Prediction of Seven Artificial Intelligence-Based Intraocular Lens Power Calculation Formulas in Medium-Long...

Area of Science:

• Ophthalmology
• Biomedical Engineering
• Artificial Intelligence in Healthcare

Background:

• Accurate intraocular lens (IOL) power calculation is crucial for achieving optimal visual outcomes after cataract surgery.
• Traditional formulas may have limitations in predicting refractive outcomes, especially in specific eye types.
• Advancements in artificial intelligence (AI) offer new possibilities for enhancing IOL power calculation accuracy.

Purpose of the Study:

• To evaluate and compare the accuracy of seven AI-based IOL power calculation formulas.
• To assess formula performance using metrics like root-mean-square absolute error (RMSAE), median absolute error (MedAE), and prediction error (PE) within ±0.5 D.
• To identify the most accurate AI formula for medium-long Caucasian eyes.

Main Methods:

• Retrospective review of biometric and refractive data from 214 Caucasian patients undergoing cataract surgery.
• Application of multiple AI-based IOL power calculation formulas including Hill-RBF 3.0, Ladas Super Formula AI (LSF AI), and PEARL-DGS.
• Analysis of prediction errors at various diopter thresholds (±0.25 D, ±0.50 D, ±0.75 D, ±1.00 D).

Main Results:

• The Hill-RBF 3.0 formula achieved the lowest RMSAE (0.368) and MedAE (0.200).
• Hill-RBF 3.0, LSF AI, and PEARL-DGS showed the highest percentage of eyes with PE within ±0.50 D (86.45%, 85.51%, and 85.05%, respectively).
• All evaluated AI formulas demonstrated good performance in IOL power calculation for the studied cohort.

Conclusions:

• The Hill-RBF 3.0 formula is highly accurate for IOL power calculation in medium-long Caucasian eyes.
• AI-based formulas represent a significant advancement in achieving precise refractive outcomes in cataract surgery.
• Further research may explore the efficacy of these AI formulas across diverse patient populations and eye characteristics.