Bayesian Sensitivity Analysis of a Cardiac Cell Model Using a Gaussian Process Emulator.

Area of Science:

• Computational biology
• Cardiac electrophysiology
• Biophysics

Background:

• Cardiac cell models offer quantitative insights into physiological processes.
• Assessing parameter uncertainties in these complex models requires extensive computational resources.

Purpose of the Study:

• To develop and validate a surrogate statistical model using Gaussian process (GP) emulators for cardiac cell models.
• To investigate the impact of parameter uncertainties on model outputs, specifically action potential characteristics.

Main Methods:

• Utilized Gaussian process (GP) emulators to build a surrogate statistical model of the Luo-Rudy 1991 cardiac cell model.
• Analyzed the dependence of eight action potential outputs on six selected input parameters (maximum conductances).

Main Results:

• GP emulators were successfully fitted using a limited number of model runs.
• The emulators accurately reflected the physiological behavior represented by the original cardiac cell model.
• Identified key input parameters influencing action potential shape and duration restitution.

Conclusions:

• Gaussian process (GP) emulators provide a powerful and efficient tool for uncertainty and sensitivity analysis in cardiac cell modeling.
• This surrogate modeling approach significantly reduces the computational cost associated with analyzing complex physiological models.