AI facial health tools estimate biological age and predict cancer survival from photographs

Mass General Brigham researchers have developed a new AI foundation model trained on more than 40 million facial images that can estimate biological age and directly predict survival risk in cancer patients from a simple photograph. The work, published in Lancet Digital Health, introduces two tools-FAHR-FaceAge and FAHR-FaceSurvival-derived from the same underlying model, offering physicians a more quantitative way to assess frailty and guide personalized treatment decisions.

"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," said Raymond Mak, MD, a radiation oncologist at Mass General Brigham Cancer Institute and corresponding author of the study. "Our goal is to make face photograph-based health assessment more quantitative, consistent and scalable so it can support personalized treatment decisions rather than replace physician judgment."

How the foundation model works

The core technical advance is a model called FAHR-Face (Foundation Artificial Intelligence for Health Recognition), which learns a general representation of the human face from more than 40 million images. From this single foundation, the team built two distinct applications without training each from scratch. FAHR-FaceAge estimates biological age with greater accuracy than prior algorithms, while FAHR-FaceSurvival predicts cancer survival outcomes directly from a photograph.

"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," said Hugo Aerts, PhD, co-senior author and Director of the Artificial Intelligence in Medicine program at Mass General Brigham. "This foundation-model approach gives us a scalable way to develop multiple facial health biomarkers rather than training each model from scratch."

Different faces, different signals

Although both tools share the same foundation model, they capture complementary information. Their predictions were only weakly correlated, and both remained independently associated with survival when analyzed together. Combining the two measures improved survival-model fit in an external validation cohort of cancer patients.

Each algorithm also relies on distinct facial regions. FAHR-FaceAge focuses on the nasolabial folds, the forehead area just above the orbits, and the temples. FAHR-FaceSurvival evaluates the region below the eyes and around the nasal bridge. "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," Aerts said. "To our knowledge, this is the first demonstration that an AI model can directly predict cancer survival from a simple photograph."

Lifestyle factors and real-world consistency

The study also found that lifestyle choices leave measurable marks on facial aging. Heavy alcohol use, drug use, and smoking were all associated with looking biologically older. Smoking showed the strongest pattern: current smokers looked older than former smokers, and former smokers looked older than those who had never smoked.

The tools proved stable across common photo variations including pose, lighting, facial expression, and makeup. They also held up across most tested plastic surgery procedures, with only a small shift observed after rhinoplasty. The research team has opened a public-facing study at faceage.bwh.harvard.edu where participants can upload a photograph and receive a FaceAge research readout while contributing data to improve the algorithms across larger, more diverse populations.

Why this matters for healthcare, research, and communications professionals

For clinicians and health systems, tools like FAHR-FaceAge and FAHR-FaceSurvival point toward a future where routine patient intake could include an objective, photograph-based frailty assessment-no blood draw or imaging required. For researchers, the foundation-model architecture offers a template for developing multiple biomarkers from a single training investment, which could accelerate work in aging and oncology studies. Professionals in AI for healthcare and AI research will need to understand how these models are validated and where they fit into clinical workflows. For marketing and communications teams, the study is a case study in translating complex AI methodology into a tangible, patient-facing application-without overpromising what the technology can do.

The study was funded by the EU European Research Council. Read the full paper at Lancet Digital Health.