Related Experiment Video
Updated: Jul 12, 2026
05:46
Correction of Presbyopia by Monocular Bi-Aspheric Ablation Profile
Published on: September 20, 2024
Machine Learning-Based Prediction of LASIK Console Inputs for Aspheric Planning (Q-factor, Defocus, Astigmatism): A
1Laser Vision Bourgogne, Chalon sur Saône, France.
Journal of Refractive Surgery (Thorofare, N.J. : 1995)
|July 10, 2026
Summary
Supervised nonlinear models accurately predict laser refractive surgery inputs like defocus and Q-factor, outperforming linear models for better refractive control and potential vision improvements.
Area of Science:
- Ophthalmology
- Laser Surgery
- Computational Modeling
Background:
- Accurate planning is crucial for laser refractive surgery outcomes.
- Current methods for planning aspheric laser refractive surgery can be improved.
Purpose of the Study:
- To frame aspheric laser refractive planning as a supervised prediction task for console-programmable inputs.
- To benchmark the performance of various regression models for this task.
Main Methods:
- A retrospective dataset of 2,448 treatments was analyzed.
- Multi-output linear and nonlinear regression models were trained and compared.
- Actuator-response analyses and external validation were performed.
Main Results:
- Linear models performed well for Defocus and Astigmatism but poorly for Q-factor.
- Nonlinear models demonstrated improved accuracy and calibration for Q-factor prediction.
- External validation confirmed the generalizability of the best nonlinear model.
Conclusions:
- Supervised nonlinear models offer a promising approach for precise control of refractive targets and asphericity.
- This method could lead to tissue sparing, enhanced contrast, and improved near vision.
- Further clinical evaluation with patient-reported outcomes is recommended.
