Generative Artificial Intelligence (AI) is poised to induce a paradigm shift in medicine and healthcare. This review provides a comprehensive overview of its foundational technologies, clinical applications, and pathways to practical integration. We first explain the principles of three core technologies: Transformers, exemplified by Large Language Models (LLMs); Diffusion Models for high-fidelity data generation; and Neural Radiance Fields (NeRF) and 3D Gaussian Splatting (3DGS) for 3D scene synthesis. We then systematically review their applications across diverse domains: automating clinical documentation, accelerating drug discovery, enhancing medical imaging diagnostics, and innovating surgical simulation. To bridge the gap to real-world implementation, we address critical system-level challenges, discussing practical solutions such as Retrieval-Augmented Generation (RAG) to mitigate hallucinations, on-premises LLMs to ensure data security, and no-code platforms to empower clinician-led development. Finally, we examine critical ethical, legal, and social issues―including data bias, interpretability, and accountability―emphasizing the need for a robust governance framework. This review underscores that generative AI is evolving beyond a mere efficiency tool into a powerful partner capable of augmenting the expertise of healthcare professionals and fundamentally shaping the future of medicine.
Nakamura et al. (Thu,) studied this question.