Artificial Intelligence in Myopia Management: Risk Prediction, Monitoring, and Clinical Decision Support
10.52725/aocl.2026.25.2.47
- Author:
Jinho JEONG
1
Author Information
1. Department of Ophthalmology, Jeju National University Hospital, Jeju National University School of Medicine, Jeju, Korea
- Publication Type:Review Article
- From:
Annals of Optometry and Contact Lens
2026;25(2):47-58
- CountryRepublic of Korea
- Language:English
-
Abstract:
Myopia has reached epidemic proportions globally, with a prevalence exceeding 80–90% among young adults in East Asia. This public health crisis has catalyzed the development of artificial intelligence (AI) applications to transform myopia management from reactive correction to predictive precision prevention. This review synthesizes the advances in AI applications across the myopia management continuum, encompassing risk prediction, automated monitoring, and clinical decision support systems (CDSS). Emphasis is placed on contributions from high-prevalence regions. We conducted a comprehensive PubMed literature search for studies published between 2018 and 2025 using the terms “myopia,” “artificial intelligence,” “machine learning,” “deep learning,” and “clinical decision support.” These studies demonstrated that deep learning systems achieve areas under the receiver operating characteristic curve exceeding 0.95 for pathologic myopia detection and up to 0.97–0.98 for 5-year high myopia risk prediction in early longitudinal cohorts. Self-supervised foundation models (RETFound) enable few-shot learning for rare complications with limited labeled data. Vision transformers outperform conventional convolutional neural networks in detecting subtle lesions such as lacquer cracks. Time-series transformers (Informer, TimesNet) demonstrate 8–12% improvements in long-term refractive prediction relative to long-short term memory architectures. Federated learning frameworks enable privacy-preserving multi-institutional collaboration. AI-guided interventions reduce the incidence of high myopia by 17.8% in simulated trials. AI has achieved expert-level performance in myopia detection, risk prediction, and progression forecasting. The convergence of foundation models, federated learning, and closed-loop management ecosystems signal a pivotal era for precision myopia care; however, real-world clinical trials and cross-ethnic validation are essential before widespread deployment.
