Premium
Recurrence Time Ratio Slope (RTRS) to estimate nonstationarity in dynamic nonlinear heart rate variability signals
Author(s) -
Wilkins Brek,
Bukkapatnam Satish,
Komanduri Ranga,
Benjamin Bruce A
Publication year - 2011
Publication title -
the faseb journal
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 1.709
H-Index - 277
eISSN - 1530-6860
pISSN - 0892-6638
DOI - 10.1096/fasebj.25.1_supplement.1130.4
Subject(s) - recurrence quantification analysis , recurrence plot , statistic , nonlinear system , heart rate variability , orthostatic vital signs , statistics , cluster analysis , variance (accounting) , mathematics , computer science , pattern recognition (psychology) , control theory (sociology) , artificial intelligence , medicine , heart rate , control (management) , physics , accounting , quantum mechanics , blood pressure , business
One of the most challenging problems facing complex systems researchers is correctly analyzing nonlinear signals that contain nonstationarities (shifts in mean and/or variance). Nonstationarity (NS) often leads to specious nonlinear statistics that can make it difficult to compare signals. In physiology these segments are intrinsic to the nonlinear control system and need to be quantified. We have developed a novel statistic, recurrence time ratio slope (RTRS) , based on the relationship of recurrence times of the first and second type that estimates NS. RTRS measures how recurrent states distribute within a recurrence plot as the % fixed amount of neighbors is increased from 2.5 to 20%. We tested this new statistic on two different Lorenz signals with identical amounts of NS that were previously indistinguishable using recurrence quantification analysis (RQA). Our results reveal two distinct clusters formed by RQA statistics vs. RTRS . Additionally, we applied RTRS to continuous heart rate variability (HRV) data during orthostatic challenge tests and are able to track NS. HRV NS can statistically (i.e. RQA) mimic clinical pathology. Therefore, it was intriguing to find a unique clustering of healthy stationary and nonstatrionary segment RQA statistics vs RTRS . Our findings indicate that RTRS is an excellent tool for researchers interested in estimating NS in dynamically complex signals.