Extracting cardiac dynamics within ECG signal for human identification and cardiovascular diseases classification.

Cardiac characteristics underlying the time/frequency domain features are limited and not comprehensive enough to reflect the temporal/dynamical nature of ECG patterns. This paper proposes a dynamical ECG recognition framework for human identification and cardiovascular diseases classification via a dynamical neural learning mechanism. The proposed method consists of two phases: a training phase and a test phase. In the training phase, cardiac dynamics within ECG signals is extracted (approximated) accurately by using radial basis function (RBF) neural networks through deterministic learning mechanism. The obtained cardiac system dynamics is represented and stored in constant RBF networks. An ECG signature is then derived from the extracted cardiac dynamics along the periodic ECG state trajectories. A bank of estimators is constructed using the extracted cardiac dynamics to represent the trained gait patterns. In the test phase, recognition errors are generated and taken as the similarity measure by comparing the cardiac dynamics of the trained ECG patterns and the dynamics of the test ECG pattern. Rapid recognition of a test ECG pattern begins with measuring the state of test pattern, and automatically proceeds with the evolution of the recognition error system. According to the smallest error principle, the test ECG pattern can be rapidly recognized. This kind of cardiac dynamics information represents the beat-to-beat temporal change of ECG modifications and the temporal/dynamical nature of ECG patterns. Therefore, the amount of discriminability provided by the cardiac dynamics is larger than the original signals. This paper further discusses the extension of the proposed method for cardiovascular diseases classification. The constructed recognition system can distinguish and assign dynamical ECG patterns to predefined classes according to the similarity of cardiac dynamics. Experiments are carried out on the FuWai and PTB ECG databases to demonstrate the effectiveness of the proposed method.

Relationship Between Persistent Excitation Levels and RBF Network Structures, With Application to Performance Analysis of Deterministic Learning.

Based on the notion of persistent excitation (PE), a deterministic learning theory is recently proposed for RBF network-based identification of nonlinear systems. In this paper, we study the relationship between the PE levels, the structures of RBF networks and the performance of deterministic learning. Specifically, given a state trajectory generated from a nonlinear dynamical system, we investigate how to construct the RBF networks in order to guarantee sufficient PE levels (especially the level of excitation) for deterministic learning. It is revealed that the PE levels decrease with the density of neural centers, denoted by explicit formulas. As an illustration, these formulas are applied to convergence analysis of deterministic learning. We present exact theoretical conclusions that a finite and definite number of centers can achieve the same performance as global centers. In addition, a tradeoff exists between a relatively high level of excitation and the good approximation capabilities of RBF networks, which indicates that we cannot always obtain better convergence accuracy by increasing the density of centers. These results provide a new perspective for performance analysis of RBF network algorithms based on the notion of PE. Simulation studies are included to illustrate the results.