Assessment of the value of wavelet analysis of Holter recordings for the prediction of sudden cardiac death

The aim of this paper is to validate wavelet analysis of Holter recordings as a tool for the detection of risk of sudden cardiac death for patients surviving an acute myocardial infarction. The study uses time-averaged HR-ECGs from the European Myocardial Infarct Amiodarone Trial (EMIAT). Each HR-ECG is transformed into 511 orthogonal Meyer wavelet coefficients, extending from 128 ms before to 384 ms after QRS onset, and their value is assessed by means of the CARTEF Time-Frequency Abnormalities Stratification method. We then perform a linear discriminant analysis to assess the discriminant power of all wavelet coefficients with a significant p-value (p<0.0001), and we compare them to the clinical parameters and also to a combination of them.