Nonlinear Modeling of Physiological Systems Using Principal Dynamic Modes
The use of a small number of properly selected principal dynamic modes (PDM) may provide the long-sought parsimony and physiological interpretability of nonlinear models of physiological systems obtained from stimulus-response experimental data. The advocated approach makes use of a filter-bank comprised of the selected FDM’s that are feeding into a multi-input static nonlinear operator. This model form facilitates the system identification task and makes nonlinear modeling possible in certain cases previously thought intractable. Since the estimation of the PDM’s is based only on 1st-order and 2nd-order kernel measurements, and the estimation of the static nonlinearity can be practically extended to high-order systems, it is hoped that this modeling method will broaden the scope of Volterra-Wiener applications and facilitate the interpretation of the obtained results. This may promote the broader use of nonlinear models in studies of physical and physiological systems.
KeywordsElliptic System Volterra Series Volterra Kernel System Kernel Laguerre Function
Unable to display preview. Download preview PDF.
- 5.Marmarelis, V.Z. (ed.) (1987). Advanced Methods of Physiological System Modeling, Vol. I, University of Southern California, Biomedical Simulations Resource, Los Angeles, California.Google Scholar
- 13.Ogura, H. (1985). Estimation of Wiener kernels of a nonlinear system and a fast algorithm using digital Laguerre filters. 15th NIBB Conf., pp. 14–62, Okazaki, Japan.Google Scholar
- 16.Sutter, E.E. (1987). A practical nonstochastic approach to nonlinear time-domain analysis. In: Advanced Methods of Physiological System Modeling, Vol. I, University of Southern Califonia, Biomedical Simualtions Resource, Los Angeles, California, pp. 303–315.Google Scholar
- 18.Volterra, V. (1930). Theory of Functional and of Integral and Integro-differential Equations, Dover, New York, New York.Google Scholar