Artificial intelligence in the prediction of glaucoma development and progression: A systematic review.

Abstract

Glaucoma remains the leading cause of irreversible blindness worldwide. Artificial intelligence (AI) may help enhance predict glaucoma development and progression. We provide a consolidated review of AI usage in predicting the (a) development of glaucoma in glaucoma suspects and normal patients, (b) progression of existing glaucoma, and (c) progression towards the occurrence of surgery. We searched PubMed, EMBASE, Scopus, ScienceDirect, and CENTRAL for observational studies and clinical trials comparing different AI models or AI models versus physician performance published in English from Aug 17, 2013, to Dec 5, 2024. We excluded studies describing AI models that required physician assistance or were designed to diagnose glaucoma. A total of 44 studies (7 studies for the development of glaucoma in glaucoma suspects and normal patients, 30 studies for progression of existing glaucoma, and 7 studies for progression towards the occurrence of surgery) were included. AI demonstrates favorable performance in predicting glaucoma development in glaucoma suspects and normal patients, as well as glaucomatous progression in diagnosed patients. There is significant potential for AI to aid the surveillance of glaucoma in those without a prior history. Moreover, its ability to predict future glaucomatous progression in diagnosed patients could improve systems-of-care targeted at halting disease progression.