Machine Learning Approaches in High Myopia: Systematic Review and Meta-Analysis.

Area of Science:

• Ophthalmology
• Medical Artificial Intelligence
• Biomedical Data Science

Background:

• Machine learning (ML) is increasingly utilized in clinical practice for disease diagnosis and prognosis.
• ML applications in ophthalmology focus on diagnosing and predicting high myopia and associated conditions.
• Evidence-based validation is crucial for ML-based diagnostic tools in ophthalmology.

Purpose of the Study:

• To evaluate the performance of ML methods in detecting high myopia and pathologic myopia.
• To provide evidence for developing and refining intelligent diagnostic and predictive tools in ophthalmology.
• To assess the accuracy of ML in diagnosing high myopia-associated glaucoma.

Main Methods:

• Systematic literature search of PubMed, Cochrane, Embase, and Web of Science up to September 2023.
• Meta-analysis using a bivariate mixed-effects model to assess diagnostic accuracy.
• Risk of bias assessment and subgroup analyses based on ML targets and methods.

Main Results:

• ML demonstrated high accuracy in diagnosing pathologic myopia (SROC 0.97, sensitivity 0.91, specificity 0.95).
• Deep learning (DL) outperformed conventional ML in diagnosing pathologic myopia.
• ML showed high accuracy in diagnosing and predicting high myopia (SROC 0.98) and diagnosing high myopia-associated glaucoma (SROC 0.96).

Conclusions:

• ML exhibits promising accuracy for diagnosing high myopia and pathologic myopia.
• ML shows potential for predicting future high myopia risk.
• DL is a viable method for intelligent image processing, supporting the development of intelligent examination tools for underserved areas.