Development of the Integrated Glaucoma Risk Index.

Area of Science:

• Ophthalmology
• Medical Imaging
• Machine Learning

Background:

• Current machine learning models for glaucoma diagnosis classify subjects as normal or glaucoma-positive but lack severity assessment.
• Existing glaucoma risk indices rely on single indicators, failing to capture overall disease progression.
• There is a need for a comprehensive index that integrates multiple diagnostic data points for accurate glaucoma severity evaluation.

Purpose of the Study:

• To develop and validate an integrated glaucoma risk index (I-GRI) that quantifies glaucoma severity.
• To incorporate data from visual field (VF) tests, optical coherence tomography (OCT), and intraocular pressure (IOP) measurements.
• To create a single, reliable measure for assessing glaucoma risk and progression.

Main Methods:

• Extracted key features from VF, OCT, and IOP examination data using machine learning algorithms.
• Developed a mathematical equation to integrate these features into a single index, the I-GRI.
• Validated the I-GRI using test cases, assessing its distribution and classification accuracy.

Main Results:

• The proposed I-GRI generates a risk score ranging from 0 to 1, with higher values indicating greater glaucoma risk/severity.
• Sanity tests demonstrated a balanced distribution of the I-GRI across both normal and glaucoma patient groups.
• Classification of normal and glaucoma cases using the I-GRI achieved a low misclassification rate of 7% (0.07).

Conclusions:

• The integrated glaucoma risk index (I-GRI) effectively diagnoses glaucoma and monitors its progression.
• This novel index provides a more comprehensive assessment of glaucoma severity than previous methods.
• The I-GRI shows significant potential for clinical application in glaucoma management.