Metody pro analýzu vektorkardiografických signálů za účelem predikování infarktu myokardu

Abstract

The aim of this master thesis is the analysis of Vectorcardiographic records measured according to Frank lead systém in order to predict Myocardial Infarction. For this purpose, pre-processing of physiological and heart attack records including the calculation of the average beat was carried out, from which selected symptoms from the QRS and T loop were subsequently extracted. Their informative value was verified by statistical methods. To classify subjects using combination of these features, machine learning Methods such as linear and quadratic discriminant analysis, linear and cubic support vector methods, and the nearest neighbor method were applied. The practical part was implemented in MATLAB.