Researchers at Harvard Medical School and Mass General Brigham have developed FAHR-Face, an AI system trained on more than 40 million facial photographs that estimates biological age and, for people with cancer, short-term mortality risk. The research was published on October 19, 2026, in The Lancet Digital Health.
The system has two components. FAHR-FaceAge estimates biological age—how old a person’s body appears to have aged relative to their chronological age—while FAHR-FaceSurvival is designed to estimate mortality risk in cancer patients. The researchers identified visual signals in areas including below the eyes and around the nose, and reported associations between accelerated facial aging and behaviors such as smoking, heavy alcohol consumption, and drug use.
In testing involving tens of thousands of cancer patients, including an independent cohort of 3,820 people, the model separated patients into sharply different 90-day risk groups. About 52% of the 122 patients assigned to the highest-risk group died within 90 days, compared with 2% of 975 people in the lowest-risk group. The reported results remained relevant across changes in lighting, facial expression, and makeup.
Those findings describe statistical associations rather than a diagnosis or a certain prediction of an individual patient’s lifespan. The researchers envision tools of this kind as potential support for clinicians assessing a patient’s overall condition before demanding treatments, alongside—not in place of—medical examinations and laboratory tests.
A public FaceAge research platform allows participants to submit three to 10 facial photos after completing consent and questionnaire steps, then receive a biological-age estimate by email. The process is described as taking less than 10 minutes, and submitted photos are used for research. The public service uses the team’s earlier FaceAge model and does not provide mortality-risk estimates.
