Artificial intelligence (AI) has emerged as a transformative force in ophthalmology, enabling automated, accurate, and efficient clinical reporting. This review summarizes recent advances in AI-driven report generation, emphasizing the integration of multimodal imaging and clinical data. Deep learning and natural language processing (NLP) models can synthesize information from diverse sources-including fundus photography, optical coherence tomography, fluorescein angiography, and patient records-to generate structured, interpretable, and personalized diagnostic reports. Such systems enhance diagnostic precision, streamline workflow, and reduce interobserver variability. We outline the technological foundations underlying these systems, including convolutional and transformer-based architectures, self-supervised and multimodal learning, and large language models. Representative applications in diabetic retinopathy, glaucoma, cataract, and age-related macular degeneration are discussed, highlighting their clinical value and emerging real-world deployment. Persistent challenges-including data heterogeneity, model interpretability, ethical governance, and clinical integration-are critically reviewed. Finally, we explore future directions such as real-time AI-assisted reporting, predictive and personalized analytics, and global scalability across healthcare ecosystems. Multimodal, explainable, and clinically integrated AI systems hold promise to redefine ophthalmic diagnostics and improve both clinician efficiency and patient outcomes.
Shen et al. (Thu,) studied this question.