Related Experiment Video
Updated: Jun 17, 2025
07:20
Inducement and Evaluation of a Murine Model of Experimental Myopia
Published on: January 22, 2019
9.8K
Machine Learning Models for Predicting Cycloplegic Refractive Error and Myopia Status Based on Non-Cycloplegic Data
Bole Ying1, Rajat S Chandra2, Jianyong Wang3
1Lower Merion High School, Ardmore, PA, USA.
Translational Vision Science & Technology
|August 9, 2024
Summary
Machine learning models accurately predict cycloplegic refractive error and myopia status using noncycloplegic data. This offers a valuable tool for large-scale eye health studies when direct measurement is not feasible.
Area of Science:
- Ophthalmology
- Data Science
- Biomedical Engineering
Background:
- Accurate refractive error assessment is crucial for understanding myopia prevalence.
- Cycloplegic refraction is the gold standard but can be challenging in large studies.
- Noncycloplegic data offers a potential alternative for refractive error prediction.
Purpose of the Study:
- To develop and validate machine learning (ML) models.
- Predict cycloplegic refractive error and myopia status.
- Utilize noncycloplegic refractive error and ocular biometric data.
Main Methods:
- Cross-sectional study of 5-18 year olds.
- Collected biometry and autorefraction data pre- and post-cycloplegia.
- Trained and validated ML models (XGBoost, Random Forest) on independent datasets.
Main Results:
- ML models achieved high accuracy in predicting cycloplegic spherical equivalent refraction (SER) (R2=0.913-0.935, MAE=0.393-0.480 D).
- Models demonstrated high performance in predicting myopia status (AUC=0.984-0.987).
- Best models showed high sensitivity and specificity for both SER and myopia prediction.
Conclusions:
- ML models effectively predict cycloplegic refractive error and myopia using noncycloplegic data.
- ML provides a viable tool for estimating refractive error and myopia prevalence in epidemiological research.
- This approach enhances the feasibility of large-scale refractive error studies.
