Radiologically Relevant Clinical History Summarization with Large Language Models: A Multireader Performance Study

The authors aimed to determine whether large language models (LLMs) could enhance the clinical utility of imaging indications derived from clinical notes, addressing the issue of incomplete clinician-provided histories. Their study found that indications generated by LLMs were significantly more comprehensive and factual than those from referring clinicians, with the proprietary LLM being rated as the most useful for protocoling and interpretation. This suggests that LLMs could improve diagnostic accuracy and workflow efficiency in radiology.