Optimized Adjustment of Single Action-potentials to Case-specific Atrial Physiology: Towards Clinical Implementation

Abstract

We present a method for modelling case-specific action potentials (APs) based on a simple model for electric wave propagation combined with an optimization procedure that searches the best combination of three model parameter values for a given action potential. The method is applied to a set of 70 APs, showing deviations around 10%. This method can be useful for prognosing possible pathological scenarios of a specific patient, based on data from electrocardiagrams. IntroductionFor modeling the propagation of action potentials in the human atria, various models have been developed, which take into account the influence of the numerous ionic currents flowing through the cell membrane. Aiming at a simplified description, the Bueno-Orovio-Cherry-Fenton (BOCF) model [1] for electric wave propagation in the ventricle has been adapted to atrial physiology [2].Recently, we studied this adapted BOCF model with respect to its capability to accurately generate spatiotemporal excitation patterns found in anatomical and spiral wave reentry [3]. We found that characteristic features of the reentrant excitation patterns seen in the more detailed model proposed by Courtemanche, Ramirez and Nattel (CRN) model [4] are well captured by the BOCF model. In particular, we showed that the size and conduction velocity of the spiral waves produced with both models differ by a constant difference smaller than 10%. Moreover, we also compared APs generated with the BOCF model with APs produced with the CRN model [5] and discussed strengths of the BOCF model, namely its lower computational costs for extensive simulations and its reduced number of differential equations.A natural step forward is to use these strengths for parameter adjustment to patient-specific electrophysiological conditions. Here, we introduce such method to model specific action potentials with a pre-given amplitude, duration and shape by only adjusting a few parameters.The specific data to be analyzed comprehends a set of 103 APs collected at the University of Dresden in the group of Prof. U. Ravens. Figure 1(a) shows one AP that we characterize by its action potential amplitude (APA) and AP durations (APDs) at four amplitude levels, namely at 90%, 50%, 40% and 20% of the APA. The distribution of each of these five AP properties for the full set of APs