Discriminating noise from chaos in heart rate variability : application to prognosis in heart failure

Abstract

This thesis examines two challenging problems in chaos analysis: distinguishing deterministic chaos and stochastic (noise-induced) chaos, and applying chaotic heart rate variability (HRV) analysis to the prognosis of mortality in congestive heart failure (CHF). Distinguishing noise from chaos poses a major challenge in nonlinear dynamics theory since the addition of dynamic noise can make a non-chaotic nonlinear system exhibit stochastic chaos, a concept which is not well-defined and is the center of heated debate in chaos theory. A novel method for detecting dynamic noise in chaotic series is proposed in Part I of this thesis. In Part II, we show that linear and nonlinear analyses of HRV yield independent predictors of mortality. Specifically, sudden death is best predicted by frequency analysis whereas nonlinear and chaos indices are more selective for progressive pump failure death. These findings suggest a novel noninvasive probe for the clinical management of CHF patients.