Related Experiment Video
Updated: Jul 12, 2026
Implantation Protocol of the Foldable Capsular Vitreous Body for Complex Vitreoretinal Surgery
Published on: April 14, 2026
Development and Validation of the KS Formula Version 5 for Predicting Vault After Implantable Collamer Lens
Shuntaro Tsunehiro1, Kimiya Shimizu1, Misae Ito1
1Eye Center, Sanno Hospital, and Department of Ophthalmology, International University of Health and Welfare, Tokyo, Japan.
Purpose:
To develop and validate the KS formula Version 5 for predicting vault after Visian Implantable Collamer Lens (ICL) (STAAR Surgical) implantation using a large clinical dataset and additional biometric parameters.
Methods:
This retrospective study included 2,343 eyes of 1,275 patients who underwent ICL implantation between September 2020 and September 2023. Preoperative anterior segment parameters-including aqueous depth, angle-to-angle distance (ATA) measurements, crystalline lens rise, and lens thickness (LT)-were obtained using the CASIA2 swept-source anterior segment optical coherence tomography. The KS formula Version 4 was first re-optimized using the current dataset. The addition of LT, the construction of size-specific regression equations, and the incorporation of horizontal and vertical ATA were each evaluated independently to determine their contributions to prediction accuracy. These components were then integrated to develop the KS formula Version 5. Predictive accuracy was evaluated using median absolute error (MedAE), and reproducibility was assessed with a 10-iteration holdout validation.
Results:
The optimized KS formula Version 4 achieved a MedAE of 0.0938 mm. Adding LT improved MedAE to 0.0908 mm (P < .001); size-specific regression to 0.0925 mm (P = .0074); and incorporating horizontal and vertical ATA to 0.0932 mm (P = .0376). The final KS formula Version 5 achieved a MedAE of 0.0856 mm, significantly better than the optimized Version 4 (P < .001), and outperformed it in all 10 validation iterations.
Conclusions:
The KS formula Version 5 is more accurate than the previous version and improves the reproducibility of vault prediction, facilitating its integration into routine clinical practice using swept-source anterior segment optical coherence tomography.