3D facial imaging: a novel approach for metabolic abnormalities risk profiling

The human face harbors a rich tapestry of complex phenotypic information spanning genetic, environmental, and physiological dimensions. While facial images excel in diagnosing genetic diseases, their untapped potential for predicting metabolic health presents an intriguing prospect. Metabolic Syndrome (MetS), marked by a constellation of metabolic abnormalities, poses a significant risk for various chronic diseases. Utilizing Face-Wide Association Studies (FaWAS) on a discovery cohort of 2,621 Chinese individuals and a replication cohort of 2,188 Chinese individuals, we investigated the associations between facial features and MetS and its related conditions. Our findings highlight half of our investigated facial features strongly correlated with MetS risk, such as a slender forehead, a broader and shorter jawline, and fuller features around the temples-eye-cheek region, with notable genetic correlations (0.55-0.58) and influences from environmental factors like age, urban residency, and educational level. The developed face-based prediction model demonstrated significant predictive robustness, achieving an AUC of up to 0.87 for MetS and 0.89 for obesity in external validations, surpassing traditional 2D imaging techniques. Our model also aids in identifying subtypes within healthy populations, with a 2.07 to 2.40-fold increased risk of developing different metabolic disorders within the next five years. This paves the way for precise risk stratification of individuals who are "at risk". Integrating 3D facial imaging for metabolic health predictions, our research introduces an innovative, non-invasive framework for health assessment and subtype identification, demonstrating high potential in personalized medicine and health monitoring.