This research evaluates how effectively large language models can summarize complex clinical documents without specific training, focusing on their ability to understand time-based relationships in patient histories.

Key Findings:

• LLMs demonstrate promising capabilities in processing lengthy clinical narratives
• Temporal reasoning enables comprehensive capture of treatment trajectories
• Zero-shot approaches offer potential for immediate application in healthcare settings
• The research addresses a critical need for efficient processing of growing volumes of clinical documentation

Business Impact: This advancement could transform healthcare data processing, improve clinical decision-making, reduce physician documentation burden, and potentially lead to better patient outcomes through more comprehensive understanding of medical histories.

Zero-shot Large Language Models for Long Clinical Text Summarization with Temporal Reasoning