DR. JANANEE RAJENDRAN

DR. JYOTI MATALIA, DR. JANANEE RAJENDRAN

Abstract

Purpose: To predict myopia progression in pediatric age group using corneal tomography and biomechanics by an artificial intelligence (AI) model.

Methods: In this retrospective cross-sectional study, 321 eyes of 321 subjects aged between 5-17 years with myopia were included. Cycloplegic refraction, Corvis-ST and Pentacam HR were assessed. Eyes were sub-grouped into stable or progression if spherical power increased by > 0.5 D, after 1 year follow-up. A decision tree AI classifier (leave one-out validated) was used to predict myopia progression using Orange AI (University of Ljublijana).

Results: Extra corneal viscosity (< 0.13 Pasec), extra corneal stiffness (>24.47 N/m), spherical aberration (>0.16 µm) were the major predictors of progression. The AI had area under the curve, accuracy, and precision of 0.73, 0.71 and 0.72, respectively.

Conclusion: Myopic progression was predicted by AI using non inverse imaging data from Corvis-ST and Pentacam HR parameters.

Full Text