Dr.POOJA KHAMAR

PROF.ROHIT SHETTY

Abstract

Purpose:To demonstrate accuracy of AI based simulation software for predicting corneal stiffness (CS) after LASIK, SMILE and PRK.

Methods: 529 eyes of 529 patients were included in study & were divided into training (n=371) & test (n=158) cohorts. In AcuSimX, iFEM virtual patient specific corneal model was built from pre-op Corvis-ST deformation & Pentacam HR data. iFEM used virtual model&aspheric ablation profile to estimate post-op CS.The computed post-op CS was refined using Lasso regression AI derived from training cohort &validated on test cohort.

Results:The mean absolute error was 6.24 N/m &intraclass coefficient (ICC) was 0.84 [95% confidence interval[0.80-0.87]in training cohort of 371 eyes. The mean absolute error was 6.47 N/m while ICC was 0.84 [0.78-0.89] in test cohort consisting of 158 eyes.

Conclusion: AcuSimX is first biomechanical simulation software for refractive surgery clinics which could serve as a “pathfinder” in choosing right procedure.