AI Cancer Selfie Tool Predicts Survival From a Photo

An AI cancer selfie tool is making headlines after doctors at Mass General Brigham showed it can predict a cancer patient's chances of survival from one photo of their face. The work, published in Lancet Digital Health, is described as one of the most comprehensive demonstrations yet of selfie-based health assessment, according to The Guardian.

What the AI actually does

The system was trained on more than forty million facial photographs across a range of ethnicities and ages. From that base model, the team built two tools. One, called FAHR-FaceAge, estimates biological age, meaning the age a person looks rather than the number on their birth certificate. The other, FAHR-FaceSurvival, predicts survival directly from a face photo.

Hugo Aerts, a co-senior author, said the key advance is that the model learns a general picture of the face that can then be adapted to different health outcomes. He called it a scalable way to build several facial health markers without starting from scratch each time.

The numbers behind the headline

Among US cancer patients treated with radiotherapy, those who looked five or more years older than their real age had a twenty-one percent higher risk of dying during the average four-year follow-up. Patients who looked five or more years younger had a seventeen percent lower risk, according to The Guardian.

Scientific American adds more detail. In a test group of three thousand eight hundred and twenty cancer patients, more than half of those with the highest risk score, sixty-four out of one hundred and twenty-two, died within ninety days of their photo being taken. A separate group of more than thirty-eight thousand patients showed that looking older than your age tracked with higher risk, according to Scientific American.

Where the AI looks on your face

The researchers used attention maps to see which parts of the face the models relied on. For age, the AI focused on the creases from the nose to the corners of the mouth, the forehead above the eye sockets, and the temples. For survival, it looked at the area under the eyes and around the bridge of the nose.

Nobody yet knows why those spots matter for risk of death. The two scores also work better together, and the lead researcher said combining them gave better survival predictions than either alone.

The tool also seemed tough to fool. Dr Raymond Mak said the age estimate stayed consistent across pose, lighting, facial expression and makeup. It held up across most plastic surgery too, with only a small shift after a nose job. Lifestyle showed up on faces as well, since heavy drinking, drug use and smoking were linked to looking biologically older.

Why doctors want it, and why people worry

Mak explained that doctors already size up how frail a patient looks when choosing between surgery, radiation or other intensive treatments, but that judgment is subjective. The goal is to make it more consistent and measurable, and to support decisions rather than replace physician judgment.

Ethical questions are already on the table. NYU bioethicist Matthew Liao told Scientific American that a wrong call could mean someone who needs care does not get it. He also raised the risk of insurers or employers using face-based health scores to deny coverage or screen job candidates.

There are limits too. The survival model was trained on a Boston-area patient group that was mostly white, and the team is working to widen its training data. Doctors say more research is needed before anything like this reaches routine care.

What to watch next

For now this is a research result, not something your doctor will run on your phone. Mass General Brigham has a patent pending on its face health algorithms, so expect more testing. If you want more health and tech stories like this one, browse our health coverage.

The big question for Gen Z is simple: if a selfie can say that much about your health, who gets to look at it? Hospitals say the tool is meant to help patients, and the next round of clinical testing will show whether that promise holds.