ABSTRACT Cancer remains a major global health challenge, and the increasing availability of multimodal biomedical data has created unprecedented opportunities for precision oncology. Recent advances in generative artificial intelligence (AI), particularly large language models (LLMs), have enabled new approaches for integrating heterogeneous data sources, including electronic health records, medical imaging, pathology, genomics, and clinical text. However, current studies remain fragmented across specific tasks, cancer types, and model architectures, and a comprehensive synthesis of how generative AI can support the entire oncology continuum is still lacking. This review provides an overview of generative AI in clinical oncology, covering LLMs, generative adversarial networks, diffusion models, and multimodal foundation models. We summarize their methodological foundations and discuss applications in cancer diagnosis, prognosis prediction, treatment planning, patient management, and clinical trial optimization. Particular attention is given to multimodal data integration, synthetic data generation, clinical reasoning, and decision support, together with current challenges related to interpretability, reliability, data privacy, regulatory governance, and real‐world implementation. By consolidating recent technological advances and clinical evidence, this review highlights future priorities toward safe, trustworthy, and clinically deployable intelligent oncology systems. Emerging agent‐based architectures and human–AI collaborative workflows may further expand the clinical utility of generative AI in oncology.
Yu et al. (Tue,) studied this question.