Page 181 - Artificial Intelligence for Computational Modeling of the Heart

Chapter 4 Data-driven reduction of cardiac models 153

Figure 4.22. Simulated depolarization of a model of human atria, obtained by
combining the regression-based model of action potential and the monodomain
model solver. Top left: t =0 ms; Top right: t = 100 ms; Bottom left: t = 200 ms;
Bottom right: t = 300 ms.

4.2.3 Discussion
This section presented methods to build high-fidelity electro-
physiology cellular models using a data-driven approach. This
enables efficient and accurate estimations of the action poten-
tial profile typical of complex, physics-based cellular models,
by regression from the model parameters. The Courtemanche–
Ramirez–Nattel (CRN) cellular model for human atria electro-
physiology is used as an example of method application.
A simple way to integrate data-driven cellular model in the nu-
merical solution of the monodomain model is also described, and
demonstrated for the computation of the electrical activation of
a realistic anatomical model of human atria. This integration ap-
proach relies on the use of a simplified, phenomenological cellular
model for the explicit computation of the transmembrane poten-
tial dynamics, before and after the action potential. A natural ex-
tension of this work is the investigation of alternative methods for
predicting upstroke, that do not rely on the explicit solution of a
cellular model.
The methods presented can be applied to simulate more com-
plex depolarization patterns, including the presence of different
tissues, leading to different action potential morphologies. The
extension to the modeling of abnormal rhythms is an important
research direction, towards the application of this data-driven ap-
proach to whole heart electrophysiology modeling and simula-
tion of pathological phenomena like arrhythmias, fibrillations or
tachycardia.