Mortality Prediction in Severe Congestive Heart Failure Patients With Multifractal Point-Process Modeling of Heartbeat Dynamics

Abstract

Multifractal analysis of human heartbeat dynamics has been demonstrated to provide promising markers of Congestive Heart Failure (CHF). Yet, it crucially builds on the interpolation of RR intervals series, which has been generically performed with limited links to CHF pathophysiology. We devise a novel methodology estimating multifractal autonomic dynamics from heartbeat-derived series defined in the continuous time. We hypothesize that markers estimated from our novel framework are also effective for mortality prediction in severe CHF. We merge multifractal analysis within a methodological framework based on inhomogeneous point process models of heartbeat dynamics. Specifically, wavelet coefficients and wavelet leaders are computed over measures extracted from instantaneous statistics of probability density functions characterizing and predicting the time until the next heartbeat event occurs. The proposed approach is tested on data from 94 CHF patients, aiming at predicting survivor and non-survivor individuals as determined after a 4 years follow up. Instantaneous markers of vagal and sympatho-vagal dynamics display power-law scaling for a large range of scales, from s to s. Using standard SVM algorithms, the proposed inhomogeneous point-process representation based multifractal analysis achieved the best CHF mortality prediction accuracy of 79.11 % (sensitivity 90.48%, specificity 67.74%). Our results suggest that heartbeat scaling and multifractal properties in CHF patients are not generated at the sinus-node level, but rather by the intrinsic action of vagal short-term control and of sympatho-vagal fluctuations associated with circadian cardiovascular control, especially within the VLF band. These markers might provide critical information in devising a clinical tool for individualized prediction of survivor and non-survivor CHF patients.

Heartbeat Dynamics Analysis under Cold-Pressure Test using Wavelet p-Leader Non-Gaussian Multiscale Expansions

“…In fact, statistical differences, yet expected, between resting and CPT sessions were associated with point-process derived series exclusively, while no significant differences were found on raw heartbeat data. This finding is in agreement with previous evidences [6] showing the importance of a proper interpolation prior to MF analysis.…”

“…Such an interaction occurs at functional, anatomical, and biochemical levels, thus motivating the scientific community to investigate signal processing methods going beyond linearity defined in the time and frequency domains [1], [3]. Common nonlinear estimates include entropy rates and non-Gaussian metrics [2]- [5], as well as multiscale and fractal indices [2], [3], [6], [7].…”

“…Recently we have proven the reliability of a multiscale MF analysis on heartbeat dynamics when properly combined with inhomogeneous point-process modelling [6]. This model interpolates the heartbeat interval series through physiologically-plausible probability density functions characterizing and predicting the timing of each event [17].…”

Investigating Phasic Activity of Time-Varying High-Order Spectra: A Heartbeat Dynamics Study During Cold-Pressure Test

“…Consequently, standard HRV time and frequency analyses, which quantify linear dynamics exclusively, are not enough to fully characterize the cardiac system, and need to be complemented by measurements from nonlinear system theory [2,3]. Of note, many psychophysiological and pathophysiological states have been successfully assessed by nonlinear heartbeat measures [2][3][4][5]. Exemplarily, good predictors of mortality following myocardial infarct or heart failure are entropy and multifractal metrics [2,4,5].…”

“…Of note, many psychophysiological and pathophysiological states have been successfully assessed by nonlinear heartbeat measures [2][3][4][5]. Exemplarily, good predictors of mortality following myocardial infarct or heart failure are entropy and multifractal metrics [2,4,5].…”