A team in Guangzhou reported an externally validated seven-variable nomogram to predict LDL cholesterol elevation risk in community-dwelling patients with type 2 diabetes, using routine clinical data to identify longer-term lipid risk windows. In related diabetic-vision research, an international effort demonstrated that an optogenetic gene therapy combined with light-projecting goggles was safe and improved visual function in some patients with advanced retinitis pigmentosa, supporting feasibility for light-controlled vision restoration. On the diagnostics side, researchers in India reported a hybrid graph neural network approach for diabetic retinopathy detection that converts retinal images into graph structures and achieved up to 96.86% accuracy and near-perfect ROC-AUC performance in their evaluation dataset. Collectively, the work spans prevention modeling, therapeutic restoration, and improved screening—positioning AI and mechanism-based interventions as near-term levers for reducing burden in retinal disease.