The PEARL-DGS Formula: The Development of an Open-source Machine Learning-based Thick IOL Calculation Formula

Area of Science:

• Ophthalmology
• Biomedical Engineering
• Optical Engineering

Background:

• Accurate intraocular lens (IOL) power calculation is crucial for achieving desired refractive outcomes after cataract surgery.
• Existing IOL formulas vary in complexity and performance, necessitating the development of new, optimized calculation methods.
• The sum-of-segments approach offers a framework for developing advanced IOL formulas.

Purpose of the Study:

• To describe an open-source, reproducible, step-by-step methodology for designing sum-of-segments thick intraocular lens (IOL) calculation formulas.
• To develop and evaluate a novel IOL calculation formula using the proposed methodology.

Main Methods:

• A retrospective, multicenter case series involving 4242 eyes implanted with Finevision IOLs was used to devise the formula design process.
• A separate test set of 677 eyes from the same center and an independent data set of 262 eyes were used for formula evaluation.
• The PEARL-DGS formula was developed and compared against other established IOL calculation formulas.

Main Results:

• The PEARL-DGS formula achieved the lowest standard deviation of prediction errors on both the primary (±0.382 D) and test (±0.269 D) data sets.
• Its performance was superior to other evaluated formulas including K6, Olsen, EVO 2.0, RBF 3.0, and BUII.
• The developed methodology yielded an accuracy comparable to state-of-the-art IOL formulas.

Conclusions:

• The open-source methodology provides a reproducible framework for developing and validating IOL calculation formulas.
• The PEARL-DGS formula, derived using this method, demonstrates high accuracy for IOL power prediction.
• The availability of open-source tools facilitates further research and customization for diverse clinical scenarios.