Facial Images May Help Forecast Cancer Treatment Survival
WHO:
CNN - Researchers at Mass General Brigham in Boston developed AI models to estimate cancer patients’ biological age and survival risk from facial photographs.
CNN - Study author Dr. Raymond Mak is a radiation oncologist at Mass General Brigham Cancer Institute and a faculty member in Mass General Brigham’s Artificial Intelligence in Medicine program.
CNN - Dr. Wilbur Lam, research chair and professor of pediatrics and biomedical engineering at Emory University, was not involved in the study.
CNN - Miroslaw Bober, professor of video processing and machine intelligence at the University of Surrey, was not involved in the study.
Fast Company - Study co-author Hugo Aerts said FAHR-FaceSurvival predicts survival risk directly from photographs; Aerts said it was, to their knowledge, the first demonstration of an AI model directly predicting cancer survival from a photograph.
Fast Company - Raymond Mak, identified as an oncologist at Mass General Brigham Cancer Institute, said the goal is to support personalized treatment decisions rather than replace physician judgment.
WHAT:
CNN - The researchers said the facial regions used by the models resemble signs Hippocrates described in *Book of Prognostics*, written around 400 B.C.: “the nose sharp, the eyes sunken, the temples fallen in.” Study author Raymond Mak said the resemblance was “mind-blowing” and “very eerie.”
Fast Company - Study co-author Hugo Aerts said FAHR-FaceAge complemented FAHR-FaceSurvival and improved the analysis of cancer survival rates. Aerts said the foundation-model approach could support developing multiple facial health biomarkers without training each model from scratch.
CNN - The study, published in The Lancet Digital Health, examined AI models that use facial images to estimate biological age and cancer-patient mortality risk.
CNN - FAHR-Face, the foundation model, was trained on more than 40 million facial images spanning diverse ages and ethnic backgrounds. It was used to build FAHR-FaceAge and FAHR-FaceSurvival.
Fast Company - The team trained FAHR-FaceSurvival using more than 40 million general face photographs and a separate set of more than 30,000 facial photographs of cancer patients linked to medical outcomes.
CNN - Among 122 patients classified as highest risk by FAHR-FaceSurvival, 52.5% died within 90 days; about 2% of 975 patients in the lowest-risk group died within 90 days.
Fast Company - About half of the 122 patients in the highest 10% of FAHR-FaceSurvival scores died within three months, compared with 2% in the lowest-score group.
CNN - Of 1,060 patients whose estimated biological age was more than 20 years above their chronological age, 21% died within 90 days. About 6% of 4,195 patients whose biological age was at least 10 years below their chronological age died within 90 days.
Fast Company - Researchers reported smoking was the variable most positively correlated with facial aging; current smokers appeared older than former smokers, and people who had quit smoking appeared older than nonsmokers.
WHERE:
CNN - The research team was at Mass General Brigham in Boston.
CNN - The study was conducted among cancer patients; Mak said further research could examine whether similar findings apply to otherwise healthy people.
WHEN:
CNN - The study assessed deaths within 90 days of evaluation. The article gives no calendar date for publication.
WHY:
CNN - Researchers said facial analysis could potentially provide doctors with additional information when assessing survival prospects and considering treatment approaches.
Fast Company - Mak said physicians need better ways to estimate physical reserve and frailty before recommending intensive treatments such as radiation, surgery, or systemic therapy.
HOW:
CNN - Mak said the team hopes that identifying accelerated aging or higher mortality scores could help flag patients for further testing for serious conditions.
Fast Company - Researchers linked cancer patients’ facial photographs to their medical outcomes to test FAHR-FaceSurvival.
CNN - FAHR-FaceAge estimates biological age from facial features; FAHR-FaceSurvival combines facial features with other clinical information to predict cancer survival.
CNN - The age model examines the temples, forehead above the eyes, and nasolabial folds. The survival model examines the bridge of the nose and area under the eyes.
Fast Company - Researchers said the age model focused on the nasolabial folds, forehead, and temples; the survival model focused below the eyes and around the bridge of the nose.
CNN - Researchers said cosmetic changes, including surgery, Botox injections, and makeup, did not affect model performance. Fast Company also reported consistency with makeup, challenging lighting, and nose surgery.
CNN - Mak said the algorithms were trained on facial photographs, age, and whether a person died, not on internet text or knowledge.
CNN - The authors said the models are not intended to replace physicians or existing clinical evaluations. Lam said clinical utility must be demonstrated prospectively; Bober said the tool needs testing against doctors’ judgments and across hospitals and patient groups.
(UPDATED October 21, 2026 06:00 UTC)
Researchers say cancer patients’ facial photographs revealed patterns linked to medical outcomes, and a new tool called FAHR-FaceAge complements their survival analysis. Study co-author Hugo Aerts says the foundation-model approach could help develop a range of facial health markers without building each model from scratch. The researchers hope the tools could flag patients for further testing, while study author Raymond Mak said the facial patterns echoed signs described by Hippocrates and called the resemblance “very eerie.”
--
(UPDATED October 21, 2026 05:09 UTC)
Researchers at Mass General Brigham have developed AI models that use facial photographs and clinical information to estimate cancer patients’ biological age and survival risk, with the findings published in The Lancet Digital Health. In the study, 52.5% of 122 patients identified as highest risk died within 90 days, compared with about 2% of 975 patients in the lowest-risk group; the researchers also found that smoking was strongly linked with facial aging. The models are intended to give doctors additional information when considering treatment, not replace clinical judgment, and researchers say their usefulness needs further testing.