MULTIVARIATE TIME-VARIANT IDENTIFICATION OF CARDIOVASCULAR VARIABILITY SIGNALS - A BEAT-TO-BEAT SPECTRAL PARAMETER-ESTIMATION IN VASOVAGAL SYNCOPE

In this paper a bivariate, time-variant model able to continuously mea sure the mutual interactions between heart rate and systolic blood pre ssure variability signals is presented, A recursive identification of the model parameters makes it possible to estimate, on a beat-to-beat basis, spectral low-frequency (LF) and high-frequency (HF) power, (LF/ HF ratio) and cross-spectral (coherence and phase relationships betwee n spectral peaks) indexes during nonstationary events, These indexes c an be helpful in: 1) physiological study of autonomic nervous system m echanisms of cardiovascular central and 2) quantification and clinical evaluation of the neural and mechanical links between the two signals , in addition, an estimate of baroreceptive activation (alpha-gain) is continuously extracted, Before applying the model to cardiovascular s ignals, the reliability of tile estimated parameters was tested on sim ulated signals, Subsequently, the model was applied to investigating v asovagal syncope episodes, aiming at the assessment of autonomic nervo us system status and autonomic role in the dynamic phenomena which lea d to syncope, The proposed model, which provides noninvasive beat-to-b eat evaluation of the autonomic events, may be useful in the descripti on of the syncopal episodes and in tile comprehension of the complex p hysiological mechanisms of syncope.