospitals from 2023 to 2024.
- Electrodes rest on the dura (硬膜外) rather than inside brain tissue, reducing surgical risk.
- Early patients regained useful hand control — drinking water, moving a cursor — with grasp decoding above 90 percent.
- Nature listed the project among notable 2025 science events; larger trials were planned for 2025.
- The system was built with Neuracle (博睿康), linking a university lab to a device maker.
What readers can do now
- Watch the trial count: The leap from three implants to the planned 30–50 cases will be the real test of safety and consistency — follow it.
- Compare routes, not just companies: Minimally invasive epidural placement is a distinct strategy from deep implanted arrays; understand which patient it fits.
- Look for peer-reviewed data: Ask whether later results appear in journals, not only in hospital news releases.
Honest limitations
This account is based on Tsinghua University's official announcements, Chinese state media coverage, and Nature's 2025 events listing; it is not drawn from an independently published clinical paper with full statistical detail. The three cases described are early feasibility implants, not a controlled trial, so we cannot state how representative the gains are across a broader population. Performance figures such as the >90 percent grasp-decoding rate come from the university's own reporting and were not re-audited here against raw data. Long-term durability of the electrode-tissue interface and whether benefits persist beyond the training period remain unverified in public sources.