Synthetic Intelligence in Fundus Images for Type 2 Diabetes: A Scoping Overview of Systemic Biomarkers and Multi-Organ Threat Prediction

Abstract

Type 2 diabetes mellitus (T2DM) is related to multi-organ problems, together with cardiovascular and renal illness. Fundus images supplies a non-invasive window into systemic microvascular well being, and synthetic intelligence (AI) has enabled extraction of retinal biomarkers for systemic danger prediction past diabetic reti-nopathy detection. We carried out a methodologically structured scoping assessment following PRISMA-ScR steering to map AI functions in retinal imaging for multi-organ danger stratification in T2DM. Studies utilizing machine studying or deep studying fashions to foretell cardiovascular, renal, or cerebrovascular outcomes had been recognized and characterised. Rather than quantitative pooling, we examined modeling methods, validation approaches, efficiency reporting, and translational readiness throughout heterogeneous research designs. AI fashions regularly demonstrated promising discrimination; nevertheless, substantial heterogeneity was noticed in cohort measurement, consequence definitions, imaging modalities, and validation methods. External validation was restricted, calibration was inconsistently assessed, and subgroup analyses ad-dressing equity and device-related area shift had been hardly ever reported. Most research emphasised discrimination metrics with out complete analysis of scientific utility.Retinal AI exhibits potential for scalable systemic danger surveillance in T2DM, however rigorous exterior validation, standardized reporting, and potential implementation research are required to allow protected and equitable scientific translation.