Fig. 5

“Collapse of complexity” with severe pathology. The MSP plots are derived from a healthy subject in sinus rhythm (top), a patient with chronic heart failure (CHF) in sinus rhythm (middle) and a patient with atrial fibrillation (AF) (bottom). The MSP plots correspond to the original time series (left column) and their derived coarse-grained time series for scales 5 (second column), 10 (third column) and 15 (right column). The “wedge” shaped appearance of the AF plots at lower scales relates to the constraints on physiologic conduction imposed by refractoriness of the AV node. (The sinus rhythm time series of the healthy subject and the patient with CHF syndrome were filtered to remove outliers using http://physionet.org/tutorials/hrv-toolkit/HRV.src/filt.c, and visual assistance). We note that the scales correspond to slightly different mean rates in each subject. For example, scale 10 corresponds to means of 10 points. Thus, if the RR mean is around ~1 s (healthy case) this coarse-grained time scale will be ~10 s, but when the mean RR is around 0.6 s (AF case), the same time scale will correspond to ~6 s. (The MSP plots are from subjects # n2nn, c3nn and a5nn, respectively, from the PhysioNet database available at http://www.physionet.org/challenge/chaos/) [33]