An ordinary face photo may contain health clues that aren’t immediately obvious. Researchers at Mass General Brigham have developed an AI system that analyzes facial photographs and estimates a cancer patient’s biological age and survival chances. The findings, published in The Lancet Digital Health, suggest facial analysis could eventually help doctors make more informed treatment decisions.
How can AI estimate cancer survival from a face?
The researchers developed FAHR-Face, a foundation model trained on more than 40 million facial images spanning different ages and ethnicities. They adapted it into two tools with different purposes. FAHR-FaceAge estimates your biological age, meaning how old you look rather than your actual age, and uses that to predict cancer survival. Whereas FAHR-FaceSurvival predicts survival risk straight from the photo.
Each tool focuses on different parts of your face. FaceAge zeroes in on creases near your nose and mouth, your forehead, and your temples, while FaceSurvival studies the area below your eyes and around your nasal bridge. The AI models showed promising results across different cancer types and demographic groups.
Patients classified in the highest-risk quartile by FAHR-FaceSurvival had more than three times the mortality risk of those in the lowest-risk group. This association remained significant even after adjusting for several clinical factors.
The age-estimation tool found another pattern. Cancer patients who appeared at least five years older than their actual age had a 21% higher risk of dying within the four-year follow-up period, while those who looked at least five years younger had a 17% lower risk.
What else did the study discover?
Researchers also found that heavy drinking, drug use, and smoking were linked to looking biologically older in these scans. Interestingly, the tool stayed accurate across lighting changes, makeup, and most plastic surgery, with only a small shift showing up after rhinoplasty. However, these results describe statistical risks across patient groups, not an individual’s exact lifespan.
Dr. Raymond Mak, a radiation oncologist involved in the study, told The Guardian that their goal is to help doctors judge a patient’s physical resilience before recommending intense treatments like radiation or surgery. The researchers also say that this AI tool is meant to support physician judgment, not replace it. For now, the findings are promising, but further research is needed before widespread clinical use
RELATED COVERAGE: