Session-by-Session Prediction of Anti-Endothelial Growth Factor Injection Needs in Neovascular Age-Related Macular...

Area of Science:

• Ophthalmology
• Medical Imaging
• Artificial Intelligence

Background:

• Neovascular age-related macular degeneration (nAMD) causes irreversible vision loss.
• Current pro-re-nata (PRN) anti-VEGF treatment requires frequent injections, burdening patients and healthcare.
• Predicting injection needs could optimize nAMD management.

Purpose of the Study:

• To develop and evaluate machine learning (ML) models for predicting intravitreal injection necessity in nAMD patients.
• To identify key clinical and OCT-derived features for accurate prediction.
• To assess the feasibility of session-by-session injection need prediction.

Main Methods:

• Machine learning models were trained and validated using clinical variables, including retinal thickness, visual acuity, and OCT features.
• A nested cross-validation approach (Leave Some Subjects Out) ensured robust model evaluation.
• SHapley Additive exPlanations (SHAP) analysis was used to interpret model predictions and feature importance.

Main Results:

• Models combining structural and functional features achieved high accuracy in predicting injection necessity (AUC = 0.747).
• Key predictors identified included subretinal fluid, intraretinal fluid, and central retinal thickness.
• The models demonstrated the ability to predict injection needs without processing entire OCT images.

Conclusions:

• Session-by-session prediction of injection needs for nAMD is feasible using ML.
• The developed ML framework can potentially be integrated into clinical workflows.
• Optimized treatment management for nAMD can be achieved through predictive analytics.