Interpretable Multimodal Zero-Shot ECG Diagnosis via Structured Clinical Knowledge Alignment

2025-10-24

AI Summary

Existing automated ECG diagnostic systems suffer from poor interpretability and limited generalizability, hindering zero-shot diagnosis of unseen cardiac conditions. Method: We propose ZETA—the first interpretable multimodal zero-shot ECG diagnosis framework. It leverages a large language model (LLM) to construct, and domain experts to validate, a structured clinical observation knowledge base—comprising positive and negative diagnostic signs—that aligns raw ECG signals with clinical semantics. ZETA performs zero-shot classification without disease-specific fine-tuning and enables evidence-level, feature-grounded reasoning. Crucially, it introduces LLM-derived structured clinical knowledge into ECG analysis for the first time and employs multimodal embedding contrastive learning to model differential diagnostic logic. Contribution/Results: ZETA achieves state-of-the-art performance on zero-shot ECG classification tasks. Qualitative analysis confirms its predictions are clinically interpretable and traceable to specific ECG features.

Technology Category

• Machine Learning: Large Multimodal Models (LMMs)
• Natural Language Processing: Language Grounding & Multi-modal NLP
• Computer Vision: Multi-modal Vision

Application Category

• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs
• Search and Retrieval-Augmented AI: Retrieval-Augmented Generation (RAG) and multi-modal RAG
• Economics, Online Markets and Human Computation: Humans versus LLMs for data annotation and labeling

Abstract

Electrocardiogram (ECG) interpretation is essential for cardiovascular disease diagnosis, but current automated systems often struggle with transparency and generalization to unseen conditions. To address this, we introduce ZETA, a zero-shot multimodal framework designed for interpretable ECG diagnosis aligned with clinical workflows. ZETA uniquely compares ECG signals against structured positive and negative clinical observations, which are curated through an LLM-assisted, expert-validated process, thereby mimicking differential diagnosis. Our approach leverages a pre-trained multimodal model to align ECG and text embeddings without disease-specific fine-tuning. Empirical evaluations demonstrate ZETA's competitive zero-shot classification performance and, importantly, provide qualitative and quantitative evidence of enhanced interpretability, grounding predictions in specific, clinically relevant positive and negative diagnostic features. ZETA underscores the potential of aligning ECG analysis with structured clinical knowledge for building more transparent, generalizable, and trustworthy AI diagnostic systems. We will release the curated observation dataset and code to facilitate future research.

Problem

• Develops interpretable ECG diagnosis via clinical knowledge alignment
• Addresses generalization to unseen conditions without disease-specific training
• Enhances transparency by grounding predictions in clinical observations

Innovation

• Zero-shot multimodal framework for ECG diagnosis
• Aligns ECG signals with structured clinical observations
• Uses pre-trained model without disease-specific fine-tuning