AI tool predicts cancer survival from a single facial photo

Cancer patients who look five or more years older face 21% higher death risk

Researchers at Mass General Brigham, a nonprofit integrated academic healthcare system in Massachusetts, built the AI system to estimate biological age and directly predict survival outcomes in patients diagnosed with cancer. The study, published in the Lancet Digital Health journal, is one of the most comprehensive demonstrations to date of how AI analysis of a selfie could help assess a patient’s health and potentially guide their medical care.

The system was trained on data from more than 40 million facial photographs across a range of ethnicities and ages. From that model, the researchers developed two complementary tools. FAHR-FaceAge estimates a patient’s biological age — the age a person looks, rather than their actual, chronological age — and predicts cancer survival. FAHR-FaceSurvival predicts survival from a facial photograph directly, rather than inferring survival through biological age. Although the two tools were derived from the same model, they captured different and complementary information about a patient’s health, the researchers said.

Each algorithm relied on distinct parts of the face to make its predictions. FAHR-FaceAge focuses on the creases that run from the sides of the nose down to the corners of the mouth, the forehead area just above the eye sockets, and the temples. FAHR-FaceSurvival evaluates the region below the eyes and around the nasal bridge.

Hugo Aerts, a co-senior author of the study, said the survival model is the first of its kind. “FAHR-FaceSurvival advances the field of facial health recognition AI a step further by learning survival risk directly from facial photographs of patients with cancer,” he said. “To our knowledge, this is the first demonstration that an AI model can directly predict cancer survival from a simple photograph.”

Among cancer patients treated with radiotherapy in the US, the people who, according to FAHR-FaceAge, looked five or more years older than their actual age had a 21% higher risk of dying during the average four-year follow-up than the rest of the group, the study found. The patients who looked five or more years younger had a 17% lower risk.

Aerts said the foundation-model approach lets the system adapt to a range of health outcomes. “The key technical advance is that FAHR-Face learns a general representation of the face from more than 40 million images, which can then be adapted to distinct health outcomes such as biological age and survival,” he said. “This foundation-model approach gives us a scalable way to develop multiple facial health biomarkers rather than training each model from scratch.”

The results suggest that people’s biological age estimates are closely linked to their physical health, which could reflect their ability to survive certain treatments, the researchers said.

Dr Raymond Mak, a radiation oncologist at Mass General Brigham, said the tool is meant to support treatment decisions for an aging population. “As the population ages, physicians need better ways to estimate physical reserve and frailty before recommending intensive treatments such as radiation, surgery or systemic therapy,” he said. “Our goal is to make face photograph-based health assessment more quantitative, consistent and scalable so it can support personalised treatment decisions rather than replace physician judgment.”

The researchers found that lifestyle factors were associated with facial aging. Heavy alcohol use, drug use and smoking were linked to looking biologically older in the AI analysis of facial photographs.

Mak said FAHR-FaceAge held up across common real-world photo variations and most tested cosmetic procedures. “Another surprising finding was how consistent FAHR-FaceAge was across common real-world photo variations, including pose, lighting, facial expression and makeup,” he said. “It was also stable across most tested plastic surgery procedures, with only a small shift observed after rhinoplasty.”

The system could help predict the risk of death or complications after major surgery or other cancer treatments, and could assist doctors in making difficult decisions about the best way forward for their patients, the researchers said. They cautioned that while single facial photographs could offer an easy, low-cost way to support personalized treatment decisions, more research was needed before the tool could be widely used.