A methodological assessment of phase-rectified signal averaging through simulated beat-to-beat interval time series

Average cardiac acceleration (AC) and deceleration (DC) capacity, as computed by Phase-Rectified Signal Averaging (PRSA), were introduced to detect quasi-periodic oscillations in RR series. Calculation of AC and DC depends on three parameters (T, Land s). The aim of the study was to provide further insights on AC/DC and on the appropriate selection of these parameters. Numerical simulations were focused on: i) changing the frequency of the oscillations detected by AC/DC; ii) testing the difference between AC and DC on synthetic data generated by AR models, fitted on real RR series; and iii) the effect of different growing and decreasing trends (lack of time-reversal symmetry). When computed on series generated by AR models, AC and DC were quantitatively equivalent, independently of the power spectrum (p < 0.05). The parameter s, more than T, affected the results, while values of L > s were equivalent. In fact, s selected the oscillations to which AC/DC resulted maximally sensitive. On the contrary, sawtooth-like series, with different growth and decrease rates, showed a marked difference between AC and DC. AC and DC are not simply related to spectral contents. Indeed, AC and DC are linked to the asymmetries between the rates of growth and decrease of heart rate, and might quantify differently underlying regulatory mechanisms.