Diving into the heart of cardiac simulation: when HPC sheds light on the mysteries of arrhythmias

The heart is much more than just a pump: it's an ultra-complex electrical network. Each heart cell, or myocyte, generates electricity across its membrane thanks to the coordinated activity of dozens of specific proteins. These proteins, sensitive to variations in electrical potential, interact to produce the rhythmic oscillations that orchestrate each heartbeat.

However, various factors can disrupt this molecular ballet. With age, muscle cells disappear and are replaced by fibrous tissue, unable to transmit the electrical flow. Certain diseases - infections, genetic abnormalities or cardiomyopathies - also accelerate this process. The result is a disorganization of cardiac tissue, paving the way for arrhythmias.

Given this complexity, numerical simulation is becoming an essential tool for understanding the invisible mechanisms of arrhythmia. Historically, scientists modeled homogeneous pieces of tissue, with each element of the model representing hundreds of cells. But to grasp the fine phenomena at the root of these disorders, we now need to go down to the level of the individual cell.

This change in approach has led to an explosion in computing power requirements. "A human heart contains around two billion myocytes. Even to model a small fragment of diseased tissue requires colossal resources, so we need to use HPC and thus supercomputers", explains Mark Potse, a researcher on the Carmen project team.

Researchers need the biggest computers on the planet - exascale machines - to carry out these increasingly detailed simulations. The aim is to represent not only each cell, but also its complex interactions with its neighbors, a key element in understanding the genesis of arrhythmias.

To meet this challenge, the European MICROCARD-2 project was launched. Coordinated by Mark Potse, it brings together researchers from different countries to build software capable of simulating the behavior of each individual cell.

The task is immense. Existing models need to be optimized to take full advantage of future calculations performed on the world's most powerful supercomputers. In addition to software optimization, geometric modeling of the tissue is a real challenge. Thanks to joint efforts with MMG platform developers in particular, the researchers are now able to generate extremely complex meshes, reproducing the tight, irregular structure of cardiac tissue. "These are particularly complex meshes. So we're making calculations that sometimes exceed the limits of what's possible with existing tools. So we're improving the tools as we go along, and for MMG, after 2 years' work, it's now working for our meshes!" boasts Mark Potse.

Beyond the heart, MICROCARD-2's advances open up prospects for simulating other tissues where cellular electrical behavior plays a key role, such as the brain.

But the immediate challenge is clear: to gain a better understanding of arrhythmias in order to provide doctors with new tools for interpreting electrocardiograms. “It's not a question of creating a new automated diagnostic tool, but of providing practitioners with a better understanding of the underlying mechanisms to refine their diagnoses,” concludes Mark Potse. To achieve this goal, the scientists are working hand-in-hand with physiologists specializing in cardiac tissue, ensuring that the models remain faithful to biological reality.