AI Tool Predicts Cancer Patient Survival From a Single Facial Photo

Researchers have developed an artificial intelligence tool that can predict cancer patients' survival odds using only a single facial photo. The model was trained on more than 40 million facial images and photos of over 30,000 cancer patients, according to a study published in The Lancet Digital Health.

The tool generates a risk score for each patient, with those scoring highest more likely to die within 90 days. The research team said the model estimates age with an error of about five years and reliably separates patients into high-risk and low-risk groups with significantly different mortality risks. It performed consistently across different ages, races, sexes, and cancer types.

Dr. Raymond Mak, a radiation oncologist at Mass General Brigham Cancer Institute and co-author of the study, said doctors are trained to assess patients through their appearance but often struggle to accurately estimate the remaining survival time of terminally ill patients. He said the team believes AI can capture part of the body's physiological reserve and mortality risk from a single facial photo.

Miroslaw Bober, a professor of video processing and machine learning at the University of Surrey who was not involved in the study, cautioned that the tool is not yet ready for clinical use. He said prediction does not equal explaining causation, and the study has not proven the tool provides additional clinically valuable information beyond doctors' overall health assessments. He called for more testing across different hospitals and patient populations.