Mass General Brigham AI Predicts Cancer Survival From 40 Million Facial Photos
Updated
Updated · The Guardian · Oct 19
Mass General Brigham AI Predicts Cancer Survival From 40 Million Facial Photos
3 articles · Updated · The Guardian · Oct 19
Summary
A Lancet Digital Health study found Mass General Brigham’s AI could estimate biological age and directly predict cancer survival from a single facial photograph.
More than 40 million images trained the system, which produced two tools: FAHR-FaceAge, linking apparent age to mortality risk, and FAHR-FaceSurvival, which predicts survival directly from facial features.
Among U.S. radiotherapy patients, people the model judged 5 or more years older than their actual age had a 21% higher death risk over roughly four years, while those looking 5 or more years younger had a 17% lower risk.
Researchers said the model stayed consistent across lighting, pose, makeup and most plastic surgery changes, while smoking, heavy alcohol use and drug use were associated with older-looking biological age.
Doctors said the tool could help gauge frailty before surgery, radiation or systemic therapy, but cautioned that more validation is needed before broad clinical use.