The effect of using a large language model to respond to patient messages (24 April 2024)

Summary

The relentless increase in administrative responsibilities, amplified by electronic health record (EHR) systems, has diverted clinician attention from direct patient care, fuelling burnout. In response, large language models (LLMs) are being adopted to streamline clinical and administrative tasks. Notably, Epic is currently leveraging OpenAI's ChatGPT models, including GPT-4, for electronic messaging via online portals.

The volume of patient portal messaging has escalated in the past 5–10 years, and general-purpose LLMs are being deployed to manage this burden. Their use in drafting responses to patient messages is one of the earliest applications of LLMs in EHRs.

Previous works have evaluated the quality of LLMs responses to biomedical and clinical knowledge questions; however, the ability of LLMs to improve efficiency and reduce cognitive burden has not been established, and the effect of LLMs on clinical decision making is unknown. To begin to bridge this knowledge gap, the authors of this study, published in the Lancet, carried out a proof-of-concept end-user study assessing the effect and safety of LLM-assisted patient messaging.