Computational Methods to Improve Clinical Decision Science Related to Pulmonary Arterial Hypertension

Pulmonary arterial hypertension (PAH) remains a deadly and rare disorder of the pulmonary vasculature. A combination of vasoconstriction and cellular proliferation of the pulmonary arterial lumen results in increased mean pulmonary arterial pressure, straining the right heart and eventually causing heart failure. Despite the development of a wide range of pharmaceutical treatments for PAH, median survival of this condition remains a paltry seven years. Treatment guidance for PAH depends significantly on a clinician’s ability to assess their patient’s risk of mortality, but all risk assessment methods remain limited in their accuracy and usability.

This dissertation examines the ways in which risk assessment can be used to improve clinical decision science related to pulmonary arterial hypertension, then explores new modeling methodologies to improve upon current risk assessment standards. Chapter 1 discusses the background of challenges related to clinical decision science, specific to pulmonary arterial hypertension. Chapter 2 examines how risk assessment tools can be used to improve the efficiency of clinical trials for pharmaceutical treatments of PAH. Chapter 3 explores how improved risk stratification can reveal differences in treatment response between low and high-risk patients. Chapter 4 examines how novel machine learning methodologies can be employed to improve PAH risk assessment tools. Finally, Chapter 5 studies the potential use of a novel physiological model of right ventricular energetics as a means of improving clinical understanding of right heart failure.