Model-driven gas exchange monitoring in the critically ill

Understanding pulmonary gas exchange performance is a dynamic process which, depending on clinical context, exhibits different levels of complexity. Global tools such as tension-based indexes yield clinically crucial information under very specific conditions. Yet, accurate mechanistic insight can only originate in model-based tools. One-parameter models such as shunt or dead space are well established in clinical practice whilst two or three-parameter models have just been advanced and their role is yet to be delineated. Although the latter provide superior accuracy, this comes at the cost of increased complexity and possibly the need for invasive data sets. Modelling gas exchange enables a quantitative and physiologically-driven management of patients with lung failure. Assumptions are inherent to each tool and can clinically mislead if not accounted for. Thorough understanding of their subjacent theoretical construct is a prerequisite for their judicious use. This manuscript aims to describe current gas exchange monitoring tools, with special reference to their mathematical framework and constituent pitfalls. A unifying perspective on their clinical role is proposed.