Diabetic Retinopathy Screening with Artificial Intelligence: A Pivotal Experience in Italian Healthcare System – Preliminary Report

Abstract

Diabetic retinopathy is the main complication of diabetes. Its prevalence is 30% and if not diagnosed in time can cause serious damages to vision, up to blindness. Screening of retinopathy is a healthy procedure with the best ratio cost/effectiveness. If done in a massive way, it allows to detect initial lesions, and so preventing or delaying evolved forms. ASL TO 5 is a Local Health District of Turin Metropolitan Area with 310.315 residents. In 2019, ASL TO5 performed an integrated care pathway dedicated to diabetic retinopathy. The clinical pathway develops technological improvements, in order to make the diagnostic and therapeutic process more effective. New knowledge and technologies, such as non- mydriatic cameras and artificial intelligence (AI) software allow to recognize elementary lesions making a first diagnosis of presence or absence of retinopathy reducing the diagnostic burden on eye care specialists and time costs for patients. Dairet, an artificial intelligence system distributed by Meteda srl (Rome, Italy) powered by Retmarker SA, demonstrated accurate sensitivity in detecting both mild and moderate retinopathy. Sensitivity ratio was 91,6% for mild retinopathy and 100% for moderate retinopathy. More over specificity was acceptable with a low false positive rate (specificity ratio 82,6%).