Complementarity-Preserving Generative Theory for Multimodal ECG Synthesis: A Quantum-Inspired Approach

arXiv cs.AI / 3/31/2026

💬 OpinionSignals & Early TrendsIdeas & Deep AnalysisModels & Research

共有:

Key Points

The paper argues that current multimodal ECG generative models often synthesize time, frequency, and time-frequency modalities independently, producing data that looks plausible but is physiologically inconsistent across domains.
It proposes a Complementarity-Preserving Generative Theory (CPGT), asserting that valid multimodal generation must explicitly preserve cross-domain complementarity rather than loosely coupling modality generation.
The authors instantiate CPGT with Q-CFD-GAN, a quantum-inspired generative model that uses a complex-valued latent space and complementarity-aware constraints to regulate mutual information, redundancy, and morphological coherence.
Experiments report substantial improvements, including a large reduction in latent embedding variance (82%), lower classifier plausibility error (26.6%), and much higher tri-domain complementarity (0.56 to 0.91) alongside low morphology deviation (3.8%).
Overall, the work claims that preserving multimodal information geometry is more important than optimizing each modality’s fidelity separately for synthetic ECG data intended for downstream clinical ML tasks.

Abstract

Multimodal deep learning has substantially improved electrocardiogram (ECG) classification by jointly leveraging time, frequency, and time-frequency representations. However, existing generative models typically synthesize these modalities independently, resulting in synthetic ECG data that are visually plausible yet physiologically inconsistent across domains. This work establishes a Complementarity-Preserving Generative Theory (CPGT), which posits that physiologically valid multimodal signal generation requires explicit preservation of cross-domain complementarity rather than loosely coupled modality synthesis. We instantiate CPGT through Q-CFD-GAN, a quantum-inspired generative framework that models multimodal ECG structure within a complex-valued latent space and enforces complementarity-aware constraints regulating mutual information, redundancy, and morphological coherence. Experimental evaluation demonstrates that Q-CFD-GAN reduces latent embedding variance by 82%, decreases classifier-based plausibility error by 26.6%, and restores tri-domain complementarity from 0.56 to 0.91, while achieving the lowest observed morphology deviation (3.8%). These findings show that preserving multimodal information geometry, rather than optimizing modality-specific fidelity alone, is essential for generating synthetic ECG signals that remain physiologically meaningful and suitable for downstream clinical machine-learning applications.