Advances in Artificial Intelligence for Diabetes Management: Predicting Hospital Readmissions and Detecting Diabetic Retinopathy

Diabetes mellitus remains a global health challenge, with increasing prevalence and complications leading to frequent hospital readmissions and vision-threatening conditions like diabetic retinopathy (DR). This review synthesizes recent developments in artificial intelligence (AI), particularly machine learning (ML) and deep learning (DL) models, for predicting ۳۰-day hospital readmissions in diabetic patients and classifying DR using vision transformers (ViTs). Drawing from studies between ۲۰۱۷ and ۲۰۲۴, we highlight hybrid models integrating support vector machines (SVM), genetic algorithms, and transformers, which achieve superior accuracy in prediction tasks. Updated statistics from the IDF Diabetes Atlas ۲۰۲۵ indicate ۵۸۹ million adults living with diabetes in ۲۰۲۴, underscoring the urgency for AI-driven interventions. Challenges such as data integration and computational costs are discussed, with recommendations for future multimodal and personalized AI systems.

کلیدواژه ها:

Diabetes mellitus ، Hospital readmission ، Diabetic retinopathy ، Machine learning ، Deep learning ، Vision transformers ، Predictive modeling ، Artificial intelligence