A Non-Invasive Patient-Specific Modeling Approach for Predicting Group II Pulmonary Hypertension as a Clinical Indicator of Diastolic Heart Failure for Patients with Uncertain Clinical Data

Diastolic dysfunction (sometimes referred to as diastolic heart failure or heart failure with preserved ejection fraction, HFpEF) is a common pathology occurring in about one-third of patients affected by heart failure. However, this condition may not be associated with a marked decrease in cardiac output or systemic pressure and therefore is more challenging to diagnose than its systolic counterpart. Compromised relaxation or increased stiffness of the left ventricle leads to elevated pressures in the pulmonary arteries. This condition is classified as Group II pulmonary hypertension or pulmonary hypertension due to left heart disease, one of the five classification sub-groups of hypertension \cite{simonneau2013updated}. In addition, the left ventricle's impaired relaxation or stiffness may increase the right ventricular afterload, which can lead to right ventricular failure. Therefore, elevated pulmonary pressures are an important clinical indicator of diastolic heart failure and significantly correlate with associated mortality. However, accurate measurements of this quantity are typically obtained through invasive cardiac catheterization. Moreover, the measurements are usually only obtained after the onset of symptoms. This thesis uses the hemodynamic consistency of a differential-algebraic circulation model to predict pulmonary pressures in adult patients from other, possibly non-invasive, clinical data. We investigate several aspects of the problem, including the ability of model outputs to represent a sufficiently broad pathologic spectrum, the identifiability of the model's parameters, and the accuracy of the predicted pulmonary pressures. We also find that a classifier using the assimilated model parameters as features is free from problems that arise from missing data and can detect pulmonary hypertension with sufficiently high accuracy. For a cohort of 82 patients suffering from various degrees of heart failure severity, we show that systolic, diastolic, and wedge pulmonary pressures can be estimated on average within 8, 6, and 6 mmHg, respectively. We also show that, in general, increased data availability leads to improved predictions.

An introduction to the cardiovascular system, its components, and the underlying mathematical relationships used to describe blood flow are contained in Chapter 1. A review of the clinical motivation for the aforementioned work is contained in Chapter 2, along with a review of heart failure and hypertension. An overview of the origins and history of modeling of the cardiovascular system, including lumped parameter models, is covered in Chapter 3. In addition, Chapter 3 includes the formulation of the 0D hemodynamic circulation model used throughout the body of this thesis. Chapter 4 is devoted to both the methodologies used for the model and the model's parameters. It introduces the statistical model, tuning of the model, and the use of Bayesian computation methods. This chapter also covers parameter estimation and optimization methods in addition to parameter sensitivity and identifiability analysis techniques. Finally, the chapter ends with an overview of the classification method implemented in this thesis, followed by a brief introduction to the computational framework utilized throughout this thesis. The final chapter, Chapter 5, introduces the two data set and presents the main research questions. The first two research questions presented focus on validating the model formulation by accessing the model's physiological admissibility and the model's ability to represent systolic and diastolic dysfunction mechanisms. The final questions address the sensitivity and identifiability analysis of the model parameters, the prediction of the pulmonary pressures, the identification of the most impactful clinical measurements, and finally, the use of patient-specific trained model parameters to identify pulmonary hypertension in patients of varying data availability. The results for each research question directly follow the questions. Finally, this chapter ends with a conclusion resulting from the implications drawn from each of the results.