# Reliability of Prediction Models for the Functional Classification of a Sinusoidal Intraocular Lens Depending on Pupil Diameter

## TL;DR

This study evaluates how consistent different models are in predicting the performance of a pupil-adaptive intraocular lens across varying pupil sizes.

## Contribution

The paper introduces a detailed comparison of six prediction models for a sinusoidal intraocular lens's functional classification based on pupil diameter.

## Key findings

- All models showed a Full-DOFi response at pupil diameters <2.5 mm.
- Functional classifications varied significantly at pupil diameters >3.5 mm depending on the model used.
- Visual acuity prediction models can provide clinically relevant metrics but require careful interpretation due to variability.

## Abstract

Background: To assess the agreement among prediction models for the functional classification of intraocular lenses (IOLs) and discuss their limitations in evaluating pupil dependency of a sinusoidal IOL. Methods: An ISO-compliant optical bench setup with modifications to characterize the modulation transfer function area (MTFa) across pupil diameters from 1.5 to 5.5 mm was used to measure the Acriva Trinova Pro C Pupil Adaptive IOL. Six prediction models (Vega et al., 2018, Fernández et al., 2019, Alarcón et al., 2016, Armengol et al., 2020 were applied to estimate visual acuity defocus curves from MTFa and functional classification based on the depth-of-field (DOFi) and the increase in visual acuity (ΔVA) from intermediate to near. Results: Defocus curves for all prediction models consistently demonstrated a Full-DOFi response (>2.3 D at 0.2 logMAR), with differences in ΔVA emerging across pupil diameters. Continuous decreases (ΔVA < 0.05 logMAR) were observed at pupil diameters <2.5 mm, while Smooth transitions (ΔVA from 0.05 to 0.14 logMAR) occurred between 2.5–3.0 mm for all models except for Vega. At pupil diameters >3.5 mm, most models transitioned to a Steep classification (ΔVA ≥ 0.14 logMAR), except Fernández, which remained Smooth, and Armengol 2020a, which shifted to Steep at 4.0 mm. Conclusions: Visual acuity prediction models provide useful means of reporting optical bench data in clinically familiar metrics. However, outcomes should be interpreted with caution as functional classifications can vary depending on the optical bench setup and prediction model used.