Hill-RBF: Improving IOL Power Selection by Artificial Intelligence

The Hill-RBF method is an artificial intelligence model-based approach designed to calculate intraocular lens (IOL) power. The calculation algorithm is completely data-driven and therefore avoids any bias introduced by measurements in a theoretical optics-based method. A selection process identified the best-predicting variables for the model: axial length, mean keratometry, anterior chamber depth, postoperative spherical equivalent and the implanted IOL power. Mathematical boundaries for a 90% precision of ±0.50 D have been defined and cases out-of-bounds are identified. The current version 3.0. has shown a weighted ±0.50 D accuracy of 91.2% with a very few cases out-of-bounds even in unusual eyes.