Estimation of respiratory gas exchange: a comparative study of linear and nonlinear model-based estimation techniques

Monitoring of respiratory gas exchange in humans is an important task in order to establish the physical condition of the patient and to control important physiological indices. In a previous work, we proposed an approximated linear dynamical model of gas exchange within a controlled volume, to be used as a basis for Kalman filtering technique in order to improve the accuracy of the estimates. In this paper, we propose an alternative nonlinear dynamical model of the same phenomenon and suggest the use of a nonlinear estimation technique. A simulation study demonstrates that operative conditions exist where the latter results are more accurate than the estimates based on the linear model. A set of controlled experiments are also designed in order to create situations in which the above difference is significant. In the paper it is evidenced that, in different operative conditions, the analysis both of simulated and experimental data, give systematically the same indications about the choice of the filtering method. The conclusive result of this paper is that a nonlinear model, and the corresponding nonlinear estimation technique, turn out to be convenient when the operative volume and the accuracy of the instrumentation of the experimental set up are both low (operative volumes of about ten liters and flows measurements errors with variances not less than 1 [liter/min]²). It should be also remarked that the proposed model-based estimation techniques, both linear and nonlinear, are highly superior to conventional methods used in medical practice. The present study provides insights and guidelines that can also be useful when dealing with similar gas exchange estimation problems in many other different application areas.