Consumer diagnostics · Brain imaging
BrainKey
A San Francisco company that turns an ordinary brain MRI into a 3D dashboard and a single number it calls BrainAge. Seven years after Y Combinator it holds no FDA clearance, and its own terms describe the service as research, informational and educational only.
Two models, one head, two answers
In 2024 the BBC reporter Lara Lewington had her brain measured twice for a documentary about how brains age. At the University of Southern California, Prof Andrei Irimia ran her MRI through models built from 15,000 brains and told her that her brain was eight months older than her birth certificate, within a two year error margin.
Then she took the same head to BrainKey. Its estimate, she wrote, “knocked a year off my brain’s biological age.” She went home with a life sized gold 3D print of it.
Two models, one brain, a gap of roughly two years, and no way to say which was closer, because there is no reference standard for a brain’s age. That gap is the story of this company and of the metric it sells.
The problem it picked
Brain care is reactive. Most people learn something is wrong only once symptoms arrive, by which point the treatable window for several conditions has closed. Phillips has made that argument in near identical words since 2019, and he is right that the raw material already exists and goes unused: brain MRIs are read for one clinical question, reported in prose and handed back on a disc. “People want to see their imaging results, but imaging data often doesn’t get to them in a way that’s understandable,” he told AuntMinnie in 2020.
The counter argument is as old. The American College of Radiology says it “does not believe there is sufficient evidence to justify recommending total body screening for patients with no clinical symptoms, risk factors or a family history suggesting underlying disease or serious injury,” warning that such scans surface findings that trigger follow up testing without improving health. BrainKey images one organ, but that problem travels.
A deep learning class at Stanford
Phillips spent more than a decade in research at UCLA, Stanford, Rome and Toulouse, and by the company’s count published more than 50 peer reviewed papers. In 2017 he helped launch the Stanford AI Club and began applying machine learning to neuroimaging.
He met Ben Kotopka, a PhD student in computation and biology, in a deep learning for genomics class. A class project became a company, and the two carried it into Y Combinator’s Winter 2019 batch, Kotopka as co-founder and chief technology officer while finishing his dissertation. By 2025 the chief technology officer named in company materials was Nathan Strong.
The founding motive has been told two ways. In 2020 Phillips said: “My grandma died with dementia and a close friend had a brain tumor that was a nightmare for him and his family.” By 2025 the website and a Microsoft case study centred on his mother, whose decline went undiagnosed until, that case study says, BrainKey’s own technology helped surface normal pressure hydrocephalus, a treatable condition. Both can be true. Only the second carries the current marketing.
From a $390 scan to imaging networks
The first commercial shape was direct to consumer. In 2020 a BrainKey scan cost $390 and bought a 15 minute MRI at one of eight California partner centres, a report from a board certified radiologist at Focus Medical Imaging, and a subscription dashboard.
That model has inverted. BrainKey now sells through clinics and imaging networks, saying in 2025 that it works with longevity clinics and networks “including SimonMed,” while a Microsoft case study cited a recent deal with one of the largest MRI providers in the United States without naming it. At one US concierge practice in 2026 the assessment retails at $950: a $500 ProScan MRI plus a $450 BrainKey analysis. The same 2025 material says the company is extending to whole body imaging.
