Artificial intelligence for precision therapeutics in age-related macular degeneration: current advances, challenges,...

Age-related macular degeneration (AMD) is a leading cause of irreversible vision loss worldwide and is characterized by substantial clinical, imaging, and molecular heterogeneity that complicates disease prediction and therapeutic management. Recent advances in artificial intelligence (AI) and precision therapeutics have created new opportunities for more individualized and data-driven AMD care. AI models trained on multimodal datasets-including fundus photography, optical coherence tomography (OCT), optical coherence tomography angiography (OCTA), genetic susceptibility loci (e.g., CFH, ARMS2/HTRA1, C3, CFI, and APOE), and longitudinal clinical information-have demonstrated promising capability in early disease detection, progression forecasting, biomarker identification, and prediction of treatment response. These developments align closely with emerging precision therapeutic strategies, including optimized anti-vascular endothelial growth factor (anti-VEGF) regimens, complement-targeted therapies, gene-based interventions, and stem cell-associated regenerative approaches. This review provides a translational overview of AI-enabled precision therapeutics in AMD, with emphasis on multimodal biomarker integration, individualized therapeutic stratification, longitudinal disease monitoring, and clinically interpretable AI systems. Importantly, we further propose a Five-Level Clinical Readiness and Translational Utility Framework for AI in AMD Precision Therapeutics, categorizing AI applications according to evidence strength, clinical maturity, validation status, interpretability, and real-world implementation potential. The framework distinguishes near-reference-standard imaging AI systems, advanced clinical decision-support tools, emerging multimodal precision therapeutic AI, supportive workflow-oriented AI systems, and currently limited or unsuitable AI applications. Despite substantial progress, important translational barriers remain, including limited external validation, retrospective study designs, dataset heterogeneity, domain shift, insufficient explainability, regulatory uncertainty, and challenges related to workflow integration and real-world clinical deployment. Future advances in multimodal longitudinal AI, explainable AI, federated learning, digital health platforms, and multi-omics integration may facilitate a transition from reactive disease management toward more proactive, predictive, and personalized ophthalmic care. Collectively, AI-enabled precision therapeutics may help establish a more scalable and clinically integrated framework for individualized AMD management and future precision ophthalmology.