US scientists develop an AI tool that can predict a cancer patient's chances of survival from a single selfie

Cancer patients whose faces appeared to be more than five years older than their chronological age were 21% more likely to die during an average four-year follow-up period. Those the tool placed in the highest-risk category faced a death rate more than three times that of patients in the lowest-risk group. The research team at Mass General Brigham in Boston built the model using upwards of 40 million facial images. Their results appeared on Monday in The Lancet Digital Health.

To assess survival risk, the tool analyses the skin under the eyes and across the nose bridge.

Smokers registered as biologically older than non-smokers, with current smokers showing the greatest age gap of all.

Professor Hugo Aerts described it as 'the first demonstration' of AI predicting cancer survival from a photograph.