Estimation of cardiorespiratory transfer under spontaneous breathing conditions: a theoretical study

Abstract

Using simulated noisy sequences of respiration and heart rate, we assessed the accuracy of the respiratory sinus arrhythmia transfer function (RSATF) estimation under three kinds of spontaneous breathing patterns: regular or tidal breathing, periodic breathing with apnea, and broadband breathing. Estimation employing the cross-power and autopower spectra of the simulated data produced RSATF estimates that were generally more variable than those computed with an autoregressive modeling approach. Variability and bias errors in the RSATF estimates became larger as respiratory bandwidth decreased when the breathing pattern changed from broadband to periodic to regular breathing. However, between frequencies of 0.1 and 0.3 Hz, these errors fell within 12% in all breathing patterns. Error in the RSATF estimates was only slightly increased, with reductions in data length to as low as 90 s. The results suggest the feasibility of obtaining accurate estimates of RSATF between 0.1 and 0.8 Hz from a wide variety of conditions, such as in different sleep-wake states where voluntary control of breathing is not possible and the ventilatory pattern may vary substantially.