Objective: To investigate the clinical efficacy of anti-vascular endothelial growth factor (anti-VEGF) monoclonal antibody therapy for wet age-related macular degeneration (AMD) using machine learning algorithms. Methods: A retrospective cohort analysis was conducted on 128 patients with wet AMD who received anti-VEGF therapy. Demographic characteristics, baseline visual acuity, imaging, and biomarker data were collected. Recursive feature elimination was used to screen variables, and the performance of multiple models, including random forest (RF) and support vector machine (SVM), was constructed and compared using 10-fold cross-validation. Results: The RF model performed best, with an internal validation accuracy of 85.9%, an AUC of 0.91, a sensitivity of 88.2%, and a specificity of 83.7%. Key predictive factors included baseline central retinal thickness [CRT, (>300 μm)], presence of subretinal fluid (SRF), age (>70 years), and baseline visual acuity (<0.5 logMAR). External validation accuracy was 82.1%, with an AUC of 0.88. SHAP analysis showed that CRT and SRF contributed most to the prediction. Conclusion: This model can assist in clinical treatment plan customization, improve diagnostic and therapeutic efficiency, and avoid the waste of ineffective treatment resources.