CLIC: Contextual Language-Informed Cardiac Pathology Classification
For cardiologists and ML practitioners, this work addresses the gap between clinical practice (using patient context) and automated ECG analysis, though the improvement is incremental over existing multimodal approaches.
CLIC enhances ECG-based cardiac pathology classification by integrating patient metadata and acquisition context as natural language descriptions, achieving consistent improvements over signal-only baselines. Template-based contextual text outperforms LLM-generated descriptions.
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.