Robust AI-ECG for Predicting Left Ventricular Systolic Dysfunction in Pediatric Congenital Heart Disease

2025-09-23

AI Summary

To address the challenge of screening left ventricular systolic dysfunction (LVSD) in pediatric patients with congenital heart disease under low-data regimes, this paper proposes a robust AI-ECG framework. Methodologically, it integrates synthetic noise augmentation, manifold regularization, and uncertainty calibration within a unified adversarial training paradigm. Its core innovations are: (i) manifold-aware adversarial perturbation, leveraging the intrinsic geometric structure of ECG signals to improve generalization from limited samples; and (ii) an uncertainty-aware adversarial training framework that jointly optimizes predictive confidence and robustness against adversarial perturbations. The method supports deployment across multiple neural architectures. Evaluated on a real-world pediatric ECG dataset, it achieves significant improvements in LVSD detection accuracy and robustness under low-labeling conditions—yielding AUC gains of ≥8.2%—and demonstrates strong potential for cost-effective deployment in resource-constrained clinical settings.

Technology Category

• Machine Learning: Adversarial Learning & Robustness
• Computer Vision: Adversarial Attacks & Robustness
• Multiagent Systems: Adversarial Agents

Application Category

• Search and Retrieval-Augmented AI: Agentic search
• Systems and Infrastructure for Web, Mobile and WoT: Applied ML and AI for Web-based mobile applications
• Economics, Online Markets and Human Computation: Economic ramifications for generative AI infrastructure and applications

Abstract

Artificial intelligence-enhanced electrocardiogram (AI-ECG) has shown promise as an inexpensive, ubiquitous, and non-invasive screening tool to detect left ventricular systolic dysfunction in pediatric congenital heart disease. However, current approaches rely heavily on large-scale labeled datasets, which poses a major obstacle to the democratization of AI in hospitals where only limited pediatric ECG data are available. In this work, we propose a robust training framework to improve AI-ECG performance under low-resource conditions. Specifically, we introduce an on-manifold adversarial perturbation strategy for pediatric ECGs to generate synthetic noise samples that better reflect real-world signal variations. Building on this, we develop an uncertainty-aware adversarial training algorithm that is architecture-agnostic and enhances model robustness. Evaluation on the real-world pediatric dataset demonstrates that our method enables low-cost and reliable detection of left ventricular systolic dysfunction, highlighting its potential for deployment in resource-limited clinical settings.

Problem

• Improving AI-ECG detection of pediatric heart dysfunction with limited data
• Developing robust training methods for low-resource pediatric ECG analysis
• Enabling reliable left ventricular dysfunction screening in resource-limited hospitals

Innovation

• On-manifold adversarial perturbation for synthetic ECG noise
• Uncertainty-aware adversarial training algorithm for robustness
• Architecture-agnostic framework for low-resource pediatric ECG analysis