Related Experiment Video
Updated: Jul 12, 2026
05:14
Comparison of Agreement and Accuracy using Binocular Wavefront Optometer with Autorefractor and Phoropter
Published on: September 16, 2025
Machine Learning Prediction Models for Classifying Myopic Versus Hyperopic Ablation Patterns in Eyes With Previous
Carel Bustos-Ampudia1, Jair Maldonado-Aparicio1, Isabel De La Fuente-Batta2
1Anterior Segment Department.
Journal of Refractive Surgery (Thorofare, N.J. : 1995)
|July 10, 2026
Summary
Machine learning models accurately classify myopic and hyperopic ablation patterns after refractive surgery using corneal topography. This aids intraocular lens power calculation in subsequent cataract surgery when records are missing.
Area of Science:
- Ophthalmology
- Artificial Intelligence
- Corneal Topography
Background:
- Corneal topography and tomography are crucial for assessing refractive surgery outcomes.
- Accurate classification of prior ablation patterns is essential for safe and effective cataract surgery planning.
- Machine learning offers potential for analyzing complex corneal data.
Purpose of the Study:
- To develop and compare machine learning models for classifying myopic versus hyperopic ablation patterns.
- To utilize corneal topographic and tomographic parameters for this classification.
- To aid intraocular lens power calculation in eyes with prior refractive surgery.
Main Methods:
- Retrospective observational study analyzing 72 eyes with prior refractive surgery.
- Machine learning models (Random Forest, Decision Tree, Logistic Regression) were trained and tested.
- Corneal parameters including P/A ratio and spherical aberration (Z 0 4) were key predictors.
Main Results:
- The Random Forest model achieved the highest performance with an AUC of 0.933.
- Posterior-to-anterior curvature ratio (P/A ratio) and Z 0 4 showed strong correlation (r = -0.793).
- Decision Tree model identified cut-off values (P/A ratio: 82.7%, Z 0 4: -0.144) for classification.
Conclusions:
- Machine learning models effectively distinguish myopic from hyperopic ablation patterns.
- The Random Forest model demonstrated superior discriminative ability.
- These models can assist in intraocular lens power calculation for cataract surgery in eyes with prior refractive surgery, especially when records are unavailable.
