Efficient Fine-Tuning of Large Language Models for Automated Medical Documentation

2024-09-14

2024 4th International Conference on Digital Society and Intelligent Systems (DSInS)

AI Summary

Excessive administrative burden on physicians—particularly EHR documentation—contributes to burnout and reduced clinical efficiency. To address this, we propose MediGen, a LLaMA3-8B–based model fine-tuned via LoRA and instruction-tuned with supervised fine-tuning (SFT). MediGen integrates clinical dialogue cleaning, structured prompt engineering, and EHR-aligned design to enable end-to-end generation of structured medical reports from unstructured physician–patient dialogues. Its key contributions are: (1) the first framework to deeply couple dialogue-driven generation with EHR structural requirements; and (2) a lightweight, robust, and interpretable fine-tuning paradigm tailored for clinical NLP. Evaluated on a real-world physician–patient dialogue test set, MediGen achieves ROUGE-L = 58% and BERTScore-F1 = 72%. It significantly reduces documentation time while improving report consistency and clinical utility.

Technology Category

• Machine Learning: Large Multimodal Models (LMMs)
• Natural Language Processing: Generation
• Data Mining & Knowledge Management: Conversational Systems for Recommendation & Retrieval

Application Category

• Search and Retrieval-Augmented AI: Retrieval-Augmented Generation (RAG) and multi-modal RAG
• User Modeling, Personalization and Recommendation: Large Language Models (LLM) for user modeling and recommendation
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs

Abstract

Scientific research indicates that for every hour spent in direct patient care, physicians spend nearly two additional hours on administrative tasks, particularly on electronic health records (EHRs) and desk work. This excessive administrative burden not only reduces the time available for patient care but also contributes to physician burnout and inefficiencies in healthcare delivery. To address these challenges, this study introduces MediGen, a fine-tuned large language model (LLM) designed to automate the generation of medical reports from medical dialogues. By leveraging state-of-the-art methodologies for fine-tuning open-source pretrained models, including LLaMA3-8B, MediGen achieves high accuracy in transcribing and summarizing clinical interactions. The fine-tuned LLaMA3-8B model demonstrated promising results, achieving a ROUGE score of 58% and a BERTScore-F1 of 72%, indicating its effectiveness in generating accurate and clinically relevant medical reports. These findings suggest that MediGen has the potential to significantly reduce the administrative workload on physicians, improving both healthcare efficiency and physician well-being.

Problem

• Reducing physician administrative workload from EHR documentation tasks
• Automating medical report generation from clinical dialogues using LLMs
• Addressing physician burnout caused by excessive paperwork burden

Innovation

• Fine-tuned LLaMA3-8B model for medical reports
• Automated medical report generation from dialogues
• Achieved high accuracy with ROUGE and BERTScore metrics