MGB AI Trained on 40M Faces Predicts Cancer Survival
Researchers at Somerville, Mass.-based Mass General Brigham have built an AI foundation model that estimates a patient's biological age and predicts cancer survival from a single photograph. According to Becker's Hospital Review, the model, called FAHR-Face, was trained on more than 40 million face images and was detailed in a study published Oct. 19 in The Lancet Digital Health.
The model powers two tools. FAHR-FaceAge estimates biological age, the idea being that a person's face can reveal how the body is aging relative to calendar years. A second tool uses that signal to forecast survival outcomes in cancer patients. The premise is that biological age, not just chronological age, carries clinical weight when clinicians weigh treatment intensity and prognosis.
The work fits a broader push to turn routine, low-cost inputs like photographs into clinical biomarkers. If validated in prospective settings, facial imaging could supplement existing risk tools. It also raises familiar questions about bias, consent, and how such predictions get used in care decisions.
Sources