Advances in the application of artificial intelligence in the diagnosis and treatment of diabetic retinopathy
10.3980/j.issn.1672-5123.2026.9.14
- VernacularTitle:人工智能在糖尿病视网膜病变诊疗中的应用进展
- Author:
Feiyu LI
1
;
GERILETU
1
Author Information
1. Inner Mongolia Medical University, Hohhot 010020, Inner Mongolia Autonomous Region, China; Department of Ophthalmology, Inner Mongolia Autonomous Region People's Hospital, Hohhot 010017, Inner Mongolia Autonomous Region, China
- Publication Type:Journal Article
- Keywords:
artificial intelligence;
diabetic retinopathy;
diabetic retinopathy screening;
machine learning;
deep learning
- From:
International Eye Science
2026;26(9):1585-1591
- CountryChina
- Language:Chinese
-
Abstract:
Diabetic retinopathy(DR)represents the most prevalent microvascular complication of diabetes mellitus(DM)and has emerged as the primary cause of blindness among working-age individuals. Timely screening and intervention are essential for preventing and treating vision loss. Recently, artificial intelligence(AI)technology has been increasingly integrated into ophthalmology, exhibiting considerable promise in the screening, diagnosis, prognosis prediction, and personalized management of DR, thus becoming a focal point of research and practice. This article systematically reviews the advancements in AI concerning the comprehensive management of DR. In the domains of screening and diagnosis, deep learning-based systems have achieved efficient and automated lesion detection and grading, demonstrating performance that is comparable to, or even exceeds, that of ophthalmologists. These systems can be deployed in grassroots and remote areas through various modalities. In the realm of prognosis prediction, AI can assess the risk of disease progression by analyzing fundus images and clinical data, thereby establishing a foundation for personalized follow-up. Regarding treatment and management, AI can aid in the development of laser treatment plans, forecast drug efficacy, and facilitate the creation of intelligent management platforms that integrate data from multiple sources. This article seeks to summarize the practical application models, key methodologies, and clinical validation of AI technology throughout the entire continuum of “screening-diagnosis-prediction-individualized management” of DR. Additionally, it examines the current technical, ethical, and promotional challenges while anticipating future developmental directions, with the aim of providing valuable insights for further research and application of AI in the prevention and treatment of DR.
