Mechanical ventilation (MV) therapy has been utilised in the intensive care unit (ICU) for 50 years to treat patients with respiratory illness by supporting the work of breathing, providing oxygen and removing carbon dioxide. MV therapy is utilised by 30-50% of ICU patients, and is a major driver of increased length of stay, increased cost and increased mortality. For patients suffering from acute respiratory distress syndrome (ARDS), the optimal MV settings are highly debated. ARDS patients suffer from a lack of recruited alveoli, and the application of positive end expiratory pressure (PEEP) is often used to maintain recruitment to maximise gas exchange and minimise lung damage. However, determining what level of PEEP is best for the patient is difficult. In particular, it involves a complex trade off between patient safety and ventilation efficacy. Currently, no clinical protocols exist to determine a patient-specific “best” PEEP. Model-based approaches provide an alternative patient-specific method to help clinical diagnosis and therapy selection. In particular, model-based methods can utilise a mix of both engineering and medical principles to create patient-specific models. The models are used for optimising ventilation settings and providing greater physiological insight into lung status than is currently available. Two model-based approaches are presented here. First, a quasi-static, minimal model of lung mechanics is presented based solely on fundamental lung physiology and mechanics. Secondly, a model of dynamic functional residual capacity (dFRC) is developed and presented based on model-based status of lung stress and strain. These models are validated with retrospective clinical data to evaluate the potential of such model-based approaches. Finally, the models are further validated with real time clinical data over a broader spectrum of pressure-volume ranges than prior studies to evaluate the clinical viability of model-based approaches to optimise MV therapy. When validated with real-time clinical trials data, the outputs of the recruitment model provide a range of optimal patient-specific values of PEEP based on different clinically and physiologically derived criteria. The recruitment model is also shown to have the ability to track the disease state of ARDS over time. The dFRC model introduces the PEEP stress parameter, β, which represents a unique population constant. The dFRC model suggests that clinically reasonable estimates of dFRC can be achieved by using this novel value of β, rather than the current, potentially hazardous, methods of deflating the lung to atmospheric pressure. Finally, a third model, combining the principles of recruitment and gas exchange is introduced. The combined model has the ability to estimate cardiac output (CO) changes with respect to PEEP changes during MV therapy. In addition, the model relates the coupled areas of circulation and pulmonary management, as well as linking these MV decision support models to oxygenation based clinical endpoints. A proof of concept is shown for this model by combining two different retrospective datasets and highlighting its ability to capture clinically expected drops in CO as PEEP increases. The model allows valuable cardiovascular circulation data to be predicted and also provides an alternative method and clinical end point by which PEEP could be optimised. The model requires further clinical validation before clinical use, but shows significant promise. The models developed and tested in this research enable rapid parameter identification from minimal, readily available clinical data, and thus provide a novel way of guiding therapy. The models can potentially provide clinicians with information to select an optimal patient-specific level of PEEP using only standard ventilation data, such as pressure-volume curves. In addition, the development of a dFRC stress model provides a unique population constant, β. Overall, the modelling approaches developed and validated in this research provide several novel methods of guiding therapy setting mechanical ventilation parameters and tracking and assess a patient’s lung condition. This research thus creates and provides novel validated methods for improving MV therapy with minimal cost or added invasiveness.