ALFRED: Ask a Large-language model For Reliable ECG Diagnosis

TL;DR

The paper tackles the challenge of reliable and explainable ECG diagnosis using large language models. It proposes ALFRED, a zero-shot framework that fuses retrieval-augmented generation with expert-curated ECG knowledge, a feature extraction module, and a rule-based diagnostic engine to ground LLM outputs in clinical expertise. Through evaluation on the PTB-XL dataset, the approach demonstrates improved diagnostic performance and interpretability, with an Ablation study highlighting the contributions of rule guidance and knowledge augmentation. The work advances practical automated ECG interpretation and provides a publicly available framework that can extend beyond the tested dataset to broader clinical contexts.

Abstract

Leveraging Large Language Models (LLMs) with Retrieval-Augmented Generation (RAG) for analyzing medical data, particularly Electrocardiogram (ECG), offers high accuracy and convenience. However, generating reliable, evidence-based results in specialized fields like healthcare remains a challenge, as RAG alone may not suffice. We propose a Zero-shot ECG diagnosis framework based on RAG for ECG analysis that incorporates expert-curated knowledge to enhance diagnostic accuracy and explainability. Evaluation on the PTB-XL dataset demonstrates the framework's effectiveness, highlighting the value of structured domain expertise in automated ECG interpretation. Our framework is designed to support comprehensive ECG analysis, addressing diverse diagnostic needs with potential applications beyond the tested dataset.

Figures (1)

• Figure 1: The proposed framework and its process pipeline.