AI-Powered Detection of Diabetic Retinopathy in Ukraine

Abstract

Purpose: To evaluate the potential of artifiial intelligence in real-world population screening to detect stages of diabetic retinopathy, identify patients with previously undiagnosed diabetes, and establish and optimise patient pathways. Materials and methods: This prospective, multicentre, open-label observational study evaluated the feasibility of automated analysis of colour fundus images to identify individuals with suspected undiagnosed diabetes mellitus (DM) by detecting signs of diabetic retinopathy in a real-world screening program. It was conducted at primary healthcare facilities in collaboration with The Filatov Institute of Eye Diseases and Tissue Therapy of NAMS of Ukraine and Oftacentro SA, Lugano-Paradiso, Switzerland. Patients were not preselected based on diabetes status, allowing assessment of fundus examination as a tool for opportunistic detection of systemic metabolic disease. Results: A total of 14,882 patients were included in the population-based screening and divided into two groups: 3 256 patients with established diabetes mellitus (used to assess detection of diabetic retinopathy in a confimed population), 11626 individuals without diagnosed diabetes but with risk factors (the key group for evaluating opportunistic detection). Fundus image analysis identifid signs of diabetic retinopathy in 622 patients (19.1 %) in DM group and 654 (5.6 %) in risk of DM group. In DM group disease staging showed: mild non-proliferative retinopathy in 58.5 % patients, moderate in 18.5 %, severe in 10 %, and proliferative retinopathy in 13 % of patients with DM. In risk of DM group disease staging showed: mild non-proliferative retinopathy in 87 % patients, moderate in 9 %, severe in 3 %, and proliferative retinopathy in 1 % of patients. Conclusions: Artifiial intelligence demonstrated high effiacy in populationbased screening for diabetic retinopathy. The screening revealed a signifiant prevalence of diabetic retinopathy among the population. Among the screened individuals with risk of DM, signs of diabetic retinopathy were detected in 5.6 % of patients. These fidings highlight the importance of early detection and large-scale screening programs. AI algorithms enable effective identifiation of at-risk patients, including those without a confimed diabetes diagnosis, creating new opportunities for early diagnosis and prevention of severe complications.