What is proven, and what is still claimed
| Evidence | What the record shows | Source type |
|---|---|---|
| FDA status | openFDA 510(k), PMA and registration searches run Sep 23, 2026 for both corporate spellings return no matches on any database. | Public record |
| What a cleared competitor looks like | Cortechs Labs holds three NeuroQuant clearances (K061855, 2006; K170981, 2017; K241098, 2024). Icometrix holds six for icobrain, K161148 (2016) through icobrain aria K240712 (2024). Quantib holds three, K153351 (2016) through Brain WMH K251527 (2025). | Public record |
| What the company says the product is | Terms of service: “BrainKey Services are for research, informational, and educational use only. We do not provide medical advice.” | Company-stated |
| Where BrainAge comes from | “The Brain Age model we use was developed by Dr. James Cole, a London-based neuroimaging researcher and Brain Key advisor, and his colleagues.” Scans align to MNI-152 space, reduce by principal components analysis to 485 values, then feed a Gaussian process regression trained on 2,001 scans. | Company-stated |
| Peer reviewed evidence for the platform | One study: Lindsay et al., Frontiers in Human Neuroscience, Aug 10, 2026, 51 elite athletes, scans segmented with BrainKey across 18 regions. Three of six authors are BrainKey employees, Phillips holds an ownership interest, and a fourth consults and holds equity. | Public record |
| The rest of the publication list | The Nature and Nature Neuroscience papers shown under research on the company site carry a BrainKey affiliation through Kevin Aquino, declared there as “a scientific advisor and shareholder in BrainKey Inc., a medical image analysis software company.” They are basic neuroscience, not evaluations of the product. | Public record |
| Independent evidence on brain age | eBioMedicine, Jan 2026, compared six brain age packages on ADNI. Agreement between packages ranged from 0.37 to 0.69, and “simple normalised grey matter volume measurements performed comparably to the more complex brain age estimates.” | Peer reviewed |
| A claim against its own caveat | The 2020 BrainAge explainer states BrainKey “is currently the only service of any kind which provides a scientifically validated Brain Age estimate,” two paragraphs before conceding that “Brain aging science is in its infancy, and it’s not yet known how best to apply it in the clinic.” | Claim against caveat |
| Trials and federal grants | No registered trial, NIH award or NSF award as of Sep 23, 2026. | Not found |
| Named customers | SimonMed named by the company in 2025. The large MRI provider in the Microsoft case study is unnamed. No customer has confirmed publicly. | Company-stated |
Read plainly: BrainKey is a wellness and engagement product with a research framing, sold into a market where the adjacent products are regulated devices. That is a deliberate legal position, stated in its own terms. The harder question is the number. Brain age is a real research construct with real published associations, including the Cole mortality work BrainKey cites, but the 2026 comparison found the best models add little over a plain grey matter volume, and the BBC episode shows what that means for one person.
What to watch
- Whether BrainKey files a 510(k), the pathway all three named competitors took.
- Named imaging network customers. SimonMed is company-asserted and not corroborated from the other side.
- A second peer reviewed study by authors holding no equity.
- The whole body expansion announced in 2025, which enters territory the ACR has cautioned against.
- Any funding filing at all. No Form D exists under either corporate name.
In their words
“People want to see their imaging results, but imaging data often doesn’t get to them in a way that’s understandable.”
Owen Phillips, CEO, to AuntMinnie, 2020 · Independent
“My grandma died with dementia and a close friend had a brain tumor that was a nightmare for him and his family.”
Owen Phillips, interview with Jess Li, 2020 · Interview
“our products are meant to promote wellness, but we can’t claim to be actually diagnosing anything!”
BrainKey BrainAge explainer, February 2020 · Company
“As far as we know, Brain Key is currently the only service of any kind which provides a scientifically validated Brain Age estimate.”
BrainKey BrainAge explainer, February 2020 · Company
“It’s becoming much more accessible for people to get an MRI scan, and the images coming off them are getting even better and better.”
Owen Phillips to Lara Lewington, BBC News, 2024 · Independent
“The most important thing to understand is that these are tools, not diagnoses.”
Jacob Brandon, MD, Priority Physicians, 2026 · Clinician
“This suggests that the current brain age estimates have limited clinical utility as a biomarker for biological brain ageing.”
Dörfel et al., eBioMedicine, January 2026 · Peer reviewed
“The results do not establish pathological or beneficial structural brain change and do not construct or validate an athlete-specific normative model.”
Lindsay et al., Frontiers in Human Neuroscience, 2026 · Company-authored study
Related companies
Sources
- Public recordopenFDA 510(k), PMA and registration queries, no matches
- Public recordopenFDA 510(k) records for NeuroQuant, icobrain and Quantib
- Public recordSEC EDGAR company and full text searches, zero results
- Public recordBrain-volume profiles in athletes, PMID 42638706
- Public recordGeometric constraints on human brain function, PMID 37258669, with BrainKey shareholder disclosure
- Peer reviewedPrediction of brain age using structural MRI, comparing six packages
- Peer reviewedCommercial volumetric MRI reporting tools, a systematic review
- IndependentStartup BrainKey offers brain MRI direct to consumers
- IndependentCan we change how our brains age?
- ClinicianA Closer Look at Your Brain Health: BrainKey and Creyos
- InterviewOn building the 23andMe for brains with Owen Phillips
- IndependentACR Statement on Screening Total Body MRI
- CompanyBrainKey terms of service
- CompanyBrain Age: how we calculate it and what it means
- CompanyAbout BrainKey, founding story and advisory board
- CompanyBrainKey and Microsoft Azure
- CompanyBrainKey exhibitor profile naming SimonMed
Profile researched and written by Healthcare Discovery. Last updated September 29, 2026.
