CLIC: Contextual Language-Informed Cardiac Pathology Classification

TL;DR

This paper introduces CLIC, a multimodal framework that improves cardiac pathology classification by integrating patient metadata and contextual information via natural language processing, leveraging large language models and template-based descriptions.

Contribution

The novel contribution is the integration of contextual patient data into ECG classification using natural language, enhancing diagnostic accuracy over traditional signal-only methods.

Findings

Template-based contextual descriptions improve classification accuracy.

Large Language Models can generate competitive clinical descriptions.

Contextual information helps disambiguate complex physiological patterns.

Abstract

The electrocardiogram (ECG) is the gold standard for non-invasive diagnosis of cardiac pathologies and is a fundamental pillar of cardiovascular medicine. Recent progress in deep learning has led to the development of robust automated classifiers that achieve high performance by processing raw physiological signals. However, in clinical practice, diagnosis is rarely based solely on the signal. Cardiologists commonly support their interpretation with the patient's characteristics and the specific data-acquisition context. Despite this, most current algorithms remain restricted to signal-only analysis, failing to integrate technical metadata and demographic variables. This paper proposes Contextual Language-Informed Cardiac pathology classification (CLIC), a multimodal framework that significantly enhances diagnostic precision by encoding these variables through natural language. We demonstrate that translating patient-level contextual data into descriptive text provides an informative anchor that helps the model disambiguate complex physiological patterns. We further investigate the use of Large Language Models to synthesize richer clinical descriptions and observe that, while these generated texts remain competitive, controlled template-based contextual clinical text leads to consistent improvements in downstream classification performance.