A prediction model for the number of intrinsic mode functions in biomedical signals: The case of electrocardiogram

In this paper, the open issue of prediction of the number of intrinsic mode functions (IMF) extracted from a time series after the application of empirical mode decomposition (EMD) is addressed and a methodology is presented directing towards an a priori prediction model. Parameters related with the time series are measured or calculated and used by the model in order to define a closed set in which the actual total number of IMFs is expected to be included after the application of EMD. The prediction model is verified by a large number of tests on simulated electrocardiogram (ECG) time series and after refinement it is validated using real ECG time series from Physionet MIT-BIH database.