Ferramentas Web para descrição e simulação de modelos de células cardíacas

Abstract

Cardiac electrophysiology modeling is an important technique for studying and simulating the electrical wave propagation on cardiac tissue. The electrical wave initiates and propagates as a pulse that is known as action potential. The action potential is responsible for synchronizing the contraction and relaxation of the cardiac cells during a heartbeat. The cellular components and functions involved in the generation of an action potential are typically described by sets of ordinary differential equations. In-silico experiments of this complex phenomenon involve the description of the mathematical model and its numerical resolution. In this respect, this work targets two different goals. First, we have implemented a Web tool to create and edit cellular components and mathematical equations, based on a XML standard named CellML. The second goal is to improve the numerical resolution of models described in CellML. To this end we implemented different improvements to a previously published tool called AGOS. AGOS translates a CellML file to a C++ code that can be used for the numerical resolution of the model via the Euler or BDF methods. In this work, we have improved the numerical methods of AGOS by implementing and testing two different ways to adapt the time step. In addition, we have implemented a parallel version of the numerical solvers based on OpenMP directives, and added two numerical techniques known as Partial Evaluation and Lookup Tables to AGOS. We compared these computational techniques in terms of execution time, memory consumption and numerical error. Our preliminary results suggest that the adaptive time step methods combined with OpenMP or Lookup Tables and Partial Evaluation can be 100 times faster than the originally implemented Euler Method.