Large Language Models in Medicine: Application Status and Challenges

In recent years, the rapid evolution of large language models (LLMs) has driven revolutionary changes in multiple medical application scenarios, showcasing vast potential. With continuous technological developments, LLMs have made substantial advancements in clinical diagnosis and support, medical education and training, clinical documentation processing, patient interaction and public education, and medical research assistance. This paper first explores the practical applications of LLMs in these medical scenarios, analyzing how LLMs contribute to improving clinical decisionmaking efficiency, optimizing medical education, improving patient interaction, and advancing medical research. The paper then discusses the key technical elements of LLMs in the medical field, including data and knowledge construction, model training methods, and multi-modal data fusion. We also focus on the challenges faced by LLMs in medical applications, including data limitations, model hallucinations, and insufficient standardization of evaluation. Finally, the paper looks ahead to future research directions, highlighting the improvement of evaluation frameworks, the enhancement of personalized medical capabilities, the development of multi-modal medical LLMs, and the strengthening of ethical and regulatory compliance. Through these analyses, this paper aims to advance the ongoing development and practical application of medical LLMs.