AI Tool FAHR-Face Predicts Biological Age and Cancer Survival
Mass General Brigham researchers unveil FAHR-Face, an AI model that estimates biological age and predicts cancer survival from facial images.
FAHR-Face utilizes over 40 million images to assess health indicators.
The model offers improved accuracy in estimating biological age and predicting cancer survival.
Lifestyle factors, including smoking and alcohol use, significantly influence facial aging.
Researchers at Mass General Brigham have developed an advanced AI tool named FAHR-Face, designed to analyze facial photographs for health assessments. This innovative model, trained on a dataset of over 40 million images, is capable of estimating a person's biological age and predicting cancer survival outcomes based solely on facial features. The findings from their study, published in Lancet Digital Health, reveal that the discrepancies between biological and chronological age are linked to patient survival rates.
The development of FAHR-Face includes two key components: FAHR-FaceAge, which accurately estimates biological age, and FAHR-FaceSurvival, which predicts survival rates for cancer patients. According to Dr. Raymond Mak, a radiation oncologist involved in the project, the goal is to enhance the precision and scalability of health assessments derived from facial images, thereby aiding personalized treatment decisions without undermining clinical judgment.
The technical innovation of FAHR-Face lies in its foundation model approach, which allows it to learn general facial representations that can be adapted for various health outcomes. While both FAHR-FaceAge and FAHR-FaceSurvival stem from the same model, they provide distinct insights into patient health, with their predictions showing only weak correlation. This independence enhances the overall accuracy of survival predictions when both tools are utilized together.
The implications of this research extend beyond individual assessments. The study highlights how lifestyle choices, such as smoking and alcohol consumption, can accelerate biological aging, with current smokers appearing significantly older than non-smokers. Furthermore, the model's robustness across different photo conditions, including lighting and facial expressions, suggests its practical applicability in real-world scenarios.
Looking ahead, the research team is expanding their efforts through the FaceAge Research Study, which invites public participation. Individuals can upload their facial photographs to receive personalized assessments while contributing valuable data for future AI health research. This initiative aims to refine AI-driven measures of facial aging and health across diverse populations, potentially revolutionizing how we approach health assessments in clinical settings.
