| Summary: | 博士 === 國立陽明大學 === 醫學工程研究所 === 101 === This paper explored the application of the Volterra series model in the estimation of baroreflex sensitivity (BRS). A spontaneous BRS analysis-the VL technique was put forth through the use of two measurement variables, namely, systolic blood pressure (SBP) and RR interval (RRI) and based on the nonlinear dynamic behavior of baroreflex regulation. This technique uses changes in SBP (ΔSBP) and changes in RRI (ΔRRI) as input and output. Through the nonlinear Volterra-Laguerre model, the baroreflex system modeling analysis was conducted to derive at the BRS values and other new indicators pertaining to the baroreflex system characteristics-the system setup time parameter and nonlinearity parameter, and based on Volterra kernels the activity of the sympathetic and parasympathetic nerves in the baroreflex system was isolated.
The VL technique can effectively rule out measured data drift and measure the baroreflex system delay and frequency response, which possess the time-domain analysis and frequency-domain analysis characteristics. After adopting numerical analog data, measured data, and EuroBaVar database data as objects for analysis, it shows that this technique not only accurately derives at BRS estimation values close to those obtained through the pharmacologic technique, but is also able to process physiological signals that fail to be processed through the existing spontaneous BRS techniques and provide related indicators that are valuable to the baroreflex mechanism.
With the respiratory variables taken into account, the Volterra series model can be supplemented by the complex demodulation method and AR-Volterra method to explicitly or implicitly process the effects of respiration on baroreflex, thereby circumventing the problem of SBP and respiratory input signal dependency when using the dual-input and single-output Volterra model. It is expected that the Volterra kernels of the baroreflex system can be more accurately estimated. However, since baroreflex is respiration-driven and that complex couplings of the blood pressure regulation mechanism takes place, actual data analysis shows that the complex demodulation method underestimated the effects caused by respiration, while the AR-Volterra method overestimated the effects. On the other hand, the VL technique that avoids the respiration band (HF band) during data fitting contributed to the acquisition of stable intermediate values. Based on this finding, it is suggested that the respiratory variables be processed with more caution. In addition, during the spontaneous BRS analysis, it is recommended that two variables, namely, SBP and RRI be adopted.
|
|---|
