IMPROVING DIABETIC RETINOPATHY SCREENING USING ARTIFICIAL INTELLIGENCE TECHNOLOGY

Purpose – to evaluate the diagnostic effectiveness of the “Retina AI” artificial intelligence system in detecting of diabetic retinopathy (DR) and its symptoms. Material and methods The study was conducted on the basis of the central multidisciplinary polyclinics of the Syrdarya region during the 1st quarter of 2025. The study consistently included 250 patients with an established diagnosis of diabetes mellitus (DM) (type 1 or type 2 diabetes), the duration of the disease is ≥ 5 years. Results The “Retina AI” system demonstrated statistically significantly higher sensitivity (94.7% vs 72.8%) and accuracy (93.5% vs 77.5%) compared to primary care physicians. The difference in specificity also reached statistical significance (91.3% vs 85.0%). The Kappa coefficient of agreement of the Retina AI system with experts was 0.88, which indicates a "very good" agreement. Conclusion The “Retina AI” artificial intelligence system has proven to be highly effective in screening DR among patients in the Syrdarya region, surpassing the assessment of primary care physicians in terms of basic diagnostic accuracy. Key words: diabetic retinopathy, artificial intelligence, diagnostics, screening, neural networks