DR. TULASI MEDA

Prof.ROHIT SHETTY, Dr.POOJA KHAMAR, Dr. Gairik Kundu

Abstract

Aim:To analyze high risk and demographic factors influencing progression of keratoconus(KC)using an Artificial Intelligence(AI)model.Methods:500 eyes of 500 KC patients with changes in anterior curvature(Kmax)between 2 visits atleast 6months apart were studied.From our previous study using Random Forest classifier,tomographic parameters were used to classify eyes into“progression”&“no progression”.Demographic data and risk factor assessment was done using a questionnaire.AI model was built to look at these factors-its area under the curve(AUC),sensitivity(SE),specificity (SP),accuracy(AC) were looked for.Results:Changes in IgE,Vit D,eye rubbing were top parameters to classify progression among other factors.~76% cases classified as “progressors” by tomographic changes were correctly predicted based on changes in clinical factors by AI model with AUC of 0.783(<0.001),SE~83%,SP of ~87%.Conclusion:AI helps in risk stratification impacting KC progression,thus helping in better management.

Full Text