T he idea that a single selfie could predict how long someone with cancer may live sounds like an incredible breakthrough. AI is getting better at spotting patterns in medical data, but can it be used to reliably predict what will happen to an individual patient? I asked a clinical colleague what she thought about the recent headlines . She found it an interesting development given that doctors have long used facial appearance to assess patient health. For example, she noted that they may write “appears stated age” or “appears older than stated age” as a proxy for general health.
Now, researchers from Mass General Brigham, a healthcare institution in Massachusetts, have taken this idea a step further. They have developed an AI model trained on more than 40 million facial images – called FAHR-Face – to assess biological age and predict survival in cancer patients. For biological age, they developed FAHR-FaceAge, which estimates age from certain facial features such as the forehead, the wrinkles between the nose and mouth and the temples. For cancer survival, they developed FAHR-FaceSurvival, trained using almost 35,000 photos of cancer patients. This model focused more closely on the area below the eyes and the bridge of the nose – regions of the face “ linked with dying ”. Dr Raymond Mak , one of the study’s authors, noted that these parts of the face are the same as those referenced by Hippocrates in 400BC as signs of muscle wasting and mortality risk: “the nose sharp, the eyes sunken, the temples fallen in”.
The results, published in The Lancet Digital Health this month, are intriguing – but far from conclusive. Researchers found that AI-estimated facial age was significantly associated with survival in 18 of 28 cancer categories. So, basically, it works for some cancers, but not for others. In addition, Prof Miroslaw Bober noted, “The tool still needs testing against doctors’ own judgments and across different hospitals and patient groups. For example, the facial-age measure was not significantly associated with survival for Black patients after adjustment for other factors.” That finding adds to the growing concern that AI models can exhibit racial bias based on the data used to train them and underlying assumptions. For example, racial bias has been shown when AI models analyse cardiac magnetic resonance (CMR) scans – used to diagnose and monitor heart disease.
Let’s go back to what the published study does and doesn’t tell us. These models are built on retrospective data looking at statistical associations across thousands of patients, and it’s a jump to then accurately predict future survival from this. A selfie may contain clues about a patient’s health, but that is a long way from being able to predict their future – and a longer way still from proving that the prediction is useful in clinical practice.
I also worry about how unreliable selfies are. Just think of the times you’ve met people in real life and compared them with their profile photos. Anyone who has tried one of the online AI “guess your age” tools will know how much the result varies with lighting, camera angle, facial expression or even time of day. And then there’s botox, fillers, facelifts and even GLP-1 drugs, which alter facial fat and shape. How much can a single photograph really tell us about someone’s underlying health?
As anti-ageing treatments become the norm , visible age and underlying health will become an increasingly complex relationship. A face in a selfie that looks younger may not be necessarily healthier. The researchers do note that they tested their age model against lighting, expression and cosmetic surgery – but if predictions are based on specific signs of ageing in parts of the face that those all affect, I can’t see how it would be possible to avoid a certain degree of error in AI estimates.
None of this means the research isn’t promising. This new tool and the research team behind it are impressive. Alongside other tests that a doctor may perform to assess a patient’s health and future prognosis, it could take some of the guesswork out of “looks stated age”. But the question is whether it’s actually useful in daily practice beyond what clinicians are already doing when using experience and years of training.
If you’re tempted to go to the project’s public website to find out your biological age, even just to brag to friends about how young it says you look, be aware that by uploading your selfie, you will become part of the study too, as their consent forms indicate. Your face is not only being assessed, it’s contributing to the data used to develop the technology. Currently, the tool will only tell you your biological age, but I would be sceptical of what that may tell you about your overall health.
Finally, a pop quiz (I’m in the middle of the teaching semester): what top three things did the researchers find made people’s face age? Alcohol, drugs and smoking. No surprise there, but possibly an explanation as to why doctors can broadly read someone’s health status from their face. Something that perhaps we didn’t need AI scanning 40m faces to tell us.
Prof Devi Sridhar is chair of global public health at the University of Edinburgh
