In a recent study published in JAMA Internal Medicine, researchers from the University of California San Francisco (UCSF) examined the potential of artificial intelligence (AI) to draft hospital discharge summaries. The research aimed to determine if AI-generated summaries could compare favorably to those written by hospitalists at The Johns Hopkins Hospital in Baltimore.
Dr. Gigi Liu, a hospitalist at Johns Hopkins, noted that while she can efficiently complete a discharge summary in as little as 20 minutes for short-stay patients, longer hospital visits can take up to an hour. These summaries play a crucial role in guiding patient care after discharge.
The UCSF study involved a comparison of discharge summaries created by hospitalists and those generated by large language models (LLMs), a form of AI capable of synthesizing extensive information into coherent narratives. The findings revealed that while LLMs produced summaries that were more concise and coherent, they were also less comprehensive and contained a higher likelihood of errors, including omissions and inaccuracies.
The study analyzed 100 randomly selected patient encounters between 2019 and 2022, with 22 attending physician reviewers evaluating the summaries without knowing the source. The reviewers found little difference in overall quality and preference between the two methods.
Dr. Charumathi Raghu Subramanian, a lead author of the study, emphasized the importance of discharge summaries in the patient care continuum, stating that they encapsulate critical aspects of a patient's hospital experience for post-acute care providers.
High-quality discharge summaries are essential for reducing medication errors and hospital readmissions, according to the study. However, a 2021 survey indicated that 44% of hospitalists felt too overwhelmed to create high-quality summaries.
The researchers believe that LLMs could alleviate some of this burden, allowing hospitalists to devote more time to direct patient care. While LLMs have shown promise in healthcare, the study highlighted the need for careful evaluation of their outputs, as errors in AI-generated summaries could still pose risks to patient safety.
Overall, the study suggests that while AI has the potential to enhance efficiency in drafting discharge summaries, further refinement is necessary to ensure accuracy and reliability in clinical settings.
Reported by HarborBeat based on Medscape (source).
0 Comments
Log in to join the conversation.