Prediction of posterior elevation stability in keratoconus

Area of Science:

• Ophthalmology
• Biomedical Engineering
• Data Science

Background:

• Keratoconus is a progressive corneal disease.
• Early detection of keratoconus progression is crucial for timely intervention.
• Predictive models can aid in managing keratoconus.

Purpose of the Study:

• To investigate features of progressive keratoconus using machine learning.
• To develop predictive models for keratoconus progression.
• To identify key corneal parameters for prognosis.

Main Methods:

• Utilized Pentacam HR for corneal topography measurements.
• Included 163 eyes from 127 patients with multiple examination records.
• Applied Support Vector Machine (SVM) and logistic regression for model construction.

Main Results:

• Age, posterior elevation (PE), and index of height deviation (IHD) showed significant changes over time.
• Logistic regression achieved an Area Under the Curve (AUC) of 0.780.
• SVM demonstrated a prediction sensitivity of 52.9% and specificity of 79.0%.

Conclusions:

• Machine learning is feasible for predicting keratoconus progression and prognosis.
• Posterior elevation (PE) is a sensitive predictor of keratoconus progression.
• Corneal topography parameters can be effectively used in predictive models.