Study suggests AI can estimate cancer survival from a face photograph

Researchers at Mass General Brigham have developed AI tools that estimate biological age and cancer survival risk from a face photograph, according to The Guardian’s account of a study published in The Lancet Digital Health. The finding is striking, but it is a research result, not a ready-made treatment guide.

Watch Desk analysis

What happened

The researchers developed two tools from a model trained on more than 40 million facial photographs. FAHR-FaceAge estimates biological age and survival; FAHR-FaceSurvival predicts survival risk directly from a photograph. The Guardian reports that the tools drew on different facial regions for their predictions.

In a group of US cancer patients treated with radiotherapy, those assessed by FAHR-FaceAge as looking at least five years older than their actual age had a 21% higher risk of dying during an average four-year follow-up than the rest of the group. Those assessed as looking at least five years younger had a 17% lower risk. These are reported associations in that patient group, not an individual prognosis or proof that the tool improves care.

Why it matters

A photograph is easy to collect; a dependable, clinically useful estimate of treatment risk is much harder. If future research validates the approach, it could give clinicians another way to assess patients’ physical reserve before intensive treatment. The researchers’ stated aim is to support personalised decisions, not replace clinical judgement.

The distinction matters. A model identifying patterns associated with outcomes does not, by itself, show what caused them, whether its predictions work across patient groups, or whether acting on them benefits patients. The Guardian says more research is needed before widespread use.

Our read

This is an intriguing use of AI, and a reminder that a selfie can carry information its subject never meant to disclose. But “predicts survival” is not the same as “knows what will happen”. Until clinical usefulness is established, the proper place for this tool is in research, not in the role of oracle at the bedside.

What to watch

• Whether the study’s methods and results are tested in independent patient groups.
• How accurately the tools perform across different cancers and patient populations.
• Whether further research shows that using their estimates improves treatment decisions or patient outcomes.

Discussion spark: Should face-based AI risk estimates be used to inform cancer treatment discussions if future studies validate them, or is the risk of misleading patients too high?

Watch Desk is operated by WittyWires as an independent cross-cutting AI news tracker. It does not speak for the organisations or people it covers.