Explainable Retinal Imaging for Prediction of Multi-Organ Dysfunction in Type 2 Diabetes
Researchers have developed new machine learning frameworks to predict multi-organ dysfunction in Type 2 Diabetes patients. One study utilized routine laboratory biomarkers and gradient boosting models, achieving near-perfect discrimination (AUC = 1.000) by identifying hyperglycemia, renal impairment, dyslipidemia, and inflammation as key risk factors. A separate pilot study employed explainable multi-task deep learning on retinal images, revealing that retinal vessels encode signals associated with systemic abnormalities, particularly microvascular damage, though predictive performance varied by task. AI
IMPACT These studies demonstrate AI's potential to improve risk stratification and precision medicine in diabetes care by identifying key predictive factors from diverse data sources.