UC Irvine

The primary goal of this dissertation was to develop a computational approach that can be used for better understanding the patient-specific consequences of obstructive sleep apnea (OSA), an increasingly prevalent sleep disorder associated with breathing disturbances and airflow limitation caused by a repeated collapse of the upper airway. To this end, Chapter 1 first addresses limitations in the clinical diagnosis and severity scoring of OSA, while highlighting observed disparities in literature, and reviews possible alterative severity metrics in the context of cardiovascular disease (CVD) risk, with a focus on those related to oxygen availability. The link between CVD and the hallmark intermittent hypoxia (IH) of OSA was also investigated. Having identified the strength of oxygen-based severity metrics and the possible issues associated with measuring oxygen using the standard pulse oximetry, in Chapter 2, the formulation of a mass transfer model to predict oxygen levels in different body compartments was detailed, and clinical application with a direct usage of patient data was demonstrated. Given the importance of output reliability on input accuracy in mathematical modeling, in Chapter 3, the lung volume input to the mass transfer model (Chapter 2) was optimized, and the impact of input error on the model oxygen output was assessed. With a better idea of output reliability, in Chapter 4, the dissolved oxygen output from the mass transfer model (Chapter 2) was employed to inform the experimental conditions of a gas chamber with an in-vitro cell culture set-up to create a patient-specific representation of OSA, as part of a demonstration of reverse translation research. In this chapter, the formulation of a mass transfer model for the experimental set-up was detailed, along with that of a novel optimization algorithm required for gas controller protocol prediction. Lastly, in Chapter 5, the achievements of the dissertation were outlined, and future directions for model-based assessment of OSA were addressed.