Shifts in emergency physicians' attitudes toward large language model-based documentation: a pre- and post-implementation study

Abstract

Large language models (LLMs) can assist physicians in writing medical notes more efficiently. This study evaluates whether using an LLM assistant for writing emergency department discharge notes can reduce doctors' workload and addresses concerns regarding the incorporation of AI in medical practice. Eight emergency doctors with an average experience of 12 years participated in our study. We surveyed them prior to, post 3 days, and post 5 weeks of their LLM usage. The results showed that doctors' concerns about using LLMs decreased significantly and remained low throughout the study period. Moreover, the LLM usage considerably reduced the perceived workload, with the time required to write each discharge note reduced by one-third of the original time. These findings demonstrate that doctors readily accept and benefit from LLM assistants in their daily practice. Our study provides the first real-world evidence of how doctors' attitudes toward AI assistants change over time in clinical settings, offering valuable insights into the future implementation of LLM-based documentation tools in healthcare.