[Can large language models answer clinical questions?]

Abstract

With the advancement of large language models (LLMs) such as ChatGPT, their application in medicine is growing, but it is crucial that the responses are aligned with international guidelines. Recent studies have shown that LLMs can be useful in the medical field, providing correct answers to questions about the management and treatment of specific diseases. However, the accuracy of these models must also include readability and thoroughness of the answers and consistency with guidelines. In addition to these characteristics, relevance, pertinence, and up-to-date nature of the sources used by the LLMs to answer questions must be ensured. Furthermore, studies are needed to investigate the consistency of responses across different LLMs and languages used by them, as well as training processes that ensure greater reliability, especially when dealing with rare or complex diseases. Although LLMs can support medical education and decision-making, their integration into clinical practice requires further validation and comparison with international guidelines.