Assessing the convolutedness of multivariate physiological time series