Advancing healthcare with generative AI and large language models: a systematic review of applications in medical imaging, patient communication, and global innovation

Large language models (LLMs), Generative AI (GAI), and Generative Vision Models (GVMs) have become transformative technologies in healthcare, offering new landscapes for diagnostics, patient communication, and research innovation. To map the applications of GAI, LLMs, and GVMs in healthcare, we conducted a systematic literature review guided by the PRISMA. After screening and quality assessment, 96 studies were synthesized to address four research questions: medical imaging applications, patient communication, specialized medical research models, and global healthcare initiatives. Findings indicate that GVMs, such as GANs, DiffMIC, and WSGAN, enhance the accuracy and segmentation of medical imaging, while GAI and LLM-based chatbots, including ChatGPT and IBM Watsonx Assistant, improve patient interaction and administrative efficiency. Specialized models, including Med-PaLM 2, BioGPT, and MedCPT, can be used to perform with high capability in drug discovery, genomic analysis, and evidence-based research. Furthermore, international case studies demonstrate regional adaptations, such as AlphaFold in the UK and scGPT in Canada, which are representative of innovation in public health and personalized care. In general, GAI, LLMs, and GVMs redefine healthcare operations and research worldwide by enabling precision diagnostics, customized communication, and accelerated biomedical innovation.