On the use of wavelets to separate dynamical activities embedded in a time series

In this work a first attempt is made to separate dynamical activity in signals by decomposing frequency sub-bands of the original observable signal. We use multiresolution wavelets to decompose electrocardiogram (ECG) signals into several frequency sub-bands and calculated the correlation dimension to determine the presence or absence of low-dimensional dynamics. Although the results must be considered as very preliminary, some evidence is presented to show that a low dimensinal activity precedes the actual incidence of the tachyarrhythmia. We hypothesize the signal decomposition by wavelet analysis may enhance the ability to locate low dimensional activity in cardiac arrhythmias.