ECG model parameters optimization and space state reconstruction

Abstract

The paper proposes a parameter optimization of a three-coupled oscillator model representing cardiac rhythm. First was detected a set of parameters which has a high or low influence on ECG signal. Then was selected a set of parameters that cause a difference in HRV signal. Hence, five parameters of Sinoatrial oscillator were designed for estimating. The optimization by particle swarm with inertia actualization presented parameters which created an ECG signal closer than real ECG. Afterward, in order to compare the ECG model with a normal ECG, the space state of a real ECG signal was reconstructed using two techniques: delay coordinates embedding and singular values decomposition (SVD). The results showed the feasibility of parameter optimization on ECG model and the benefits of using SVD to reconstruct the space state instead of delay coordinates method in this case of ECG signal.