Fitting membrane resistance along with action potential shape in cardiac myocytes improves convergence: application
Jaspreet Kaur1, Anders Nygren1, Edward J Vigmond2
1Electrical and Computer Engineering, University of Calgary, Alberta, Canada.
Plos One
|September 25, 2014
Summary
Fitting cardiac models is improved by including membrane resistance (Rm) alongside action potential (AP) data. This approach enhances parameter fitting convergence and reduces variability, leading to more robust models for predicting tissue behavior.
Area of Science:
- Computational Biology
- Cardiac Electrophysiology
- Biophysics
Background:
- Parameter fitting for cardiac single cell ionic models is typically time-consuming and relies on matching action potentials (APs).
- Discrepancies exist as different parameter sets can yield similar APs, and single-cell accuracy doesn't guarantee accurate tissue behavior.
- Existing methods struggle with parameter uncertainty and preserving tissue-level dynamics.
Purpose of the Study:
- To investigate if incorporating membrane resistance (Rm) fitting alongside AP morphology improves the accuracy and robustness of cardiac ionic model parameterization.
- To reduce uncertainty in parameter sets and enhance the predictability of tissue-level behavior from single-cell models.
Main Methods:
- Developed a genetic algorithm incorporating both AP morphology and membrane resistance (Rm) data at specific voltages.
- Compared the performance of this dual-objective algorithm against a standard algorithm using only AP morphology.
- Validated the approach by fitting models to themselves, to other models, and to experimental rabbit AP data.
Main Results:
- The genetic algorithm with added Rm fitting demonstrated significantly faster convergence, achieving a smaller mean square error (MSE) in fewer generations.
- Parameter set variability was substantially reduced, with many parameters showing an order of magnitude decrease in variability.
- The inclusion of Rm improved the overall goodness of fit and robustness compared to using AP data alone.
Conclusions:
- Incorporating membrane resistance (Rm) fitting into the objective function alongside action potential morphology significantly enhances the efficiency and robustness of cardiac ionic model parameterization.
- This improved fitting strategy leads to reduced parameter uncertainty and better preservation of tissue-level behavior.
- The study recommends including Rm as an objective in fitting protocols for more reliable cardiac modeling.
