AI Predicts Cancer Survival from Patient Photos
Researchers at Mass General Brigham have created AI tools that use facial images to predict cancer survival rates and biological age.
Researchers at Mass General Brigham in Boston have developed artificial intelligence tools capable of predicting cancer survival rates and biological age using merely a photograph of a patient’s face. This approach was documented in a study published in The Lancet Digital Health.
The model, known as FAHR-Face, utilizes over 40 million facial images to make its assessments. Two AI models were created: FAHR-FaceAge estimates biological age, while FAHR-FaceSurvival predicts survival likelihood for cancer patients.
Dr. Raymond Mak, a radiation oncologist involved in the research, expressed surprise at how these algorithms could detect mortality risk from facial characteristics. He stated, “I was pretty surprised to find out that we could pick up signs of mortality risk from just a facial photograph.”
The algorithms focus on different facial regions. For biological age, they analyze features such as the temples and forehead, while for survival predictions, areas like the bridge of the nose and under the eyes are evaluated. This research aligns with observations made by Hippocrates centuries ago concerning facial signs and health.
In a study involving 122 cancer patients, the AI model accurately predicted that over half of those flagged as high-risk would die within 90 days. In contrast, only about 2% of those assessed as low-risk had the same outcome. This indicates a potential for the AI to significantly aid in treatment decisions.
The FaceAge tool is now publicly accessible, allowing users to upload selfies to receive estimates of their biological age in under ten minutes. Users must consent before participating, as their images are used for research purposes. Dr. Mak noted that assessing biological age could help identify patients who may benefit from further testing or support.
Dr. Wilbur Lam, a professor at Emory University, commented on the study’s findings, emphasizing the need for clinical evaluation to determine its utility across diverse populations and specific health conditions.
Since its launch, over 10,000 people have signed up for the FaceAge study, which also examines the influence of lifestyle choices on perceived age. Preliminary results indicate that factors like smoking and heavy alcohol use contribute to an older biological appearance.
Dr. Hugo Aerts, another researcher in the project, stated, “To our knowledge, this is the first demonstration that an AI model can directly predict cancer survival from a simple photograph.” However, he also cautioned about the variability in performance across different cancer types, indicating that further research is required in this area.
