Evaluating Visual Field Phenotypes in Glaucomatous Patients with Cardiovascular Disease

Abstract

Although glaucoma is a leading cause of irreversible blindness, the pathophysiology of primary open angle glaucoma (POAG) is poorly understood. Emerging evidence points towards an association between cardiovascular disease (CVD) and POAG, with CVD reported as an independent risk factor for rapid disease progression. This thesis aims to further understand this relationship and to determine how baseline visual field characteristics relate to CVD and subsequent glaucoma progression.

First, existing literature on the relationship between CVD and POAG was reviewed. Further original studies were conducted: the exploration of baseline Humphrey Visual Field (HVF) phenotypes in POAG patients with CVD and other vascular risk factors, the use of artificial intelligence models to predict CVD status from HVFs, and the use of unsupervised archetypal analysis in identifying baseline HVF patterns in rapid and non-rapid POAG progressors.

CVD and other vascular risk factors may be associated with specific HVF phenotypes in POAG patients such as superior arcuate and central defects. Machine learning models outperformed deep learning models in predicting CVD status from HVF input. Archetypal analysis successfully identified clinically relevant HVF patterns in rapid and non-rapid progressors.