AI-guided personalized predictions on myopia progression and interventions.

Area of Science:

• Ophthalmology
• Artificial Intelligence
• Genetics

Background:

• Myopia is a significant global health issue, particularly in children and adolescents.
• Accurate prediction of myopia progression is crucial for effective management.
• Current management strategies lack personalization and precise outcome forecasting.

Purpose of the Study:

• To develop and validate a Transformer-based AI model, the Myopia Progression Predictive Model (MPPM), for predicting myopia progression.
• To enable personalized myopia management through accurate forecasting of both natural progression and treatment effects.
• To evaluate the efficacy of different myopia interventions using an AI-driven approach.

Main Methods:

• Developed a two-module AI model (MPPM): Natural Progression Module (NPM) and Intervention Progression Module (IPM).
• Trained NPM on over 1.1 million refractive records from children and adolescents for predicting spherical equivalent (SE) and axial length (AL).
• Trained IPM on intervention cohorts (atropine, orthokeratology, defocus spectacles, RLRL therapy) using a Transformer-based causal machine learning framework.

Main Results:

• NPM achieved high predictive accuracy for SE (R²=0.94) and AL (R²=0.91) over 10 years.
• IPM accurately predicted myopia changes under interventions (SE: R²>0.88; AL: R²>0.80).
• Repeated low-level red light (RLRL) therapy showed potential for myopia reversal, while other interventions slowed progression.

Conclusions:

• The MPPM demonstrates strong potential as an AI-driven platform for personalized pediatric myopia management.
• The model enables accurate prediction of myopia progression under various conditions and interventions.
• AI-powered predictive modeling can optimize myopia management strategies for individual patients.