Early postoperative vault after implantable collamer lens surgery: retrospective model development and an equation recheck

Purpose.

To develop and internally evaluate models of early postoperative vault after implantable collamer lens (ICL) implantation, and to recheck published sizing equations under explicit meridian and input rules. The compact model used recorded implant orientation.

Methods.

We retrospectively analysed 200 patient records (397 eyes) implanted during 2022-2025. The original modelling analysis compared nine regression algorithms using patient-grouped five-fold cross-validation and derived compact vault and high-vault equations. A source-workbook recheck recalculated cohort summaries, paired follow-up, selected correlations and published-equation errors with stated input and orientation rules. Fold-level model predictions remain those of the original modelling analysis.

Results.

Early postoperative vault averaged 611 ± 233 μm. Near-horizontal and near-vertical eyes had mean vaults of 651 and 534 μm; the patient-cluster mean difference was -117 μm (95% CI -176 to -59). Recalculated MAE was 178.4 μm for Zhu et al.'s hSTS equation and 236.5 μm for KS with STS substituted for ATA. Binary and elliptical meridian matching reduced the latter to 205.7 and 197.1 μm. These STS versions are rewritten implementations. The rounded display of the original four-variable equation was Vault (μm) = -2056 + 285·ACD + 139·ICL size − 30·cylinder − 127·(vertical = 1), with reported CV R² = 0.32 and MAE = 150 μm.

Conclusions.

Recorded meridian was associated with early vault and transferred-equation error. The compact equation remains an internally derived model whose cross-validation has not been rerun after the workbook recheck. Preoperative use requires a documented implantation plan, reconciled predictor coding and independent evaluation.