A new artificial intelligence tool can estimate a cancer patient’s chances of survival from a single photograph of their face, according to a study published in Lancet Digital Health. Patients whose faces looked five or more years olderthan their actual age had a 21% higher risk of death, the researchers found. The system was trained on more than 40 million facial images spanning a wide range of ethnicities and ages.
Who developed it
The AI system was created by specialists at Mass General Brigham, a non-profit, integrated academic healthcare system in Massachusetts, to estimate biological age and directly predict survival outcomes in patients diagnosed with cancer.
It 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.
How it works
The system can estimate a person’s biological age, meaning the age they appear to be rather than their actual, chronological age, and predict survival chances for cancer patients from a simple photo of their face.
The researchers used the technology to develop two complementary tools:
- FAHR-FaceAge estimates a patient’s biological age and predicts cancer survival
- FAHR-FaceSurvival predicts survival directly from a facial photo, rather than through biological age
Although both were derived from the same model, they captured different and complementary information about a patient’s health.
Different parts of the face
Each algorithm also relied on different areas of the face to make its predictions about age and mortality risk.
FAHR-FaceAge focuses on the lines running from the sides of the nose to the corners of the mouth, the area of the forehead just above the eye sockets, and the temples. FAHR-FaceSurvival, on the other hand, assesses the area under the eyes and around the bridge of the nose.
“The first evidence”
“FAHR-FaceSurvival takes the field of facial health AI a step further, as it learns survival risk directly from facial photographs of cancer patients,” said Hugo Aerts, one of the study’s two lead authors. “To our knowledge, this is the first evidence that an AI model can directly predict cancer survival from a simple photograph.”
What the numbers show
Among cancer patients who underwent radiotherapy in the US, over an average follow-up of four years:
- those whom FAHR-FaceAge judged to look five or more years older than their real age had a 21% higher risk of death than the rest of the group
- those who looked five or more years younger had a 17% lower risk
“The key technical innovation is that FAHR-Face learns a general representation of the face from more than 40 million images, which can then be adapted to specific health outcomes such as biological age and survival,” Aerts said. “This foundation model approach gives us a scalable way to develop multiple facial health biomarkers, rather than training each model from scratch.”
Helping doctors decide on treatment
The findings suggest that estimates of a person’s biological age are closely linked to their physical health, which could reflect their ability to withstand certain treatments.
The discovery could pave the way for AI facial analysis to help doctors make difficult decisions about the best course of treatment. It could also help better predict the risk of death or complications after major surgery or other cancer treatments.
Dr Raymond Mak, a radiation oncologist at Mass General Brigham, said: “As the population ages, doctors need better ways to assess physical resilience and frailty before recommending intensive treatments such as radiotherapy, surgery or systemic therapy.”
“Our goal is to make facial photo-based health assessment more quantitative, consistent and scalable, so it can support personalised treatment decisions, rather than replace a doctor’s judgement.”
Lifestyle and facial ageing
The researchers found that lifestyle factors were associated with facial ageing. For example, heavy drinking, drug use and smoking were linked to a biologically older appearance, according to the AI analysis of the photos.
Reliable across everyday photos
“Another surprising finding was how consistent FAHR-FaceAge was across common real-life photo variations, such as posture, lighting, facial expression and make-up,” Mak said. “It was also stable across most of the cosmetic surgeries tested, with only a small change observed after rhinoplasty.”
More research needed
The doctors cautioned that, while single facial photos could offer an easy and low-cost way to support personalised treatment decisions, further research is needed before the tool can be widely used.
With information from The Guardian
