This study evaluates specialized clinical large language models (cLLMs) for extracting medical history entities from unstructured clinical notes, enabling better healthcare data structuring for downstream tasks.

• Extracts entities related to chief complaints, history of present illness, and past/family/social history
• Fine-tuned clinical LLMs can be deployed on-premises to protect sensitive patient data
• Provides a comparative analysis of different model capabilities for medical text processing

This research is significant for healthcare organizations seeking to convert free-text clinical documentation into standardized electronic health records, improving continuity of care, medical coding accuracy, and quality metrics tracking.

Extracting Patient History from Clinical Text: A Comparative Study of Clinical Large Language Models