A Statistical Model to Analyze Clinician Expert Consensus on Glaucoma Progression using Spatially Correlated Visual...

Area of Science:

• Ophthalmology
• Biostatistics
• Medical Imaging

Background:

• Glaucomatous visual field (VF) progression is a leading cause of irreversible blindness.
• Accurate detection of VF progression is crucial for timely intervention and management.
• Current methods may not fully capture the complex spatial correlations in VF changes.

Purpose of the Study:

• To develop and validate a statistical model for improved detection of glaucomatous VF progression.
• To define progression based on expert clinician consensus.
• To account for spatially correlated sensitivities in visual field testing.

Main Methods:

• Developed a Bayesian statistical model incorporating spatial probit regression.
• Jointly modeled spatially correlated changes in visual sensitivities across VF locations.
• Accounted for structural similarities between neighboring VF regions.

Main Results:

• The proposed model demonstrated superior model fit and predictive ability compared to existing models (DIC: 198.15 vs. 201.29-213.38).
• Achieved significantly higher area under the ROC curve (0.80 vs. 0.59-0.72; P < 0.018) and optimal sensitivity (0.92 vs. 0.28-0.82).
• Simulation studies confirmed improved parameter estimation and inference with spatial modeling.

Conclusions:

• A novel statistical model effectively detects glaucomatous VF progression by considering spatial correlations.
• The model outperforms competing methods in key performance metrics.
• This model is readily applicable to clinical practice for enhanced glaucoma management.