Related Experiment Videos
Nonlinear identification of the PCO2 control system in man
M Noshiro1, M Furuya, D Linkens
1Division of Electronic Engineering, Tokyo Medical and Dental University, Japan.
Computer Methods and Programs in Biomedicine
|July 1, 1993
Summary
This study compared two methods for identifying the human PCO2 system. A nonlinear NARMAX model best fit the data, outperforming a Belville model, suggesting advanced identification techniques are suitable for respiratory modeling.
Area of Science:
- Physiology
- Biomedical Engineering
- Systems Biology
Background:
- Accurate modeling of the human respiratory system is crucial for understanding gas exchange.
- Previous models often simplify complex physiological dynamics.
- The partial pressure of carbon dioxide (PCO2) system regulation is a key aspect of respiratory control.
Purpose of the Study:
- To compare two distinct methods for identifying the human PCO2 system.
- To evaluate the effectiveness of nonlinear black-box versus structured compartmental models.
- To determine the optimal modeling approach for respiratory subsystem analysis.
Main Methods:
- Utilized a nonlinear Nonlinear AutoRegressive Moving Average with eXogenous inputs (NARMAX) identification package.
- Employed a structured two-compartment Belville model for comparison.
- Collected data from volunteers breathing controlled gas mixtures, measuring respiratory gas flow and PCO2.
Main Results:
- A low-order dynamic model with nonlinear polynomial expansion provided the best fit to the experimental data.
- The Belville model performed best when linearized, due to optimization challenges with its nonlinear form.
- NARMAX identification demonstrated superior performance in modeling the PCO2 system.
Conclusions:
- Modern systemic excitation and identification methods are effective for modeling human respiratory subsystems.
- Nonlinear dynamic models offer a more accurate representation of the PCO2 system compared to traditional compartmental models.
- The findings support the use of advanced computational techniques in respiratory physiology research.