The stroke survivor could not close his hand. Not because the muscles were gone, but because the message from his brain no longer reached them. A cap of sensors on his scalp picked up the faint intent to move; a robot guided his fingers; the loop closed, and the brain began to relearn.
This is the unglamorous front line of brain–computer interfaces (脑机接口) in China — not implants, but wearable sensors, and not science fiction, but the hospital ward.
The scale of the problem
Stroke is a leading cause of death and disability among adults. Of those who survive, about 75% struggle with everyday activities and need long-term rehabilitation. Traditional rehab equipment helps, but motivation is low and outcomes are uneven. That is exactly where non-invasive BCI fits: it reads intention from the scalp and drives a device, turning passive exercise into active, brain-led training.
BrainCo's decade on the non-invasive path
BrainCo (强脑科技) is, by its own description, China's first BCI unicorn. Founded in 2015 at the Harvard Innovation Lab, it moved its headquarters to Hangzhou in 2018 under founder and CEO Han Bicheng (韩璧丞). Where Neuralink went inside the skull, BrainCo stayed outside it.
The company says the neural signal it chases is tiny — about ±50 microvolts, which it compares to picking up a mosquito's wing-flap from 50 kilometres away. Its pitch is a "super sensor" that reads nerve and brain signals through the skin and decodes them with algorithms. It claims signal-recognition accuracy above 95% of medical-grade EEG equipment.
BrainCo's products already sit in more than 400 hospitals and rehabilitation centres, according to a Xinhua-affiliated interview, and the company says it aims to help over 6,000 people with disabilities this year. Its bionic hand sells for roughly one-seventh to one-tenth the price of comparable Western prosthetics, and Zhejiang province has begun folding non-invasive BCI adaptation fees into basic medical insurance.
A robot that reads the intention to move
The most concrete clinical program is the "NCyborg Project," a collaboration between Tongji Hospital (affiliated with Huazhong University of Science and Technology) and Zhejiang BrainCo. Reported in the KeAi journal Brain Hemorrhages, the project builds a stroke-rehabilitation robot driven by BCI and brain-inspired control.
Its stated goals are specific: recognise at least eight distinct hand-movement intentions, at an accuracy of 90% or better, with a response time under 300 milliseconds. The team starts with the hand — often left weak after a stroke — and hopes, within five years, to put such a robot in front of millions of survivors.
Bournatech's hospital-grade rehab systems
Shanghai's Bournatech (博睿康), founded by researchers from Tsinghua University's neuroengineering group, builds EEG-based rehabilitation products — motor rehab through EEG-linked virtual reality, and motor-imagery VR training — used at institutions such as Huashan Hospital, Xuanwu Hospital and the China Rehabilitation Research Center. The company's strength is less the headline demo and more the unglamorous work of fitting EEG acquisition, decoding and clinical workflow into real wards.
The competitive shift here is telling: the win condition is moving from "can we decode a signal" to "can we embed it in a hospital's daily routine." Hardware needs regulatory clearance; algorithms must handle huge individual variation; hospitals need training and outcome measures. The firms that win will be the ones that integrate, not just impress.
Non-invasive technology has a structural advantage in scale. With no surgery, no implant and no infection risk, the same headset can be used on patient after patient. That is why specialists repeatedly argue the largest near-term volume in neurotechnology will come from safe, reusable, non-invasive systems rather than from the operating theatre — exactly the segment BrainCo and Bournatech occupy.
What readers can do now
- For families of stroke survivors, ask rehab departments whether they offer BCI-assisted training — it is increasingly available at major tertiary hospitals.
- Treat consumer "brain-training" headsets as wellness, not therapy; real rehab is supervised and registered.
- Watch Zhejiang's insurance coverage as a bellwether for whether non-invasive BCI rehab becomes broadly reimbursable.
Why this is a quiet turning point
Non-invasive BCI rehab rarely makes headlines because it is not a dramatic implant — it is a slow, repetitive training loop made a little more engaging. That is exactly why it matters. Stroke leaves millions of survivors needing months of rehab that health systems struggle to staff, and a device that turns that grind into something a patient will actually stick with can change real recovery odds. The China angle is scale: a domestic supply chain and, in Zhejiang, early insurance coverage that could make the therapy normal rather than exceptional. For readers, the signal to watch is not a single breakthrough but whether payers and hospitals adopt these tools as standard care.
Honest limitations
BrainCo's patient counts, hospital coverage and accuracy figures come from a Xinhua-affiliated interview and are company-disclosed; they are not independently audited. The NCyborg Project details are drawn from a peer-reviewed paper summary via EurekAlert and are research-stage goals, not proven clinical outcomes. Bournatech's (博睿康) hospital-client and product claims rely on industry reporting rather than primary regulatory filings, and should be read as indicative. Non-invasive BCI rehab improves engagement and may aid recovery, but it complements — and does not replace — standard physiotherapy.