An evaluation of the clinical potential of a comprehensive model of human respiration in artificially ventilated patients.

Abstract

1. We have investigated the feasibility of accurately simulating the respiratory function of artificially ventilated patients, using a computer model of the respiratory system. Twelve patients artificially ventilated after uncomplicated cardiac bypass surgery was studied. 2. The basic information required to simulate individual ventilated subjects was measured or derived. A program was written to enable key model parameters to be adjusted automatically to match model predictions to these clinical measurements. On completion of this matching procedure all the variables computed by the model were compared with patient values (measured or derived) and their accuracy was assessed. 3. The matching algorithm successfully optimized parameters of the model representing metabolic activity, tissue respiratory quotient, venous admixture, physiological dead space and total body bicarbonate to match measured values for oxygen consumption, carbon dioxide production, Pa, O2, PaCO2, and arterial HCO3- respectively. Other variables compared arise from the solution of equations within the model and correlation between model and patient values is generally good (r greater than 0.9). However, values of Pv-, O2 correlate less well (r = 0.85). Factors affecting the accuracy of patient simulation are discussed and some deficiencies analysed. 4. The creation of an accurate, steady-state representation of a patient by the model opens up the possibility of using it interactively as an aid to clinical management. Some possible future developments of the technique are discussed.