Simulation of parametrized cardiac electrophysiology in three dimensions using physics-informed neural networks

2025-06-18

AI Summary

This study addresses the sensitivity to hyperparameters and training instability of physics-informed neural networks (PINNs) in strong-form PDE solving for 3D parametric myocardial electrophysiological modeling, arising from coupling between boundary conditions and material parameters. We propose a 3D PINN framework based on the Aliev–Panfilov equations, pioneering the embedding of strong-form physical constraints into parametric cardiac geometry. To enhance robustness, we introduce a synergistic strategy combining input normalization and adaptive loss weighting. Evaluated across multiple 3D scenarios featuring heterogeneous conduction and complex boundaries, our method achieves sub-3.2% error in predicting action potentials and recovery variables against finite-element benchmarks, accurately reproducing boundary effects and conduction anisotropy. This work represents the first high-fidelity, robust implementation of PINNs for 3D parametric cardiac electrophysiology simulation, establishing a novel, interpretable, data- and physics-driven paradigm for preclinical mechanistic investigation.

Technology Category

• Machine Learning: Bio-inspired Learning
• Cognitive Modeling & Cognitive Systems: Neural Spike Coding
• Computer Vision: Low Level & Physics-based Vision

Application Category

• User Modeling, Personalization and Recommendation: Psychology-informed user models and recommender systems
• Graph Algorithms and Modeling for the Web: Graph neural networks and deep learning approaches for Web-related graphs
• Systems and Infrastructure for Web, Mobile and WoT: Applied ML and AI for Web-based mobile applications

Abstract

Physics-informed neural networks (PINNs) are extensively used to represent various physical systems across multiple scientific domains. The same can be said for cardiac electrophysiology, wherein fully-connected neural networks (FCNNs) have been employed to predict the evolution of an action potential in a 2D space following the two-parameter phenomenological Aliev-Panfilov (AP) model. In this paper, the training behaviour of PINNs is investigated to determine optimal hyperparameters to predict the electrophysiological activity of the myocardium in 3D according to the AP model, with the inclusion of boundary and material parameters. An FCNN architecture is employed with the governing partial differential equations in their strong form, which are scaled consistently with normalization of network inputs. The finite element (FE) method is used to generate training data for the network. Numerical examples with varying spatial dimensions and parameterizations are generated using the trained models. The network predicted fields for both the action potential and the recovery variable are compared with the respective FE simulations. Network losses are weighed with individual scalar values. Their effect on training and prediction is studied to arrive at a method of controlling losses during training.

Problem

• Optimize PINN hyperparameters for 3D cardiac electrophysiology simulation
• Predict myocardial activity using Aliev-Panfilov model with boundary parameters
• Compare neural network predictions with finite element simulations

Innovation

• Physics-informed neural networks for 3D cardiac modeling
• Finite element method generates training data
• Optimal hyperparameters control network losses