Presentation Information
[5Yin-A-25]Parameter estimation of an ECG model based on delay-coupled oscillators using physics-informed neural networks
〇Daiki Watanabe1, Kentaro Takeda1 (1. Kagawa University)
Keywords:
Electrocardiogram (ECG),Physics-informed neural networks (PINNs)
Cardiac rhythm arises from periodic self-excitation of the sinoatrial node, which propagates with delays through the atrioventricular node and the His-Purkinje complex, producing coordinated contraction and relaxation of the heart.
Electrocardiograms (ECGs) record this electrical propagation using surface electrodes. Previous studies have modeled ECG signals as delay-coupled systems of modified van der Pol oscillators.
In this study, we propose a method to estimate coupling parameters from ECG waveforms generated by numerical simulations using physics-informed neural networks (PINNs).
We demonstrate that the proposed method successfully estimates coupling parameters for different cardiac rhythms including sinus rhythm and ventricular flutter.
These results suggest that PINNs may enable automatic diagnosis of cardiac conditions from ECG signals.
Electrocardiograms (ECGs) record this electrical propagation using surface electrodes. Previous studies have modeled ECG signals as delay-coupled systems of modified van der Pol oscillators.
In this study, we propose a method to estimate coupling parameters from ECG waveforms generated by numerical simulations using physics-informed neural networks (PINNs).
We demonstrate that the proposed method successfully estimates coupling parameters for different cardiac rhythms including sinus rhythm and ventricular flutter.
These results suggest that PINNs may enable automatic diagnosis of cardiac conditions from ECG signals.