Nonparametric block-structured modeling of rat lung mechanics |
| |
Authors: | Geoffrey N Maksym Jason H T Bates |
| |
Institution: | (1) Meakins-Christie Laboratories, McGill University, 3626 St. Urbain Street, H2X 2P2 Montréal, Québec, Canada;(2) Department of Biomedical Engineering, McGill University, Montréal, Québec, Canada |
| |
Abstract: | The quasistatic and dynamic pressure volume characteristics of the lungs were measured in five anesthetized, paralyzed open-chest
rats. Psuedo-random volume perturbations over a frequency range of 0.25 to 25 Hz and having peak-peak amplitudes of 1 to 4
ml were applied after the lungs were allowed to expire against 0.2, 0.4, 0.6, and 0.8 kPa positive endexpiratory pressure
(PEEP). The lung mechanics were partitioned in two ways: a linear dynamic block followed by a static nonlinearity (Wiener
model) and a static nonlinearity ahead of a linear dynamic block (Hammerstein model). It was found that a Hammerstein model
featuring a third-order polynomial static nonlinearity and a linear impulse response function of 1-sec duration accounted
for the greatest amount of the output variance (98.8±06%, mean±SD from perturbations of 4 ml amplitude and PEEP-0.8 kPa).
The static nonlinear behavior matched the measured quasistantic pressure volume behavior obtained at the same amplitude and
at the same level of PEEP, provided that all direct current gain of the model was located within the static nonlinearity.
Under these conditions, the linear resistance was inversely dependent on the PEEP, whereas little PEEP or amplitude dependence
of the linear compartment elastance was observed. Thus, of the two block-structured models tested, the Hammerstein model accounted
better for the large amplitude nonlinear mechanical behavior. However, neither model could account for the dependence of the
linear block resistance on PEEP. |
| |
Keywords: | Wiener model Hammerstein model System identification Pressure-volume curve |
本文献已被 SpringerLink 等数据库收录! |
|