JMIR Publications

ABSTRACT

Background:

Effective healthcare communication is crucial in the medical field. Text-based large language models offer a variety of possibilities for improving communication. To date, there are no published reviews on the use of large language models in healthcare communication.

Objective:

To summarize the applications and challenges of large language models in healthcare communication and to identify directions for future research.

Methods:

A comprehensive literature search of PubMed, Embase, Web of Science, and the Cochrane Library was conducted from January 2018 to July 2024. Eligible studies are those that use large language models to facilitate healthcare communication between the public, patients, and Clinicians. All articles selected and data extracted were double-checked. The data was analyzed using an inductive descriptive approach, and presented in a table and narrative form.

Results:

Thirty-seven studies were included in this review, summarizing three patterns of large language model application in healthcare communication: healthcare information conversion (n=19), interactive content generation (n=17), and healthcare training support (n=1). Research findings indicate that large language models can effectively improve access to medical information, simplify clinical workflows, and innovate medical education models. However, challenges remain in terms of accuracy and security, legal and ethical issues, and limitations in emotional interaction.

Conclusions:

This review provides evidence that large language models have broad prospects in the field of healthcare communication. Future research should focus on optimizing evaluation systems, improving model performance, refining ethical regulatory frameworks, and optimizing human-machine collaboration models to fully unleash the potential of large language models in healthcare communication. Clinical Trial: Open Science Framework websites (https://osf.io/7vf2u/, registration DOI: https://doi.org/10.17605/OSF.IO/YVXSP).