Generative and Image-Processing Artificial Intelligence in Ophthalmic Screening for Early Detection of Chronic Diseases and Improving Patient Education

Abstract

Background: Diabetes and hypertension represent a growing public health burden worldwide. Their common ocular complications – diabetic and hypertensive retinopathy – are leading causes of preventable vision loss among the working-age population. Early detection – via regular fundus examinations – can significantly reduce the risk of severe visual impairment. Widespread implementation of screening faces challenges (shortage of specialists, low participation).

Objective: The study reviews the role of artificial intelligence, particularly modern deep learning-based image processing and generative tools, in the early detection of chronic diseases (highlighting diabetes and hypertension) and the improvement of patient education.

Methods: A narrative literature review was conducted using PubMed-indexed clinical studies, meta-analyses, and regulatory documents.

Results: Autonomous algorithms approved by the FDA are capable of detecting referable lesions in diabetic retinopathy screening without ophthalmologist involvement (sensitivity 87–96%, specificity 85–94%). Generative models have also emerged in patient education, but accuracy issues limit their use.

Conclusions: Artificial intelligence-based ophthalmic screening systems can significantly increase the efficiency and accessibility of screening. The European Union regulatory environment imposes strict requirements to ensure safe operation.

Keywords: artificial intelligence; ophthalmic screening; diabetic retinopathy; generative models; patient education; health informatics