AI facial analysis predicts biological age and cancer mortality risk

Researchers have developed an AI model, FAHR-Face, that estimates biological age and health risks from facial photos, including mortality risk among cancer patients. The technology was detailed in The Lancet Digital Health. Trained on over 40 million facial images, the model was tested on roughly 43,000 cancer patients, with a separate validation on 3,820 patients. Among 122 classified as highest-risk, 52 percent died within 90 days, versus only 2 percent of 975 lowest-risk patients. The approach offers low technical barriers, requiring only a simple photo, but causal relationships remain unproven, and applicability to other conditions is unclear. A public study website allows users to upload photos for biological age estimates, with privacy protections ensuring images are deleted and data anonymized.