Parameter identifiability of cardiac ionic models using a novel CellML least squares optimization tool.

ABSTRACT

Published models of excitable cells can be used to fit to a range of action potential experimental data. CellML is a well-defined standard for publishing and exchanging such models, but currently there is a lack of software that utilizes CellML for parameter analysis. In this paper, we introduce a Java-based utility capable of performing model simulation, identifiability analysis, and parameter optimization of ionic cardiac cell models written in CellML. Identifiability analysis was performed in seven CellML models. Parameter identifiability was consistently improved by using the compensatory membrane current as opposed to the membrane voltage as the residual. as well as through the introduction of an additional stimulus set used in the fitting process.