Parameter estimation of a phenomenological cardiac model based on a biophysically detailed model of human atria: A method for model complexity reduction using extended Kalman filter

Atrial fibrillation (Afib), the most common type of cardiac arrhythmia, arises from abnormal electrical activity of atrial membrane. Electrophysiological dynamics of human atrial tissue is represented by biophysically detailed models. Such models are really complex and hard to use for analysis purposes such as parameter estimation based on patient data. Hence, there is a need for simpler yet biophysically accurate and mathematically tractable models to be used for personalized simulation of electrical activity in human atria. Kalman filter, using the principle of recursive least square estimation, is a powerful tool for optimal estimation of states under different noise definitions. Model parameters can be augmented as state variables and can be estimated by Kalman filter. Kalman filter is also applicable for nonlinear systems with some modifications. One such widely use variant is extended Kalman filter (EKF). In this study, we propose a method for parameter estimation of a phenomenological cardiac model to match a targeted behavior first generated from a complex model, by using extended Kalman filter. Its performance is then compared to that of particle swarm optimization (PSO), a method that has been widely used for parameter optimization.