AI Tool Analyzes Facial Photos to Assess Patient Health

US researchers are developing artificial intelligence tools that can assess a patient's health and estimate their remaining lifespan using just a photograph of their face. The AI systems, known as 'FaceSurvival' and 'FaceAge,' analyze specific facial features to quantify physical reserve and predict biological age, according to a study led by Dr. Raymond Mak from Mass General Brigham in Boston.

The 'FaceSurvival' tool focuses on the areas under the eyes and around the bridge of the nose, which are identified as vital for estimating how long a patient might live. This metric has shown applicability across various cancer types, including those that have metastasized. The patients in the study, from the US and the Netherlands, spanned a wide age range, from children to individuals in their 90s, with many in their 50s and 60s.

These AI tools are intended to assist physicians in their work, not replace their judgment. "As the population ages, physicians need better ways to estimate physical reserve and frailty before recommending intensive treatments such as radiation, surgery or systemic therapy," Mak explained. The goal is to make facial photograph-based health assessment more quantitative, consistent, and scalable to support personalized treatment decisions.

Meanwhile, the 'FaceAge' tool evaluates biological age, or how old a person appears, by examining the skin below the nose, near the mouth, and on the forehead, temples, and cheeks. Factors like heavy alcohol use, drug use, and smoking can accelerate aging, which this tool aims to identify. The system has demonstrated consistency across real-world photos despite variations in pose, lighting, and facial expression, and even showed stability after some plastic surgery procedures like rhinoplasty.

However, the researchers emphasize that these tools are still under development and require significantly more data, particularly from diverse ethnic groups, to validate their effectiveness. Professor Miroslaw Bober, an expert in video processing and machine learning at the University of Surrey, described the work as a "promising proof of concept" but cautioned that it is not yet ready for clinical use. He stressed the need for further testing against physicians' judgments and across different patient populations, adding that "predicting an outcome is not the same as explaining it."

The research team is inviting public participation through their website, where individuals can upload facial photos to receive a personalized biological age estimate.