FaceAge, a face-reading AI tool that estimates biological age from facial photographs, predicts cancer outcomes with an impressive 81% accuracy rate. This surpasses traditional methods and outperforms doctors in survival assessments.
Developed by researchers at Mass General Brigham, FaceAge shows early promise in predicting short-term life expectancy. The findings suggest the AI tool can identify high-risk patients and provide a low-cost, accessible way to inform care decisions, enhancing clinical decision-making.
How FaceAge translates facial features into health and age predictions
Trained on nearly 59,000 images of presumed-healthy adults aged 60 and older, FaceAge can identify subtle facial features associated with aging beyond obvious traits such as hair coloring or balding.
Researchers tested FaceAge on 6,196 cancer patients in the US and the Netherlands, and it consistently estimated their biological age to be about five years older than their actual chronological age. This means that, on average, cancer patients appeared older in terms of physiological aging than their birthdate would suggest.
By analyzing facial features like wrinkles, skin texture, and pigmentation, the AI tool captures aging markers invisible to the naked eye. It provides a quantifiable measure of biological aging that correlates with survival outcomes and health risks.
FaceAge vs. clinicians’ predictions
In testing, FaceAge outperformed clinicians in predicting short-term life expectancy for patients receiving palliative radiotherapy.
When 10 clinicians and researchers were asked to estimate survival based on 100 patient photos, their predictions were only slightly better than chance. Adding FaceAge data, however, significantly improved accuracy, raising prediction reliability from 61% to 80%.
Interestingly, the AI healthcare tool alone surpassed doctors’ predictions, scoring 81%.
This improvement was observed across multiple cancer types and stages, with FaceAge delivering a reliable prognostic signal, offering doctors more objective and precise survival estimates.
Medical community weighs in on the AI tool
FaceAge has drawn a mix of cautious optimism and concern from medical experts.
According to The Washington Post, Irbaz Riaz, assistant professor of medicine and artificial intelligence consultant at Mayo Clinic, called it “a promising early-stage tool.”
Hugo Aerts, director of the AI in Medicine program at Mass General Brigham and a lead researcher on the tool, acknowledged its potential benefits but warned it could also cause harm if misused. He stressed that hospitals have strict rules and oversight to ensure AI is used properly and only to benefit patients, not insurers or other parties.
FaceAge’s broader potential
Mass General Brigham researchers say more studies are needed before FaceAge can be routinely used in clinical settings. Their goal is to turn it into an early detection system for various health issues, supported by strong ethical and regulatory standards.
Read eWeek’s coverage of the top AI healthcare software solutions.