Open AccessComparison of nonparametric and parametric methods for time-frequency heart rate variability analysis in a rodent model of cardiovascular disease
Author(s) -
Emily Wong,
Fern Tablin,
Edward S. Schelegle
Publication year - 2020
Publication title -
plos one
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.99
H-Index - 332
ISSN - 1932-6203
DOI - 10.1371/journal.pone.0242147
Subject(s) - autoregressive model , spectral density , nonparametric statistics , statistics , heart rate variability , fourier transform , discrete fourier transform (general) , parametric statistics , short time fourier transform , mathematics , fourier analysis , heart rate , medicine , mathematical analysis , blood pressure
The aim of time-varying heart rate variability spectral analysis is to detect and quantify changes in the heart rate variability spectrum components during nonstationary events. Of the methods available, the nonparametric short-time Fourier Transform and parametric time-varying autoregressive modeling are the most commonly employed. The current study (1) compares short-time Fourier Transform and autoregressive modeling methods influence on heart rate variability spectral characteristics over time and during an experimental ozone exposure in mature adult spontaneously hypertensive rats, (2) evaluates the agreement between short-time Fourier Transform and autoregressive modeling method results, and (3) describes the advantages and disadvantages of each method. Although similar trends were detected during ozone exposure, statistical comparisons identified significant differences between short-time Fourier Transform and autoregressive modeling analysis results. Significant differences were observed between methods for LF power ( p ≤ 0.014); HF power ( p ≤ 0.011); total power ( p ≤ 0.027); and normalized HF power ( p = 0.05). Furthermore, inconsistencies between exposure-related observations accentuated the lack of agreement between short-time Fourier Transform and autoregressive modeling overall. Thus, the short-time Fourier Transform and autoregressive modeling methods for time-varying heart rate variability analysis could not be considered interchangeable for evaluations with or without interventions that are known to affect cardio-autonomic activity.