Multivariable approach to dynamic ECG classification

The first step towards a heart attack is a condition called Ventricular Arrhythmia. This paper proposes a system that uses the advanced digital signal processing techniques to analyse electrocardiogram (ECG) signals and recognise the Arrhythmia condition. In addition, proposed method can differentiate the more dangerous condition, Ventricular Arrhythmia from a simple Arrhythmia. The proposed technique combines the classical ECG signal parameters (e.g. Heart Rate Variability) with the standard statistical signal parameters, nonlinear parameters used in the fields of Chaos Theory and parameters obtained using Symbolic Analysis techniques. Linear Discriminant Analysis (LDA) is employed in order to reduce the size of ECG parameter set and is followed by a clustering algorithm.