Personalization of mathematical models of human atrial action potential

Atrial cardiomyocytes demonstrate a wide spectrum of patient-specific, tissue-specific, and pathology-specific action potential (AP) phenotypes due to differences in protein expression and posttranslational modifications. Accurate simulation of the AP excitation and propagation in healthy or diseased atria requires a mathematical model capable of reproducing all the differences by parameter rescaling. In the present study, we have benchmarked two widely used electrophysiological models of the human atrium: the Maleckar and the Grandi models. In particular, patch-clamp AP recordings from human atrial myocytes were fitted by the genetic algorithm (GA) to test the models' versatility. We have shown that the Maleckar model results in a more accurate fitting of heart rate dependence of action potential duration (APD) and resting potential (RP). On the other hand, both models demonstrate the poor fitting of the plateau phase and spike-and-dome morphologies. We propose that modifications to L-type calcium current-voltage relationships are required to improve atrial models' fidelity.