965: Computational Model of Gas Exchange for Closed-Loop Control of Inspired Oxygen Concentration

Abstract

INTRODUCTION: Autonomous physiologic closed-loop control (PCLC) of the fraction of inspired oxygen (FiO2) during mechanical ventilation has important implications for the management of critically ill patients, especially in austere environments that lack an intensivist-lead care team. In this study, we present a computational model of gas exchange for the testing and validation of PCLC algorithms. METHODS: The model incorporates gas exchange between the alveoli and pulmonary circulation, pulmonary shunt and dead space, and metabolic gas exchange throughout the systemic circulation. Alveolar ventilation was calculated according to the product of the respiratory rate and the tidal volume - dead space difference. Alveolar O2 and CO2 tensions were numerically solved to ensure conservation of mass between CO2 delivery to the lung and its elimination via alveolar ventilation. Mixed arterial oxyhemoglobin saturation (SaO2) was computed as a function of body temperature, O2 and CO2 tensions, and pH. Peripheral O2 saturation (SpO2) was time delayed from SaO2 by a set lung-to-finger delay. For metabolic exchange in the systemic circulation, the mixed arterial blood was compartmentalized for the continuous exchange of gas content with a metabolic source. Blood contents in the systemic circulation was filtered according to a moving average technique to simulate uniform mixing throughout the intravascular volume. Each time-step of the simulation was equivalent to the duration of one breath. RESULTS: We simulated and recorded the performance of a proportional-derivative (PD) FiO2 PCLC algorithm, with the gas exchange model being incorporated into the feedback loop. The model generated SaO2 according to the FiO2 adjusted by the algorithm, as well as the previous oxygen tension of mixed arterial blood. Results from our simulations demonstrated system stability of the PCLC algorithm in the face of rapid, severe physiologic changes that ordinarily would result in severe hypoxemia. Despite these changes, the PCLC algorithm prevented the SpO2 of the model from desaturating below 88%. CONCLUSIONS: Our simulations demonstrate the feasibility of our gas exchange model to establish an appropriate level of safety required for the implementation FiO2 PCLC systems in clinical trials.