Researchers at Mass General Brigham in Boston have developed an artificial intelligence model that predicts the likelihood of survival for cancer patients using a simple photograph of their face. This novel approach could become a valuable tool in assessing patient health and treatment options, according to a study published in the journal The Lancet Digital Health.
The AI system was trained on over 40 million images, analyzing various facial features to estimate a patient’s biological age and their chances of survival. Dr. Raymond Mak, a radiation oncologist involved in the study, noted the efficacy of the AI, stating, “I was pretty surprised to find out that we could pick up signs of mortality risk from just a facial photograph.”
The technology incorporates two algorithms: FAHR-FaceAge estimates biological age based on facial features, while FAHR-FaceSurvival predicts cancer survival directly. Each model focuses on different areas of the face: FAHR-FaceAge looks at the nasolabial folds, forehead, and temples, whereas FAHR-FaceSurvival analyzes the bridge of the nose and areas below the eyes. This dual approach yields more accurate predictions about a patient’s health.
In a recent study, it was found that among 122 patients whose photographs indicated a high risk of death, 52.5% died within 90 days. In contrast, only about 2% of patients identified as having a lower risk died in the same period. The study highlighted that individuals who appeared biologically older than their chronological age faced a 21% higher risk of death within 90 days, showing the potential implications of biological age assessments on survival predictions.
Researchers have launched clinical trials to investigate this technology further. One ongoing trial focuses on older patients with early-stage lung cancer, aiming to identify frail individuals who might require additional support before undergoing treatment. Another trial examines the use of this AI system in combination with data from wearable devices to flag patients at risk of surgical complications.
While the AI model shows promising capabilities, experts have cautioned that its effectiveness across diverse populations needs further validation. Dr. Wilbur Lam from Emory University emphasized the importance of prospective evaluations to ascertain the model’s clinical utility beyond the existing research context.
Doctors believe that integrating this technology into clinical settings could improve patient assessments and inform treatment decisions. Dr. Mak stated, “As the population ages, physicians need better ways to estimate physical reserve and frailty before recommending intensive treatments.”
Despite the tool’s potential, there are concerns about ethical implications, including privacy and misuse in insurance and employment contexts. Matthew Liao, a bioethics expert, noted the necessity for careful considerations as this technology advances.
The researchers view this AI tool as an emerging resource for medical professionals that could complement existing diagnostic methods rather than replace them. Although further testing is required, the study signifies a significant step forward in harnessing AI for health assessments.
