AI-based oculomics for trajectory-driven risk stratification of pathologic myopia in paediatric high myopia

Area of Science:

• Ophthalmology
• Medical Imaging
• Artificial Intelligence

Background:

• High myopia (HM) in children poses a significant risk for developing pathologic myopia (PM), a leading cause of vision impairment.
• Early identification of children at risk for PM is crucial for timely intervention and prevention of sight-threatening complications.

Purpose of the Study:

• To identify early oculomic biomarkers that predict the onset of pathologic myopia (PM) in children with high myopia (HM).
• To develop an artificial intelligence (AI)-based model for individualized risk stratification of PM in pediatric HM patients.

Main Methods:

• A prospective longitudinal study of 375 children with HM followed for a median of 15 years.
• Deep learning-based retinal vascular phenotyping combined with ocular biometry and OCT metrics.
• Multivariable models and AI were used to identify predictors and assess predictive performance (AUROC).

Main Results:

• Pathologic myopia (PM) developed in 17.1% of children.
• Key predictors included reduced retinal vessel density, fractal dimension, narrower arterioles/venules, thinner choroid, faster axial elongation, and accelerated choroidal thinning.
• The AI oculomic model demonstrated excellent predictive discrimination (AUROC 0.96-0.98).

Conclusions:

• Early signs of pathologic myopia (PM) risk in pediatric high myopia (HM) can be detected through retinal microvascular changes and ocular growth patterns.
• AI-powered oculomic profiling provides a scalable method for early risk stratification.
• This approach supports targeted surveillance and timely prevention strategies for sight-threatening conditions.