The consultation is short, and it repeats across the country. A child does not turn toward sound. Imaging shows an inner ear that never formed, or a nerve that never arrived, and the standard device is offered and then withdrawn, because it needs something intact to stimulate.
For those families the answer used to be a school for the deaf, or a special-zone import.
On 20 October 2026 China's drug regulator approved a domestically made device for exactly this group, and the terms of the approval explain more than the headline does.
Key takeaways
- NMPA approved Nurotron's auditory brainstem implant on 19 October 2026 and announced it the next day, under registration number 国械注准20263122152 (NMPA).
- It is cleared only for prelingual children aged 2 to under 6 with severe inner-ear malformation or an absent cochlear nerve, who cannot receive a cochlear implant (NMPA).
- The trial reported a 96.67% total effective rate 12 months after switch-on; the approval notice does not say how many children were implanted.
- The array carries 16 electrodes on the brainstem's cochlear nucleus and uses a four-tone coding strategy built for Mandarin (Shanghai Ninth People's Hospital).
- Nurotron says it will price below imported devices, which the Yuhang district government put at more than RMB 400,000 (≈ US$56,300 / HK$440,000).
What was approved, and why the route is different
The device, 诺尔康-WH-01A, is an active Class III implant: a stimulator, a receiving coil, an intracranial electrode array, an extracranial plate electrode, a ring electrode and an internal magnet. It works with the speech processor Nurotron already sells for cochlear implants. The external unit encodes sound; the implant turns that stream into current pulses at the cochlear nucleus.
The indication is narrow. It covers non-tumour, non-syndromic prelingual deafness where the inner ear is severely malformed — Michel deformity, common cavity, cochlear aplasia or severe hypoplasia classified as CH-1 — or where the cochlear nerve is absent and a cochlear implant is therefore impossible. Patients must be two or older and under six.
NMPA calls it the first auditory brainstem implant made in China, and the first Chinese product to stimulate and record inside the brainstem. The application entered the regulator's innovation-device review channel at the end of 2024.
A cochlear implant does not restore an ear; it drives the auditory nerve directly, which means the nerve has to exist and the inner ear has to be navigable.
In a subset of children neither holds. Science and Technology Daily, setting out the clinical background on 29 October 2026, put the share at 8% of children with congenital severe-to-profound sensorineural hearing loss who cannot receive an implant because of severe inner-ear malformation or bilateral auditory nerve damage.
The brainstem implant moves the stimulation point inward, to the surface of the cochlear nucleus, bypassing both the cochlea and the nerve. Two design choices follow from that target:
- Sixteen electrodes, shaped to the anatomy of the nucleus rather than to the spiral of the cochlea.
- A four-tone coding strategy, because in Mandarin the pitch contour of a syllable carries meaning; a stimulation pattern that discards pitch delivers sound without words.
The array can also record neural responses, letting clinicians check the tissue, not only the patient's report. The programme took seven years: team in 2019, patent licence in July 2021, trial from 2022, surgery and follow-up done in 2024.
What 96.67% does and does not mean
The figure is a total effective rate, measured 12 months after the device was switched on, in a trial run at Shanghai Ninth People's Hospital with Sanbo Brain Hospital in Beijing as a sub-centre.
It is neither a speech-recognition score nor a population estimate. Neither the approval notice nor the hospital releases state how many children were implanted, so the denominator behind 96.67% is not public. A rate from a small series of children with a rare anatomical condition is evidence that the device can work, not a forecast for the next patient.
Twelve months is also short. The implant sits against the brainstem of a child who will grow, and no public document addresses how the interface behaves over five or ten years.
The bottleneck is the operating theatre
Brainstem implantation is difficult surgery, and the developer's release notes that only a handful of centres worldwide perform it, which the developer says has long limited the technology's introduction into China.
Capacity, not hardware, is the constraint. Three must line up:
- Detection. The malformation or missing nerve must be found inside a two-to-six-year window, which depends on newborn screening followed by imaging.
- Referral. A family has to reach a centre with a neurosurgical team willing to operate near the brainstem.
- Money. Imported brainstem implants cost more than RMB 400,000 (≈ US$56,300), according to the Yuhang district government, home to Nurotron. The company says its version will be cheaper. Neither that release nor the approval notice mentions reimbursement.
Why it is filed under brain-computer interface
Chinese policy treats hearing restoration as a brain-computer interface (脑机接口) problem, because the electrode sits inside the skull. The technology was named a future industry in the 15th Five-Year Plan and appeared in the 2026 Government Work Report for the first time (Inside BCI, 30 September 2026).
The label is accurate, but the economics invert the motor implants that dominate the field: a spinal-cord interface serves adults who have already lived with speech, while a hearing interface serves children who have not yet acquired language.
Honest limitations
The evidence is one registration trial reported by the hospital and the company; no peer-reviewed paper with patient-level data exists. The number of implanted children, their ages and the definition of "effective" are not in the public record. Durability in a growing skull, the learning curve at new centres, reimbursement and price are unknown. The 8% share is a clinical estimate quoted in a newspaper, not a registry figure.
What readers can do now
- Pull the NMPA record by registration number and compare its indication wording with the company's marketing: the age band and anatomical exclusions are where off-label use will start.
- For adoption modelling, wait for a provincial pricing or insurance decision: an approval without a payment route produces pilots, not volume.
- In a neurotech tracker, separate sensory-restoration devices from motor ones. They differ in patient count, surgical capacity and regulatory history.