A systematic review of AI for predicting glaucoma progression: challenges and recommendations towards clinical implementation

Category	Detail
Condition	Glaucoma
Key Mechanisms	Irreversible optic nerve damage with variable progression rates influenced by factors such as cup-to-disc ratio, baseline visual acuity, genetic variants, and systemic vascular conditions
Target Population	Patients with glaucoma, glaucoma suspects, and ocular hypertension
Care Setting	Ophthalmology clinical settings with potential integration of AI tools for monitoring and treatment planning

Glaucoma remains the leading cause of irreversible blindness worldwide with increasing total cases despite declining age-standardised prevalence.
AI models, including machine learning and deep learning, show promise in predicting glaucoma progression by extracting latent features from complex clinical data.
Current AI applications face challenges such as heterogeneous definitions of progression, inconsistent reporting, and integration into clinical workflows.

Utilize AI-assisted diagnostic tools for early detection of glaucoma and related conditions where validated.

Incorporate AI predictions of glaucoma progression to guide personalized monitoring intensity and treatment adjustments for high-risk patients.

Implement regular visual field and optic nerve assessments combined with AI-based progression forecasting to identify rapid progressors.

Recognize variability in AI model transparency and the need for rigorous validation before clinical deployment.
Be aware of heterogeneous study designs and lack of standardized progression definitions impacting AI reliability.

Glaucoma patients and suspects from diverse international cohorts, predominantly from the United States.

AI models may enable earlier identification of rapid progression, allowing timely intensification of therapy to preserve vision.

Adopt AI tools that have undergone rigorous validation and regulatory approval for clinical use.
Ensure consistent reporting standards and transparent AI model outputs to facilitate clinical interpretation.
Integrate AI predictions within existing clinical pathways to support, not replace, clinician decision-making.
Maintain awareness of patient heterogeneity and tailor AI application accordingly.

Clinical Scorecard: A comprehensive analysis of artificial intelligence applications in forecasting glaucoma progression: obstacles and suggestions for clinical application