Karhunen-Loeve representation of ECG data

Quantitative representation of mathematical functions or random data provides a useful and often necessary step for analyzing complex physical or physiological phenomena. In general, the representation process converts complex functions or observations into weighted combinations of simple, elementary components. Rationale for use of representation lies in the simplification of restructuring the original functions or data in terms of these basic elements, which can facilitate understanding or analysis of the phenomena under study. Representation provides a basis for feature extraction in classification problems, filtering for noise reduction, and information assessment. In this paper the author describes some of the classical representation methods and demonstrates specific implementation of the statistical representation of electrocardiographic data using the Karhunen-Loeve method.