Leveraging a Large Language Model for Streamlined Medical Record Generation: Implications for Health Care Informatics

Area of Science:

• Medical Informatics
• Artificial Intelligence in Healthcare

Background:

• Medical record documentation is crucial but time-consuming for clinicians.
• Large language models (LLMs) offer potential solutions for improving documentation efficiency and quality.

Purpose of the Study:

• To assess the efficiency and quality of LLM-generated medical record summaries compared to physician-authored records.
• To evaluate the impact of AI-assisted documentation on clinical staff workload and healthcare delivery.

Main Methods:

• A multidisciplinary team developed guidelines for LLM (Llama 3.1) generated medical summaries.
• LLM generated admission, weekly, and discharge notes for physician review and editing.
• The Physician Documentation Quality Instrument (PDQI-9) was used for quality assessment.

Main Results:

• No significant difference in PDQI-9 scores for weekly summaries and discharge notes (p=0.129, p=0.873).
• Significant difference in PDQI-9 scores for admission notes (p=0.004), with item-level variations.
• The AI note-assisted function gained popularity post-implementation.

Conclusions:

• LLMs demonstrate significant potential for enhancing medical record summary efficiency and quality.
• Successful LLM integration requires ongoing quality assessment, training, and support.
• AI-assisted documentation can free up clinical staff for more critical patient care tasks.