Cancer · Prediction and risk
A face photo that estimates your 'biological age' and cancer survival
Researchers trained an AI on more than 40 million face photos to estimate 'biological age' and, in people who already have cancer, survival risk. It held up in a separate group of patients. This is a study of past data. It hasn't been tested in real care, and it can't diagnose anything. Rating: Early research.
How far away is this?
A peer-reviewed study that built and checked the model on past data. No one has tested whether using it changes care or outcomes. Funded by the EU European Research Council; the paper states 'The funders of the study had no role in the study design, data collection, data analysis, data interpretation, or writing of the report.' Three of the authors report that their hospital system, Mass General Brigham, has filed a provisional patent application on face-based health algorithms, and several authors report advisory, consulting or research-funding ties to drug and medical-tech companies such as AstraZeneca, Varian and Siemens (some described as unrelated to this work).
Get the weekly How Far
One email every Friday morning (Sydney time) on which medical AI actually reached patients, and what it means for you. Free, in plain English.
Where is it approved?
| Region | Regulator | Status | Notes |
|---|---|---|---|
| United States | FDA | Not applicable | Research model; not a medical device. |
| European Union | CE mark / EMA | Not applicable | No register record checked. |
| United Kingdom | MHRA | Not applicable | No register record checked. |
| Australia | TGA | Not applicable | No ARTG record checked. |
Research model only; not on any register. "Register not searched yet" means we haven't searched that regulator's register; it doesn't mean the product is or isn't available. Approval in one country isn't approval in another.
What is it?
You take an ordinary photo of someone's face. The AI estimates how old their body 'looks' and, for people already diagnosed with cancer, how their survival risk compares with others.
What's the evidence?
FAHR-FaceAge estimated age with an average error of 5.1 years on public datasets. In plain words: on average it was about five years off, which was the best of the models compared.[1]
Evidence stage: Early research · #c1
In people with cancer, those the AI placed in the highest-risk quarter had more than three times the death rate of the lowest-risk quarter (adjusted hazard ratio 3.23), after accounting for usual clinical factors. This was an analysis of past patient records, so it shows a link with survival. It doesn't show that the score changes anyone's treatment or outcome. The researchers report it worked similarly across age, sex, race and cancer subgroups in the groups they tested, but it hasn't been tested in real care.[1]
Evidence stage: Early research · #c2
What does it mean for me?
No one can run this on your selfie at the doctor's today. If a company offers you a 'face age' score, ask what study backs it and who sees your photo.
What could I ask my doctor?
- Who would see my photo and the score?
- Has this been tested in people like me?
These are general starting points, not a recommendation to choose any particular test or treatment.
Short videos, under a minute
We make short, plain-English videos about AI in medicine, including what it can't do yet. Follow How Far for the next one.
Change history
- : First published.
- : Editor pass before publication: labels aligned to site categories; sources and rating reviewed.
General information only, not medical advice. Talk to your doctor about your situation. AI-assisted source check. Not clinically reviewed.
References
- FAHR-Face study, The Lancet Digital Health (2026). doi.org (2026). Peer-reviewed study. https://doi.org/10.1016/j.landig.2026.101056
Publisher shown as the website hosting each source. Dates are given where the source label includes them.