Detecting diabetic retinopathy in individuals with objective indicators suggestive of possible undiagnosed diabetes mellitus using an artificial intelligence-based software platform

Abstract

Objectives: To examine the potential for the detection of diabetic retin- opathy ( DR) using the artificial intelligence ( AI)-based software platform in patients with risk factors of diabetes mellitus. Methods: This was an open-label, prospective, observational case-con- trol study for the detection of DR in patients with risk factors using an AI- based software platform. The study was conducted at the sites of health- care facilities in Chernivtsi region, Lviv region, Kyiv region and Kyiv-city. 5655 individuals (11.310 eyes) were involved in the study. All fundus imag- es were analyzed using the artificial intelligence ( AI)-based software plat- form Retina- AI CheckEye©. Results: All patients were divided into two groups. The first group consist- ed of 1841 patients, who had diabetes mellitus, and the second group had 3814 patients with risk factors of diabetes mellitus. Using the AI- based software platform, in the first group signs of D R were detected in 366 dia- betics (19.88 % of the diabetics). In the second group signs of D R were de- tected in 33 individuals, who had not diabetes mellitus (0.87 % of patients with risk factors). The diagnoses of DR were verified by expert ophthal- mologists in each patient. Conclusions: AI- based software platform, Retina- AI CheckEye© system helps to diagnose the presence of diabetic retinopathy not only in pa- tients with diabetes, but also in patients with risk factors, and can be used for mass screening of the disease. Health care system need to pay more at- tention for patients with risk factors of diabetes mellitus and perform pro- phylactic examinations of the eye fundus.