A post hoc analysis of the phase III HAWK and HARRIER studies evaluated an AI-based disease activity (DA) model for detecting disease activity in patients with neovascular age-related macular degeneration (nAMD). The authors published their findings online in Ophthalmology Science. The model generated DA scores from optical coherence tomography (OCT) images and other parameters, classifying assessments as “easy,” “noisy,” or “difficult” based on agreement with investigator decisions. A panel of 10 retina specialists reviewed 425 OCT assessments, providing majority votes to evaluate model performance. The DA model showed 80% overall accuracy, with 96% accuracy for “easy” cases, comparable to investigators and panelists. For “noisy” cases, it matched panelist performance (84% vs 86%) and exceeded investigator accuracy (16%). In “difficult” cases, the model outperformed investigators (74% vs 53%) but was less specific than panelists (86%). Subretinal and intraretinal fluids were key parameters.