Kullback-Leibler clustering of continuous wavelet transform measures of heart rate variability

Donald E. Mager, Marcellus M. Merritt, Jyotsna Kasturi, Lisa R. Witkin, Mirna Urdiqui-Macdonald, John J. Sollers, Michele K. Evans, Alan B. Zonderman, Darrell R. Abernethy, Julian F. Thayer

Research output: Contribution to journalArticle

Abstract

Power spectral analysis of beat-to-beat heart rate variability (HRV) has provided a useful means of understanding the interplay between autonomic and cardiovascular functionality. Despite their utility, commonly employed frequency-domain techniques are limited in their prerequisite for stationary signals and their inability to account for temporal changes in the power spectral and/or frequency properties of signals. The purpose of this study is to develop an algorithm that utilizes confinuous wavelet transform (CWT) parameters as inputs to a Kohonen self-organizing map (SOM), providing a method of clustering subjects with similar wavelet transform signatures. Continuous interbeat-intervals were recorded (Portapres monitor at 200Hz) during a perception of affect test in 79 African-American volunteers (ages 21-83). where after a 5-min baseline, participants evaluated emotional expressions in sentences and pictures of faces, followed by a 5-min recovery. Individual HRV biosignals from each session were pre-processed (artifact replacement and signal resampling at 2Hz) and a CWT was applied (db9 wavelet basis function over 32 scales). Standard deviations of resulting wavelet coefficients at each scale were calculated, normalized, and used as inputs into a SOM with Kullback-Leibler divergence as the dissimilarity measure used for clustering. Differences in subject demographics between two final clusters were assessed via two-independent-groups t-tests or chi-square or Fisher's exact tests of contingency tables. Significant differences were found for age, initial systolic blood pressure, smoking status, and mean s.d. of coefficients in the high frequency band (0.15-0.4Hz). These findings may have clinical significance and the developed algorithm provides an alternative means of analyzing HRV data originating from populations with complex covariates.

Original languageEnglish (US)
Pages (from-to)337-342
Number of pages6
JournalBiomedical Sciences Instrumentation
Volume40
StatePublished - 2004
Externally publishedYes

Fingerprint

Wavelet Analysis
Wavelet transforms
Cluster Analysis
Heart Rate
Self organizing maps
Blood Pressure
Blood pressure
Chi-Square Distribution
African Americans
Artifacts
Spectrum analysis
Frequency bands
Volunteers
Smoking
Demography
Recovery
Population
Power (Psychology)

Keywords

  • African-Americans
  • Continuous wavelet transform
  • Heart rate variability
  • Self-organizing maps

ASJC Scopus subject areas

  • Hardware and Architecture

Cite this

Mager, D. E., Merritt, M. M., Kasturi, J., Witkin, L. R., Urdiqui-Macdonald, M., Sollers, J. J., ... Thayer, J. F. (2004). Kullback-Leibler clustering of continuous wavelet transform measures of heart rate variability. Biomedical Sciences Instrumentation, 40, 337-342.

Kullback-Leibler clustering of continuous wavelet transform measures of heart rate variability. / Mager, Donald E.; Merritt, Marcellus M.; Kasturi, Jyotsna; Witkin, Lisa R.; Urdiqui-Macdonald, Mirna; Sollers, John J.; Evans, Michele K.; Zonderman, Alan B.; Abernethy, Darrell R.; Thayer, Julian F.

In: Biomedical Sciences Instrumentation, Vol. 40, 2004, p. 337-342.

Research output: Contribution to journalArticle

Mager, DE, Merritt, MM, Kasturi, J, Witkin, LR, Urdiqui-Macdonald, M, Sollers, JJ, Evans, MK, Zonderman, AB, Abernethy, DR & Thayer, JF 2004, 'Kullback-Leibler clustering of continuous wavelet transform measures of heart rate variability', Biomedical Sciences Instrumentation, vol. 40, pp. 337-342.
Mager DE, Merritt MM, Kasturi J, Witkin LR, Urdiqui-Macdonald M, Sollers JJ et al. Kullback-Leibler clustering of continuous wavelet transform measures of heart rate variability. Biomedical Sciences Instrumentation. 2004;40:337-342.
Mager, Donald E. ; Merritt, Marcellus M. ; Kasturi, Jyotsna ; Witkin, Lisa R. ; Urdiqui-Macdonald, Mirna ; Sollers, John J. ; Evans, Michele K. ; Zonderman, Alan B. ; Abernethy, Darrell R. ; Thayer, Julian F. / Kullback-Leibler clustering of continuous wavelet transform measures of heart rate variability. In: Biomedical Sciences Instrumentation. 2004 ; Vol. 40. pp. 337-342.
@article{da2c148a911a485abefd0e97c616d75d,
title = "Kullback-Leibler clustering of continuous wavelet transform measures of heart rate variability",
abstract = "Power spectral analysis of beat-to-beat heart rate variability (HRV) has provided a useful means of understanding the interplay between autonomic and cardiovascular functionality. Despite their utility, commonly employed frequency-domain techniques are limited in their prerequisite for stationary signals and their inability to account for temporal changes in the power spectral and/or frequency properties of signals. The purpose of this study is to develop an algorithm that utilizes confinuous wavelet transform (CWT) parameters as inputs to a Kohonen self-organizing map (SOM), providing a method of clustering subjects with similar wavelet transform signatures. Continuous interbeat-intervals were recorded (Portapres monitor at 200Hz) during a perception of affect test in 79 African-American volunteers (ages 21-83). where after a 5-min baseline, participants evaluated emotional expressions in sentences and pictures of faces, followed by a 5-min recovery. Individual HRV biosignals from each session were pre-processed (artifact replacement and signal resampling at 2Hz) and a CWT was applied (db9 wavelet basis function over 32 scales). Standard deviations of resulting wavelet coefficients at each scale were calculated, normalized, and used as inputs into a SOM with Kullback-Leibler divergence as the dissimilarity measure used for clustering. Differences in subject demographics between two final clusters were assessed via two-independent-groups t-tests or chi-square or Fisher's exact tests of contingency tables. Significant differences were found for age, initial systolic blood pressure, smoking status, and mean s.d. of coefficients in the high frequency band (0.15-0.4Hz). These findings may have clinical significance and the developed algorithm provides an alternative means of analyzing HRV data originating from populations with complex covariates.",
keywords = "African-Americans, Continuous wavelet transform, Heart rate variability, Self-organizing maps",
author = "Mager, {Donald E.} and Merritt, {Marcellus M.} and Jyotsna Kasturi and Witkin, {Lisa R.} and Mirna Urdiqui-Macdonald and Sollers, {John J.} and Evans, {Michele K.} and Zonderman, {Alan B.} and Abernethy, {Darrell R.} and Thayer, {Julian F.}",
year = "2004",
language = "English (US)",
volume = "40",
pages = "337--342",
journal = "Biomedical Sciences Instrumentation",
issn = "0067-8856",
publisher = "ISA - Instrumentation, Systems, and Automation Society",

