Neuromodulation is now used across a range of disorders, from sacral nerve stimulation for overactive bladder to non-invasive stimulation for depression, and sensory stimulation approaches explored in Alzheimer’s. But while neuromodulation use in psychiatry is growing, it has been far from prominent in autism spectrum disorder. A Chinese startup is trying to change that, combining MRI-guided brain mapping with a form of TMS to reduce core symptoms associated with ASD.

That ambition is ethically and clinically sensitive. TMS has been explored in ASD for more than a decade, including by international providers such as BrainsWay, but few startups have explicitly committed to treating ASD’s core symptoms. The absence of approved drugs for those symptoms is not simply a gap waiting to be filled. It reflects a changing view of autism itself. ASD is considered a spectrum condition with highly individualized manifestations, often closely tied to identity, bringing into question: what will they treat?

Autism as a Spectrum Disorder

Autism spectrum disorder is a relatively new diagnosis. Earlier systems separated autism-related presentations into categories such as autistic disorder and Asperger’s disorder. In 2013, the DSM-5 collapsed those labels into autism spectrum disorder, reflecting diagnostic inconsistency and wide variation in how autism presents. ASD is now understood as a heterogeneous neurodevelopmental condition, spanning different levels of language, cognition, sensory processing, independence, and support needs.

Core diagnostic features are usually grouped into two domains: differences in social communication and interaction, and restricted or repetitive patterns of behavior, interests, or sensory experience. But those features often sit alongside a broad range of associated challenges, including anxiety, sleep problems, ADHD-like symptoms, epilepsy, language delay, self-injury, and intellectual disability. Most current care is aimed at support, development, education, communication, daily functioning, or those co-occurring symptoms.

Medicine use in autism largely follows that pattern. Risperidone and aripiprazole are used for irritability associated with autism, but do not treat social communication differences or restricted and repetitive behaviors. Drug programs aimed at core symptoms, including oxytocin and balovaptan, have struggled to show convincing benefits. That absence of approved drugs for core symptoms reflects inherent biological heterogeneity, difficult endpoints, and an unresolved ethical question over what improvement in ASD means.

Neural Galaxy’s Treatment Approach

In China, a startup called Neural Galaxy is exploring whether neurotechnology can treat core ASD symptoms. Neural Galaxy, also known as Galaxy Brain Scientific or 优脑银河, is a Chinese neurotech company building around personalized brain functional mapping and precision neuromodulation. The startup was founded by MIT and Harvard scientists and, after receiving over $90 million in funding, now operates a team of more than 50 researchers and over 100 product hires. Its systems are used for a range of psychiatric and neurodegenerative disorders.

In a recently published Nature Molecular Psychiatry paper, a Chinese research group tested Neural Galaxy’s neuronavigated TMS system in autistic participants aged 6 to 30. The study used accelerated intermittent theta-burst stimulation (iTBS), a patterned form of TMS that delivers repeated bursts of magnetic pulses over a compressed treatment schedule, to target individualized fronto-parietal control network nodes.

The trial enrolled 59 participants in a young cohort, where the mean age sat around 10 years old. 88% of participants were male, and all had cognitive or adaptive delays. Participants were assigned to active or sham stimulation, while response was measured through the ADOS-2 Social Affect domain, a clinician-observed measure of social communication and reciprocal interaction. At week 12 of the research, the active group showed a higher response rate than placebo, at 55% versus 29%.

The protocol was intensive. Participants received three daily iTBS sessions, five days per week, over 12 weeks, alongside speech therapy and behavioral training. Treatment targets were personalized, selected using structural and functional MRI. 

Neural Galaxy is not the first company-linked effort to bring TMS into autism. BrainsWay conducted research into deep TMS for ASD back to 2014, when a small double-blind trial tested bilateral dorsomedial prefrontal stimulation in adults with high-functioning autism or Asperger’s syndrome. Since then, BrainsWay has listed autism among its CE-marked treatment areas for adults outside the US, although autism remains outside of FDA-cleared indications. In the US, Yale is currently researching the use of TMS in ASD.

More broadly, non-invasive brain stimulation has been tested in ASD for more than a decade, across prefrontal, motor, and social-brain targets. Yet Neural Galaxy’s work stands out by combining individualized network targeting, company-linked hardware, an intensive treatment schedule, and a trial endpoint explicitly framed around core ASD symptoms in a young, high-support cohort.

An Identity or Manageable Symptoms?

Treating core ASD symptoms produces a range of ethical and clinical questions. Autism interventions are not ethically sensitive because support is wrong, or because high-support autistic people do not need care. Many would benefit greatly from more support. But where does support end, and where does normalization begin? When an intervention targets distress, self-injury, communication barriers, sleep disruption, sensory overload, or severe functional impairment, the clinical rationale can be clear. When it targets core ASD traits, interwoven with the person’s identity and personality, the question becomes less settled.

The pediatric setting of the Chinese research makes those questions harder. Neural Galaxy’s trial included participants as young as six, in a cohort with cognitive or adaptive delays, and involved an intensive 12-week stimulation protocol. Some patients had to be sedated during the MRI phase of the study. Parent or guardian consent is required in that context, but hardly sufficient on its own. How should assent be assessed when a child has limited verbal communication? What counts as dissent? How should sensory burden, distress, or procedural tolerance be weighed against potential clinical benefit?

The same uncertainty applies to outcomes. If a clinician-rated social-affect score improves, does that reflect greater autonomy, lower distress, better communication, or simply behavior that appears more neurotypical in a structured assessment? If caregivers report improvement but the person later experiences more masking or exhaustion, how should that be interpreted? And if treatment is aimed at core ASD symptoms rather than a bounded problem such as self-injury or anxiety, how can one determine when to stop treatment? 

The results coming out of China reflect renewed interest in neurotech’s potential in improving the lives of ASD populations, but also surface a range of tough ethical questions that are still very hard to answer.