Selfies can be used to predict cancer patients' survival rate

Telegraph

09-05-2025

Selfies could predict a person's chance of surviving cancer, a study has suggested.
Doctors believe a new artificial intelligence tool that measures the 'biological age' of a patient based on a photo of their face, could inform the type of cancer treatment they receive.
Knowing someone's biological age, rather than their actual age, is a better predictor of someone's overall health and life expectancy, a team from Mass General Brigham, a non-profit research group in the United States claims.
The FaceAge AI tool scans an image of a person's face to estimate their biological age, which is based on factors including lifestyle and genetics.
It is akin to what doctors call an 'eyeball test', in which doctors make judgements about overall health based on appearance, which in turn informs decisions about whether a person is strong or fit enough to undergo intensive cancer treatment.
Researchers said they wanted to see whether they could 'go beyond' the 'subjective and manual' eyeball test by creating a 'deep learning' AI tool that could assess 'simple selfies'.
The new AI algorithm was trained using 59,000 photos.
Dr Hugo Aerts, one of the authors, said it was the first study to show 'we can really use AI to turn a selfie into a real biomarker source of ageing'.
He said the tool is low cost, can be used repeatedly over time and could be used to track an individual's biological age over 'months, years and decades'.
'The impact can be very large, because we now have a way to actually very easily monitor a patient's health status continuously and this could help us to better predict the risk of death or complications after, say, for example, a major surgery or other treatments,' he added.
Explaining the tool, academics showed how it assessed the biological age of actors Paul Rudd and Wilford Brimley based on photographs of the men when they were both 50 years old. Mr Rudd's biological age was calculated to be 42.6, while Mr Brimley, who died in 2020, was assessed to have a biological age of 69.
The new study, published in the journal Lancet Digital Health, saw the FaceAge tool used on 6,200 patients with cancer using images taken at the start of their treatment.
The academics found that the biological age of patients with cancer was, on average, five years older than actual age.
They also found that the older biological age the tool gave them, the worse survival rates the cancer patients had, especially if FaceAge gave them a reading of more than 85.
The authors concluded: 'Our results suggest that a deep learning model can estimate biological age from face photographs and thereby enhance survival prediction in patients with cancer.'
Dr Ray Mak, another author of the research, said turning the photo into a measure of health was 'like having another vital sign data point' and said it was 'another piece of the puzzle like vital signs, lab results or medical imaging'.
But he added: 'We want to be clear that we view AI tools like FaceAge as assistance to provide decision support and not replacements for clinician judgement.'
More studies assessing FaceAge are under way, including whether it could be used for other conditions or diseases and what impact things like cosmetic surgery or Botox have on the tool.