Analisi parametrica, multivariata, tempovariante dei segnali di variabilità nel sistema cardiovascolare

This thesis approaches the problem of short-term mechanisms that regulate Heart Rate (HRV), Blood Pressure (BPV) and Respiratory (RV) Variabilities, by focusing their evident interactions and the changes of their frequencies during transients. In this study we suggest some different models based on AutoRegressive (AR) multivariate identification of the interactions among the signals. These models take into account the specific roles played by the different signals in the complex control system that regulates the cardiovascular status allowing the estimation of several parameters of physiological interest. As an example, a good estimation of the baroreflex gain is obtained by using a closed-loop approach in the bivariate and trivariate models. In the AutoRegressive with eXogenous input (ARX) model the respiratory signal has the specific role of exogenous input to the HRV-BPV loop in good agreement with the physiology of the system. On the other hand, a time-variant algorithm based on an oblivion coefficient has been developed in order to follow the changes of the signal frequency contents during non-stationary episodes. The oblivion coefficient is chosen to be constant or variable in time according to the specific characteristics of the recorded series. All the algorithms successfully tested on simulated series have been used to analyse the recorded data according to the different protocols. In particular we performed a study on transient ischemic attacks, analysing 27 episodes of ischemia of unknown origin, X Syndrome (SX) and 30 episodes of Chronic Stable Angina (CSA). The results show that in SX there is an activation of the sympathetic system at the onset of the episode and the vagal reflex that usually ends the attack comes out later in respect to the CSA. In the other protocol 64 young Patients affected by Essential Hypertension (EHP) and 35 Normal Subjects (NS) have been studied during an Head Up Tilt Test (HUTT). The EHP showed a greater sympathetic activation during rest and all the indices associated with the baroreceptive gain showed a reduced sensibility of this mechanism in respect to the NS. The comparison of several indices of the baroreceptive gain extracted from the literature and evaluated on our data with the indices resulted from our models showed a better sensibility and performance of the gain when calculated from a trivariate closed-loop ARX model evaluated in the HF (high frequency) band. Important results have been achieved on the structure of the transfer functions involved in the models, suggesting, for future development, the possibility of reducing the parameters’ number. The application of these methods in clinical studies and the evaluation of the introduced parameters performances in different physiological and pathological situations will be not only possible but also useful for further studies.