October 19, 2026

expert reaction to study testing an AI tool using a face photograph to predict cancer survival

Scientists comment on a study, published in Lancet Digital Health, testing an AI tool using a face photograph to predict cancer survival.

Prof Miroslaw Bober, Professor of Video Processing and Machine Learning, Institute for People-Centred AI & CVSSP, University of Surrey, said:

“This is an interesting and substantial study, with an important strength being the validation of its findings in an independent group of nearly 5,000 cancer patients in the Netherlands.

“This study provides convincing evidence of a statistical link between facial appearance and survival in cancer patients receiving radiotherapy. In the US test group, over half of the patients given the highest survival-risk scores died within three months, which shows the tool can separate patients by risk.

“Predicting an outcome is not the same as explaining it. Despite the references to ‘biological age’, the AI estimates age from facial appearance rather than directly measuring any underlying biological ageing process.

“The authors link the facial regions the AI relies on to the ‘look of dying’ described by doctors. However, the study did not test whether the AI recognises anything a doctor would not. It has not shown that these predictions add clinically useful information beyond doctors’ assessments of a patient’s general health and fitness, which were not measured in the study.

“The press release broadly reflects the results, but its headline goes further than the evidence. The tool ranks patients by relative risk within a group, but its scores have not been validated as individual survival probabilities or for clinical decision-making.

“Smiling or changing the lighting altered the AI’s estimate of a person’s age by around two years on average, and considerably more in some cases. Claims that the tool could track ageing or lifestyle changes over time are therefore premature, as this was not tested in the study.

“This is a promising proof of concept, but it is not ready for clinical use. The tool still needs testing against doctors’ own judgement 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, although the separate survival model was.”

‘Foundation artificial intelligence models for health recognition by use of face photographs (FAHR-Face): a model development and validation study’ by Fridolin Haugg et al. was published in The Lancet Digital Health at 23:30 UK time on Monday 19^th October.

Declared interests

Prof Miroslaw Bober:

Prof Bober is Chief Technology Officer of Forecom.AI, which develops AI methods for biological imaging and clinical risk modelling, and has provided AI consultancy to life-sciences companies including Pfizer and Bristol Myers Squibb (BMS), as well as NHS organisations. He has also undertaken research collaborations with the NHS.