Fast Parameterization of Human Ventricular Ionic Models Using CardioFit

Area of Science:

• Computational biology
• Cardiac electrophysiology
• Biophysics

Background:

• Cardiac action potential (AP) models require accurate parameterization for reliable simulation.
• Tuning these parameters to match experimental data is often challenging and time-consuming.

Purpose of the Study:

• To develop an efficient computational tool, CardioFit, for fitting cardiac AP model parameters.
• To facilitate the optimization of detailed human ventricular models using experimental time-series data.

Main Methods:

• CardioFit employs particle swarm optimization (PSO) for parameter fitting.
• The tool is a web-based application utilizing JavaScript and WebGL for GPU acceleration.
• Parallel processing capabilities of PSO are leveraged for rapid computation.

Main Results:

• CardioFit rapidly identifies conductance parameter values for detailed ionic models (e.g., ten Tusscher et al., O'Hara et al.).
• The tool successfully matches model outputs to experimental data within model limitations.
• Parameter fitting is achieved in minutes on consumer hardware, despite requiring thousands of model runs.

Conclusions:

• CardioFit offers a fast and accessible solution for optimizing cardiac electrophysiology models.
• The web-based, GPU-accelerated approach significantly reduces the time needed for model parameterization.
• This tool aids researchers in achieving better fits of ionic models to experimental AP data.