Predictive Modeling of Long-Term Glaucoma Progression Based on Initial Ophthalmic Data and Optic Nerve Head...

Area of Science:

• Ophthalmology
• Glaucoma research
• Medical imaging analysis

Background:

• Open-angle glaucoma (OAG) is a leading cause of irreversible blindness.
• Predicting the rate of disease progression is crucial for effective patient management.
• Retinal nerve fiber layer (RNFL) thinning is a key indicator of glaucoma progression.

Purpose of the Study:

• To develop a predictive model for long-term rapid RNFL thinning in OAG patients.
• To identify initial optic nerve head (ONH) characteristics that predict glaucoma progression.
• To utilize machine learning for glaucoma prognosis.

Main Methods:

• Evaluated 712 eyes with OAG over 5 years with annual RNFL thickness measurements.
• Incorporated baseline ophthalmic features into a random forest (RF) machine learning model.
• Interpreted model using Shapley additive explanations and evaluated factors with a decision tree.

Main Results:

• The RF model identified greater lamina cribrosa (LC) curvature, higher intraocular pressure (IOP), visual field mean deviation, and thinner peripapillary choroid as significant predictors.
• LC curvature >12.0 strongly predicted faster RNFL thinning.
• Higher IOP (>26.5 mm Hg), greater laminar curvature (>13.95), and thinner peripapillary choroid (≤117.5 µm) were key determinants of RNFL thinning rate.

Conclusions:

• Baseline ONH characteristics and ophthalmic data can predict OAG eyes at risk of faster progression.
• Combinations of IOP, LC curvature, and choroidal thickness can stratify patients by RNFL thinning rate.
• This model provides a foundation for estimating glaucoma prognosis based on initial ONH features.