Therapy rooms are usually quiet places of repetition: a nurse guides a limp arm, a machine flexes a frozen knee, a patient tries and tries again. In a growing number of Chinese hospitals, a new instruction has entered that routine — not "lift your arm," but "imagine lifting your arm."
That small change is the whole point of a wave of non-invasive brain–computer interface (脑机接口) rehabilitation equipment now reaching hospital wards. One of the companies pushing it hardest is Xiangyu Medical (翔宇医疗), a Henan-based rehabilitation-device manufacturer that has spent the last decade building what it calls its Sun-BCI Lab.
From passive motion to active intention
Traditional rehab is largely passive. A device moves the patient; the patient's brain is a spectator. Brain–computer interface rehab flips the order. Electrodes on the scalp capture the patient's motor intent, a decoder translates it, and the rehab machine executes the motion the patient was trying to make. The company says the loop is meant to re-engage the brain's own repair mechanisms rather than simply exercising unused muscle.
Xiangyu's approach is deliberately multi-paradigm. It mixes motor imagery (运动想象), where the patient imagines a movement; motor attempt, where they try; and steady-state visual evoked potential, where flickering visuals let a locked-in patient spell words or control a device by gaze. The throughline is Hebbian-style learning: neurons that fire together, wired by feedback, grow new connections.
What is actually deployed
This is not a lab curiosity. At the 93rd China International Medical Equipment Fair in April 2026, Xiangyu showed close to a hundred rehabilitation products and a dedicated brain–computer interface line built on a decade of in-house brain-science research. The company told China Securities Journal that, as of May 2026, it held two registration certificates for electroencephalography acquisition devices and had submitted more than ten brain–computer interface rehab devices for certification, with several expected to clear review within the year.
The product matrix is broad: upper- and lower-limb active-passive rehab trainers, hand-function devices, swallowing-stimulation units for dysphagia (吞咽障碍), and cognitive and speech modules. The intended users are people with stroke, traumatic brain injury, spinal-cord injury, cognitive decline, or speech loss. In one often-cited use, a patient who cannot speak uses a visual-evoked-paradigm system to type or to operate home devices — turning a wheelchair, placing a call — by intention alone.
Scale is the part that surprises visitors. According to the company's January 2026 investor disclosures, reported by Securities Times, its non-invasive brain–computer interface products had entered more than 600 tertiary hospitals, with a 2026 target of over 1,000. Those numbers are company-reported and should be read as commercial footprint rather than clinical proof.
The hospital partnerships are the real signal
Beyond scattered placements, Xiangyu is building structured clinical centers. On September 6, 2026, Zhejiang Provincial Hospital of Traditional Chinese Medicine (浙江省中医院) opened a "Lingshu" brain–computer interface rehabilitation diagnosis-and-treatment center with Xiangyu as the core partner. The center put into clinical use a brain-interactive rehab ward-care system, a brain-controlled wheelchair, non-invasive EEG-driven upper- and lower-limb active-passive trainers, and a brain-interactive lower-limb exoskeleton walking trainer — covering movement, hand function, and swallowing.
A few weeks earlier, Wenzhou Central Hospital launched what was described as southern Zhejiang's first clinical brain–computer interface ward, where patients "just think" to drive rehab equipment without surgery or implants. In both cases the appeal is exactly that word: no opening of the skull, no implantation, just a cap and a machine.
Why non-invasive won at the bedside first
Invasive brain–computer interfaces grab headlines because implanted electrodes read cleaner signals. But for a hospital treating hundreds of stroke patients a week, an electrode you can clip on and take off wins on safety, cost, and throughput. Xiangyu's pitch is integration: it already makes the rehab beds, bikes, and robots, so bolting a decoder onto them is a product-line extension rather than a moonshot. The company says it designs the chip, electrode, algorithm, and整机 (whole device) itself, which it argues avoids the compatibility problems of cobbling a third-party brain sensor onto someone else's rehab machine.
Honest limitations
Almost every adoption and certification figure above is company-disclosed, not independently audited, and "placed in a hospital" is not the same as "proven to restore function." Non-invasive scalp signals are weak and person-dependent, so consistency across patients remains the central unsolved problem. Several of these devices are still in the certification pipeline rather than on the market, and the clinical evidence for durable recovery is thinner than the deployment headlines suggest. Brain–computer interface rehab should be read as a promising adjunct to conventional therapy, not a replacement.
What readers can do now
- If a relative is in neurological rehab, ask whether the facility offers brain–computer interface–assisted training and which specific certified device it uses.
- Check a device's registration status on the National Medical Products Administration database before assuming a "brain-controlled" rehab product is a cleared medical device.
- Watch for peer-reviewed or hospital-published outcomes on stroke and dysphagia cohorts, not just vendor demonstrations, when judging whether to pursue this route.