Question Answering on Patient Medical Records with Private Fine-Tuned LLMs

TL;DR

This paper presents a privacy-preserving approach for semantic question answering over electronic health records using fine-tuned LLMs, demonstrating superior performance over larger models in specific medical QA tasks.

Contribution

It introduces a novel method combining resource identification and question answering on EHRs with private, fine-tuned LLMs, outperforming benchmark models like GPT-4 in accuracy.

Findings

Fine-tuned LLMs outperform GPT-4 by 0.55% in F1 score.

Smaller models achieve 42% better Meteor scores.

Sequential fine-tuning and self-evaluation improve model performance.

Abstract

Healthcare systems continuously generate vast amounts of electronic health records (EHRs), commonly stored in the Fast Healthcare Interoperability Resources (FHIR) standard. Despite the wealth of information in these records, their complexity and volume make it difficult for users to retrieve and interpret crucial health insights. Recent advances in Large Language Models (LLMs) offer a solution, enabling semantic question answering (QA) over medical data, allowing users to interact with their health records more effectively. However, ensuring privacy and compliance requires edge and private deployments of LLMs. This paper proposes a novel approach to semantic QA over EHRs by first identifying the most relevant FHIR resources for a user query (Task1) and subsequently answering the query based on these resources (Task2). We explore the performance of privately hosted, fine-tuned LLMs, evaluating them against benchmark models such as GPT-4 and GPT-4o. Our results demonstrate that fine-tuned LLMs, while 250x smaller in size, outperform GPT-4 family models by 0.55% in F1 score on Task1 and 42% on Meteor Task in Task2. Additionally, we examine advanced aspects of LLM usage, including sequential fine-tuning, model self-evaluation (narcissistic evaluation), and the impact of training data size on performance. The models and datasets are available here: https://huggingface.co/genloop