Validation and application of a high-fidelity, computational model of acute respiratory distress syndrome to the examination of the indices of oxygenation at constant lung-state.

Abstract

Calculated venous admixture (Qs/Qt) is considered the best index of oxygenation; surrogates have been developed (Pa(O(2))/Fi(O(2)), respiratory index, and arterioalveolar PO(2) difference), but these vary with Fi(O(2)), falsely indicating a change in lung-state. Using a novel model, we aimed to quantify the behaviour of the indices of oxygenation listed above during physiological and treatment factor variation. The study is the first step in developing an accurate and non-invasive tool to quantify oxygenation defects. We present the static and dynamic validation of a novel computational model of gas exchange in acute respiratory distress syndrome (ARDS) based upon the Nottingham Physiology Simulator. Arterial gas tension predictions were compared with data derived from ARDS patients. The subsequent study examined the indices' susceptibility to variation induced by independent changes in Fi(O(2)) (0.3-1.0), haemoglobin concentration (Hb: 6-14 g dl(-1)), oxygen consumption (VO(2): 250-350 ml min(-1)), and Pa(CO(2)) (4-8 kPa). Static validation produced a mean error of -0.3%, a 10-fold improvement over previous models. Dynamic validation produced a mean prediction error of -0.05 kPa for Pa(O(2)) and 0.09 kPa for Pa(CO(2)). Every parameter, especially Fi(O(2)), induced variation in all indices. The least Fi(O(2))-dependent index was Qs/Qt (variation: 5.1%). In contrast, Pa(O(2))/Fi(O(2)) varied by 77% through the range of Fi(O(2)). We have improved simulation of gas exchange in ARDS by using a sophisticated respiratory model. Using the validated model, we have demonstrated that the current indices of oxygenation vary with alteration in Hb, Pa(CO(2)), and VO(2) in addition to their previously well-documented dependence on Fi(O(2)).