AI tool on a single face photo estimates biological age and cancer survival risk

What the model does

Researchers at Mass General Brigham, a nonprofit academic health system in Massachusetts, have built an artificial intelligence system that works from a single photograph of a person's face. The system, called FAHR-Face, was trained on more than 40 million facial photographs covering a range of ethnicities and ages. Its purpose is to estimate biological age, meaning how old a person's body appears in relation to their birth date, and, in people with cancer, to estimate survival chances. The findings were published in The Lancet Digital Health, and the study appeared on Monday, 19 October. The team's stated aim is to make health assessment from facial photos more quantitative, consistent and scalable.

Two tools, two facial regions

From the same model the researchers developed two complementary tools. FAHR-FaceAge estimates biological age and uses that estimate to predict survival in cancer patients. FAHR-FaceSurvival predicts survival directly from a photograph, without going through biological age. Although both come from one model, they examine different parts of the face. FAHR-FaceAge focuses on the creases running from the sides of the nose to the corners of the mouth, the forehead above the eye sockets and the temples. FAHR-FaceSurvival weighs the area below the eyes and the region around the bridge of the nose.

Results in cancer patients

In a test group of 3,820 cancer patients, more than half of those with the highest risk score died within 90 days of their photograph being taken. A second test group of more than 38,000 cancer patients showed a similar pattern, with patients who looked older than their chronological age also facing a higher risk of death. Among patients treated with radiotherapy in the United States, those whom FAHR-FaceAge judged to look five or more years older had a 21% higher risk of dying over an average follow-up of four years. Patients who looked at least five years younger had a 17% lower risk, and the results suggest biological age estimates are closely tied to physical health.

The cohort ages ranged widely. Most cancer patients in the study were in their 50s and 60s, though it also included some children and people over 90.

Lifestyle links and robustness

The analysis also linked lifestyle to facial appearance. Heavy alcohol use, drug use and smoking were all associated with a biologically older look, and smoking showed the strongest association. The researchers tested stability under typical photo conditions, varying lighting, face position, expression and makeup. The model stayed robust in most cases, including after cosmetic surgery, with a small change recorded after nose correction. It is not yet clear why those particular facial areas correlate with a person's risk of death.

Limits and next steps

The researchers say the tools are designed to support doctors, not replace their judgment. The authors expect the approach could help assess physiological reserve and frailty before intensive treatments such as surgery, radiotherapy or systemic therapy. Further testing is still required before the system can be used in hospitals, and a patent application for the technology is pending. The tools could also help predict the risk of death or complications after surgery and other cancer treatments, though that use remains to be validated.