@article{AES9226,
author = {Sanil Joseph and Yueye Wang and Mingguang He},
title = {Artificial intelligence in ophthalmology: current applications, challenges, and future directions—a narrative review},
journal = {Annals of Eye Science},
volume = {11},
number = {0},
year = {2026},
keywords = {},
abstract = {Background and Objective: Ophthalmology is particularly well suited to artificial intelligence (AI) applications due to its reliance on imaging and quantitative diagnostics. Over the past decade, advances in machine learning (ML) and deep learning (DL) have driven rapid growth in AI-based tools for detecting and managing major causes of vision impairment. While algorithmic performance has been widely reported, there remains a need for integrative reviews that critically examine real-world implementation, ethical considerations, and health system implications. This narrative review aims to synthesize current developments in AI for ophthalmology, evaluate disease-specific applications and impact, identify key implementation challenges, and outline future directions for equitable and responsible integration.Methods: This narrative review draws on peer-reviewed literature, regulatory reports, and real-world implementation studies relevant to AI in ophthalmology. A literature search was performed using PubMed, Scopus, and Web of Science to identify relevant peer-reviewed articles published in English between January 2013 and March 2025. Evidence was selected based on clinical relevance, maturity of application, and implications for health systems rather than through a formal systematic search or meta-analysis. Key themes were identified across disease applications, implementation experiences, ethical and equity considerations, and emerging technological and regulatory trends.Key Content and Findings: AI systems have demonstrated expert-level performance in detecting diabetic retinopathy (DR), age-related macular degeneration (AMD), and glaucoma, with expanding applications in retinopathy of prematurity (ROP), cataract, and refractive error. Regulatory approvals of autonomous AI systems and real-world validation studies indicate growing clinical readiness. Beyond diagnostics, AI has potential to improve access to screening, reduce diagnostic delays, and enhance efficiency, particularly in underserved settings. However, persistent challenges include dataset bias, limited explainability, workflow integration barriers, reimbursement constraints, regulatory uncertainty, and risks of technology-driven health inequities.Conclusions: AI is transitioning from experimental innovation to a meaningful adjunct in ophthalmic care. This review provides a conceptual and practical framework to guide future research, policy development, and responsible clinical adoption. Addressing ethical governance, equity, and system-level integration alongside technical advancement will be essential to ensure that AI contributes to safe, effective, and equitable eye care globally.},
issn = {2520-4122}, url = {https://aes.amegroups.org/article/view/9226}
}