An artificial intelligence (AI) tool may be able to assist doctors in judging a patient's illness and estimating survival using just one photograph of their face, US researchers have reported. The preliminary findings have been published in the Lancet Digital Health journal.
Named 'FaceSurvival', the AI tool was trained using millions of healthy facial images and 35,000 photographs of cancer patients. Scientists indicate that it makes estimates by closely examining features around the eyes and the bridge of the nose.
While the tool is currently a long way from being ready for use in hospitals, researchers aim for it to help doctors recommend appropriate treatments. Dr Raymond Mak, lead researcher from Mass General Brigham in Boston, explained that the goal is to make face photograph-based health assessment more quantitative and consistent to support personalised treatment decisions.
The research team is also evaluating a separate AI tool, 'FaceAge', which predicts biological age based on facial features like nasolabial folds, marionette lines, frown lines, crow's feet, and cheeks. However, experts like Prof Miroslaw Bober from the University of Surrey caution that any claims about tracking ageing or lifestyle changes over time would be "premature", and the tool is not yet ready for clinical use.