}

TY - JOUR

T1 - Kullback-Leibler clustering of continuous wavelet transform measures of heart rate variability

AU - Mager, Donald E.

AU - Merritt, Marcellus M.

AU - Kasturi, Jyotsna

AU - Witkin, Lisa R.

AU - Urdiqui-Macdonald, Mirna

AU - Sollers, John J.

AU - Evans, Michele K.

AU - Zonderman, Alan B.

AU - Abernethy, Darrell R.

AU - Thayer, Julian F.

PY - 2004

Y1 - 2004

N2 - Power spectral analysis of beat-to-beat heart rate variability (HRV) has provided a useful means of understanding the interplay between autonomic and cardiovascular functionality. Despite their utility, commonly employed frequency-domain techniques are limited in their prerequisite for stationary signals and their inability to account for temporal changes in the power spectral and/or frequency properties of signals. The purpose of this study is to develop an algorithm that utilizes confinuous wavelet transform (CWT) parameters as inputs to a Kohonen self-organizing map (SOM), providing a method of clustering subjects with similar wavelet transform signatures. Continuous interbeat-intervals were recorded (Portapres monitor at 200Hz) during a perception of affect test in 79 African-American volunteers (ages 21-83). where after a 5-min baseline, participants evaluated emotional expressions in sentences and pictures of faces, followed by a 5-min recovery. Individual HRV biosignals from each session were pre-processed (artifact replacement and signal resampling at 2Hz) and a CWT was applied (db9 wavelet basis function over 32 scales). Standard deviations of resulting wavelet coefficients at each scale were calculated, normalized, and used as inputs into a SOM with Kullback-Leibler divergence as the dissimilarity measure used for clustering. Differences in subject demographics between two final clusters were assessed via two-independent-groups t-tests or chi-square or Fisher's exact tests of contingency tables. Significant differences were found for age, initial systolic blood pressure, smoking status, and mean s.d. of coefficients in the high frequency band (0.15-0.4Hz). These findings may have clinical significance and the developed algorithm provides an alternative means of analyzing HRV data originating from populations with complex covariates.

AB - Power spectral analysis of beat-to-beat heart rate variability (HRV) has provided a useful means of understanding the interplay between autonomic and cardiovascular functionality. Despite their utility, commonly employed frequency-domain techniques are limited in their prerequisite for stationary signals and their inability to account for temporal changes in the power spectral and/or frequency properties of signals. The purpose of this study is to develop an algorithm that utilizes confinuous wavelet transform (CWT) parameters as inputs to a Kohonen self-organizing map (SOM), providing a method of clustering subjects with similar wavelet transform signatures. Continuous interbeat-intervals were recorded (Portapres monitor at 200Hz) during a perception of affect test in 79 African-American volunteers (ages 21-83). where after a 5-min baseline, participants evaluated emotional expressions in sentences and pictures of faces, followed by a 5-min recovery. Individual HRV biosignals from each session were pre-processed (artifact replacement and signal resampling at 2Hz) and a CWT was applied (db9 wavelet basis function over 32 scales). Standard deviations of resulting wavelet coefficients at each scale were calculated, normalized, and used as inputs into a SOM with Kullback-Leibler divergence as the dissimilarity measure used for clustering. Differences in subject demographics between two final clusters were assessed via two-independent-groups t-tests or chi-square or Fisher's exact tests of contingency tables. Significant differences were found for age, initial systolic blood pressure, smoking status, and mean s.d. of coefficients in the high frequency band (0.15-0.4Hz). These findings may have clinical significance and the developed algorithm provides an alternative means of analyzing HRV data originating from populations with complex covariates.

KW - African-Americans

KW - Continuous wavelet transform

KW - Heart rate variability

KW - Self-organizing maps

UR - http://www.scopus.com/inward/record.url?scp=2342523142&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=2342523142&partnerID=8YFLogxK

M3 - Article

C2 - 15133981

AN - SCOPUS:2342523142

VL - 40

SP - 337

EP - 342

JO - Biomedical Sciences Instrumentation

JF - Biomedical Sciences Instrumentation

SN - 0067-8856

ER -