Published on
AI-generated reports using large language models (LLMs) seem inevitable in healthcare, especially when it comes to the time-consuming task of writing up patient discharge summaries, a commentary post in JAMA Internal Medicine states. Right now, the responsibility of compiling summaries is one of the many administrative burdens leading to burnout among clinicians, and leveraging LLM tools could reduce that burden if done thoughtfully and with appropriate data integration. In a related article, researchers found no significant difference in overall quality between LLM- and physician-generated narratives. Although the LLM-generated summaries were more concise and coherent than those created by physicians, the LLM versions were less comprehensive, the authors found. The AI narratives contained more unique errors (mean [SD] errors per summary, 2.91 [2.54]) than those written by humans (mean [SD] errors per summary, 1.82 [1.94]). Yet there was no significant difference in the potential for harm between LLM- and physician-generated narratives across individual errors (mean [SD] of 1.35 [1.07] vs 1.34 [1.05]; P = .99). Harm was scored from 0 to 7, based on an adapted Agency for Healthcare Research and Quality scale.
Coming soon: “This is relevant to urgent care,” says Alan A. Ayers, MBA, MAcc, President of Urgent Care Consultants and Senior Editor of JUCM. “However, in a hospital setting, discharge instructions can be more complicated compared to the lower acuity found in urgent care. Also, in urgent care, the top 20 diagnoses represent 80% of visits, so there is less variability compared to hospital discharge. It is indeed a matter of when, not if, AI-generated discharge summaries will be used in patient care.”