Semantic Compensation via Adversarial Removal for Robust Zero-Shot ECG Diagnosis

2026-04-01

AI Summary

Existing ECG-language pretraining methods suffer from severe degradation in semantic alignment when leads or temporal segments are missing, hindering robust zero-shot diagnosis. To address this, this work proposes SCAR, a novel framework that introduces, for the first time, a differentiable adversarial masker to simulate the absence of critical diagnostic information and designs a semantics-supervised adaptive selector to reweight visible tokens, thereby enabling a semantic compensation mechanism tailored for missing-data scenarios. We also introduce CMRS, a new metric to quantify semantic robustness. Experiments demonstrate that SCAR significantly improves semantic alignment under joint missing conditions across six datasets, excelling particularly when primary diagnostic evidence is absent, and effectively enhances linear-probe transfer performance.

Abstract

Recent ECG--language pretraining methods enable zero-shot diagnosis by aligning cardiac signals with clinical text, but they do not explicitly model robustness to partial observation and are typically studied under fully observed ECG settings. In practice, diagnostically critical leads or temporal segments may be missing due to electrode detachment, motion artifacts, or signal corruption, causing severe degradation of cross-modal semantic alignment. In this paper, we propose \textbf{SCAR}, a robust ECG--language pretraining framework for \textbf{S}emantic \textbf{C}ompensation via \textbf{A}dversarial \textbf{R}emoval. SCAR improves robustness by explicitly training the model to remain semantically aligned with semantically critical missingness and to recover diagnostic meaning from the remaining visible evidence. Specifically, we introduce a differentiable adversarial masker to remove the most alignment-critical spatio-temporal ECG tokens during training, forcing the ECG encoder to learn representations that remain semantically aligned with clinical text even when primary diagnostic evidence is missing. Under such adversarial corruption, we equip the ECG encoder with a semantically supervised adaptive selector that learns to reweight the remaining visible tokens and compensate with secondary yet diagnostically informative morphological cues. To evaluate robustness beyond classification accuracy, we further introduce Counterfactual Missingness Resolution Score (CMRS), which quantifies how well feature preserve diagnostic semantics under missingness. Experiments on $6$ datasets show that SCAR consistently improves semantic robustness under joint lead and temporal missingness, with particularly clear advantages in harder cases where primary diagnostic evidence is unavailable, while also yielding stronger linear-probing transferability.

Problem

• zero-shot ECG diagnosis
• partial observation
• missingness robustness
• semantic alignment
• ECG-language pretraining

Innovation

• adversarial removal
• semantic compensation
• zero-shot ECG diagnosis
• missingness robustness
• ECG-language pretraining