Key points are not available for this paper at this time.
Extensions to a previously published nonlinear model for generating realistic artificial electrocardiograms to include blood pressure and respiratory signals are presented. The model accurately reproduces many of the important clinical qualities of these signals such as QT dispersion, realistic beat to beat variability in timing and morphology and pulse transit time. The advantage of this artificial model is that the signal is completely known (and therefore its clinical descriptors can be specified exactly) and contains no noise. Artifact and noise can therefore be added in a quantifiable and controlled manner in order to test relevant biomedical signal processing algorithms. Application examples using Independent Component Analysis to remove artifacts are presented.
Clifford et al. (Tue,) studied this question.