AI facial analysis predicts cancer survival from photos

A new AI model, FAHR-Face, can predict cancer survival outcomes from facial photographs, potentially transforming prognostic assessments. The model, trained on over 40 million images, estimates biological age and directly predicts mortality risk, with high-risk patients showing over three times the mortality hazard of low-risk patients. The model's age estimation achieved a mean absolute error of 5.1 years, outperforming benchmarks. Patients appearing older than their chronological age had poorer survival, with a 21% higher mortality hazard for those with a facial age deviation of at least five years. Smoking and heavy substance use were also linked to older apparent age. The survival model showed striking early mortality differences, with 52.5% of the highest-risk group dying within 90 days versus 2.1% in the lowest-risk group. However, the models require prospective validation across diverse populations and clinical settings before routine use, and privacy concerns regarding facial biometric data must be addressed.