ECGTwin: Personalized ECG Generation Using Controllable Diffusion Model

2025-07-31

AI Summary

This study addresses two key challenges in personalized ECG digital twin generation: (1) disentangled extraction of individual physiological features without access to real subject-specific data; and (2) controllable, non-interfering injection of diverse clinical conditions (e.g., arrhythmia, ischemia) alongside individual traits during synthesis. To this end, we propose a two-stage framework: first, a robust individual feature encoder trained via contrastive learning; second, an AdaX Condition Injector integrated into a diffusion model, which employs a dual-path mechanism to jointly disentangle and conditionally inject subject-specific and pathological features. Experiments demonstrate that our generated ECGs significantly outperform baselines in signal fidelity, cross-condition diversity, and individual specificity. Moreover, downstream automated diagnosis achieves a 3.2% average F1-score improvement. The framework provides an interpretable, controllable foundation for synthetic ECG data in precision medicine.

Technology Category

• Natural Language Processing: Code Generation / Program Synthesis from Natural Language
• Machine Learning: Deep Generative Models & Autoencoders
• Search and Optimization: Mixed Discrete/Continuous Search

Application Category

• User Modeling, Personalization and Recommendation: User privacy protection in personalized systems
• Economics, Online Markets and Human Computation: Economic ramifications for generative AI infrastructure and applications
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs

Abstract

Personalized electrocardiogram (ECG) generation is to simulate a patient's ECG digital twins tailored to specific conditions. It has the potential to transform traditional healthcare into a more accurate individualized paradigm, while preserving the key benefits of conventional population-level ECG synthesis. However, this promising task presents two fundamental challenges: extracting individual features without ground truth and injecting various types of conditions without confusing generative model. In this paper, we present ECGTwin, a two-stage framework designed to address these challenges. In the first stage, an Individual Base Extractor trained via contrastive learning robustly captures personal features from a reference ECG. In the second stage, the extracted individual features, along with a target cardiac condition, are integrated into the diffusion-based generation process through our novel AdaX Condition Injector, which injects these signals via two dedicated and specialized pathways. Both qualitative and quantitative experiments have demonstrated that our model can not only generate ECG signals of high fidelity and diversity by offering a fine-grained generation controllability, but also preserving individual-specific features. Furthermore, ECGTwin shows the potential to enhance ECG auto-diagnosis in downstream application, confirming the possibility of precise personalized healthcare solutions.

Problem

• Simulate personalized ECG digital twins for specific conditions
• Extract individual features without ground truth data
• Inject diverse conditions without confusing generative models

Innovation

• Contrastive learning extracts individual ECG features
• AdaX Condition Injector integrates personalized conditions
• Diffusion model ensures high fidelity ECG generation