Evaluation of a mathematical model to predict intrapulmonary shunt non-invasively

Purpose: We have previously published a mathematical model of oxygen transp ort. Using several physiological assumptions, the model provides a non-inva sive estimate of intrapulmonary shunt. During a larger study of lung injury in a pig model, we had the opportunity to check the validity of our assump tions and the accuracy of the model's predictions. Methods: We used six female pigs, average weight 12.8 kg. Following general anesthesia, tracheostomy and insertion of pulmonary venous and arterial li nes, lung injury was induced by repeated saline lung lavage. Using hemodyna mic measurements made at different levels of inspired oxygen, intrapulmonar y shunt was calculated both by the traditional shunt equation and also by o ur mathematical model based on non-invasive measurements of F1O2 and SaO(2) . Results: There was good agreement between the two methods of shunt calculat ion. Using linear regression the correlation coefficient was 0.95. Bland an d Altman analysis showed a bias of -0.8 and precision of 12%. Conclusion: In a controlled setting, intrapulmonary shunt can he estimated from non-invasive measurements to a reasonable degree of accuracy. However, the calculation requires too many assumptions to be of general clinical va lue. The equations used provide a validated physiological model that acts a s a useful tool for teaching cardiorespiratory physiology.