Problem of power spectra estimation in application to the analysis of heart rate variability

• Published in: The European physical journal. ST, Special topics Vol. 232; no. 5; pp. 615 - 624
• Main Authors: Ishbulatov, Yurii M., Gridnev, Vladimir I., Ponomarenko, Vladimir I., Ezhov, Dmitry M., Prokhorov, Mikhail D., Kiselev, Anton R., Karavaev, Anatoly S.
• Format: Journal Article
• Language: English
• Published: Berlin/Heidelberg Springer Berlin Heidelberg 01.05.2023; Springer Nature B.V
• Subjects: Atomic; Autoregressive models; Brain Physiology Meets Complex Systems; Classical and Continuum Physics; Condensed Matter Physics; Females; Fourier transforms; Heart rate; Hypertension; Materials Science; Measurement Science and Instrumentation; Molecular; Optical and Plasma Physics; Parameter sensitivity; Parameterization; Patients; Physics; Physics and Astronomy; Power spectra; Regular Article; Sensitivity; Spectrum analysis; Standard deviation; Time series; Window functions
• ISSN: 1951-6355; 1951-6401
• DOI: 10.1140/epjs/s11734-022-00731-2

We investigated how the parameters of the spectral analysis affect standard deviation and error of the estimation of well-known indices for the heart rate variability. We compared the nonparametric Fourier transform to the parametric approach based on autoregressive models. We also investigated how the precision of the indices estimation depends on the choice of the window function, parameterization of the Bartlett’s method, and the lengths of time series. For each set of parameters, we calculated the sensitivity and specificity of the resulting indices when diagnosing arterial hypertension. To isolate and investigate the errors caused by inaccuracy of the spectral analysis itself, we conducted our study using the mathematical models of heart rate variability for healthy subjects and arterial hypertension patients, for which the correct values of the spectral indices are known. The obtained results suggest that the analysis of 20-min signals, comparing to 5-min signals, significantly decreases the standard deviation of the estimations and increases both their sensitivity and specificity. We found no advantages of using the parametric approach over the Fourier transform. We have shown that application of the Hann’s window function and normalization of the spectral indices decreases the sensitivity and specificity of the medical diagnostics.