Dr. REMYA EDACHERY

Dr. JYOTI MATALIA, Dr.Pratibha Panmand, Dr.Nuti Shah, Dr.Shetty Bhujang K

Abstract

Aim :To predict the myopic progression using Pentacam HR and Corvis-ST data based on artificial intelligence. Methods:150 eyes of children <18 years were randomly chosen. Progression was defined as increase in spherical equivalent by > 0.5D/ year. Corvis-ST and Pentacam HR parameters were added to a random forest AI algorithm. Results:Prediction was 87.5% accurate of progressing eyes and 51.4% of stable eyes.Significant parameters in prediction were high convex time (0.143), average densitometry anterior 120µm annulus 10.0 – 12mm (0.114), constant corneal stiffness (0.1), wavefront error astigmatism 00 (0.09), wavefront error coma 00 (0.08), applanation 2 secant intersession length (0.08) and astigmatism (0.08). Rate of progression in 4 to 8 years 0.77±0.2 D/year, 9 to 13 years 0.71±0.26 D/year and 14 to 18 years 0.58±0.06 D/year.Conclusion: AI model predicted myopic progression with a good accuracy. Pentacam HR parameters showed the most influence in predicting progression.

Full Text