A foundation of the stochastic representation of Electrocardiograms

Abstract

The purpose of this report is to provide a normal electrocardiogram (ECG) with a stochastic representation and to get various new techniques for character detecting in the numerical diagnosis of ECG by digital computers.
An individual ECG can be assumed as a continuous functions on a closed unit interval [0, 1], in which an R-wave voltage peak had to be located at 0.4.The sequence of its values at k/100 (k=0, 1, …, 99) can be regarded as a hundred dimensional vector.
After the computer calculations;(1) we obtained sample vectors from normal EGG in lead II out of 69 normals and their mean vector x= (x₁, …, x₁₀₀) as well as the standard deviation σ= (σ1, …, σ₁₀₀).
(2) It was statistically tested that each i-th component xi obtained from normal vector was normally distributed around the averagex_i.
(3) Using some additional vectors from normal ECGs and those from myocardial infarctions, the distance between the vector and the mean vector x and the extent of the eccentricity were calculated.Values of eccentricity were zero in almost all components from the normal ECG. On the other hand, in ECG from the myocardial infarctions, there could be observed many extremely high values.Concerning distances, we had results the same as for the case of eccentricities.