Comparison of autoregression and fast Fourier transform techniques for power spectral analysis of heart period variability of persons with sudden cardiac arrest before and after therapy to increase heart period variability.

Clinical studies have used two types of analysis of the power spectral estimates of heart period variability: autoregressive and fast Fourier transform techniques. Controversy exists regarding which method is the most valid. The specific aims of this study are: (1) to describe the power spectra of heart period variability before and after an intervention designed to increase heart period variability in persons after sudden cardiac arrest; (2) to compare the integral of power spectral density between autoregressive and fast Fourier transform techniques within low and high-frequency bands; (3) to compare the magnitude of the spectral peak values determined by autoregressive and fast Fourier transform techniques approaches within low and high-frequency bands; and (4) to compare the aforementioned parameters using 4-minute, 1-hour, and 24-hour blocks of heart period data. Results indicated high correlations between spectral estimations by autoregressive and fast Fourier transform techniques using integrals or peak values within either the low or high-frequency ranges. The autoregressive technique demonstrated better resolution of sharp peaks than the fast Fourier transform technique, and makes a smoother, more interpretable curve. Lastly, people can cognitively change their heart rate variability.