Related Experiment Video
Updated: Jul 12, 2026
08:27
Quantification of Oculomotor Responses and Accommodation Through Instrumentation and Analysis Toolboxes
Published on: March 3, 2023
Binocular Non-Cycloplegic Ocular Parameters for Modeling Cycloplegic SER in Children
Ranran Chen1, Juan Zhang2, Jinming Lei3
1Department of Ophthalmology, Shanghai Fifth People's Hospital, Fudan University, Shanghai, China.
Translational Vision Science & Technology
|July 10, 2026
Summary
This study developed an interpretable binocular model to accurately predict children's refractive error, improving vision screening tools. The model accounts for inter-eye correlation, enhancing prediction accuracy for pediatric myopia management.
Area of Science:
- Ophthalmology
- Biomedical Engineering
- Data Science
Background:
- Accurate refractive error prediction in children is crucial for early detection and management of vision disorders like myopia.
- Existing methods may not fully account for the correlation between the two eyes, potentially limiting prediction accuracy.
Purpose of the Study:
- To evaluate an interpretable binocular modeling approach for predicting cycloplegic spherical equivalent refraction (SER) in children.
- To assess the model's performance while considering inter-eye correlation.
Main Methods:
- Collected clinical data from children aged 4-17 years across two centers.
- Developed a binocular approach incorporating inter-eye features and applied a linear mixed-effects model (LMM).
- Utilized SHapley Additive exPlanations (SHAP) for feature interpretability and validated models internally and externally.
Main Results:
- Optimal binocular models achieved high R² values (up to 0.969) and low Mean Absolute Error (MAE) (as low as 0.232 D) across age groups in internal validation.
- External validation showed comparable performance with R² up to 0.969 and MAE as low as 0.287 D.
- SHAP analysis identified key predictors varying by age, including axial length and non-cycloplegic SER.
Conclusions:
- The interpretable binocular modeling approach shows promise for enhancing refractive status prediction accuracy in children.
- This method can serve as a valuable auxiliary tool for large-scale vision screening programs.
- The study translates ocular biometric modeling into clinical practice for pediatric myopia risk assessment and screening.