Predictive Accuracy of the Hill-RBF 2.0 Formula in Pediatric Eyes: Comparison of 5 Intraocular Lens Formulas

Abstract

Purpose: To compare the predictive accuracy of the Hill-Radial Basis Function (Hill-RBF) 2.0 formula with the Barrett Universal II, Hoffer Q, SRK/T, and Holladay 1 formulas in pediatric eyes. Methods: In this observational study, 99 eyes of 70 children 4 to 18 years old with clinically significant congenital or developmental cataracts attending the pediatric ophthalmology clinic in Guru Nanak Eye Centre were included. Optical biometry was performed in all patients with the Lenstar LS-900 (Haag-Streit). Intraocular lens (IOL) power predicted by the Hill-RBF formula was selected. Mean absolute prediction error (MAE) at 8 weeks of follow-up was calculated for the Hill-RBF, Barrett Universal II, Hoffer Q, SRK/T, and Holladay 1 IOL power formulas. Percentages of eyes having residual refraction within ±0.50, ±0.50 to ±1.00, and greater than ±1.00 diopters (D) of target refraction were calculated. Results: The MAEs were 1.08 ± 1.00 D for the Hill-RBF, 1.24 ± 1.20 D for the Barrett Universal II, 1.25 ± 1.06 D for the Hoffer Q, 1.25 ± 1.10 D for the SRK/T, and 1.28 ± 1.01 D for the Holladay 1 formulas. The Hill-RBF formula had the lowest MAE, which was significantly lower than the Holladay 1 and Hoffer Q formulas. However, the difference in MAE between the Hill-RBF and the SRK/T and Barrett Universal II formulas was not statistically significant ( P > .05). The Hill-RBF group had the maximum percentage of eyes with residual error within ±0.50 D of the target refraction. Conclusions: The overall evidence from the studies indicates that the Hill-RBF method that uses artificial intelligence and works independent of specific anatomical features is non-inferior to the Barrett Universal II, Hoffer Q, SRK/T, and Holladay 1 formulas in pediatric eyes. [ J Pediatr Ophthalmol Strabismus . 20XX;X(X):XX–XX.]