The lungs are the primary organ of the respiratory system. Their main function is to provide freshly breathed oxygen (O²) to the blood capillaries, while taking carbon dioxide (CO²) from them and expelling it to the atmosphere. Lung conditions such as Acute Respiratory Distress Syndrome (ARDS), Idiopathic Pulmonary Fibrosis (IPF), Coronavirus Disease (COVID-19), etc., cause impaired gas exchange that is life-threatening. In this dissertation, I developed 1) a physiology-based dynamic pulmonary system to study the lung normo- and patho-physiology, and 2) a model-based constrained optimization algorithm to do parameter estimation in order to non-invasively assess lung health.
The goals of this work are 1) to accomplish a respiratory personalized medicine example for clinical decision support, and 2) to further the understanding of respiratory physiology, via a mechanistic physiology-based model and system identification techniques. The mechanistic model presented in this thesis comprises six subsystems: 1) a lung mechanics module that computes airflow transport from the mouth and nose to the alveoli (gas exchange units), 2) a respiratory muscles and rib cage mechanics module that simulates the effect of the respiratory muscle contraction on the lungs and the rib cage, 3) a microvascular exchange system that describes fluid (water) and mass (albumin and globulin) transport between the pulmonary capillaries and the alveolar space, 4) an alveolar elasticity module that computes alveolar compliance as a function of the pulmonary surfactant concentration and the elastic properties of the lung tissue fiber, 5) a pulmonary blood circulation that describes blood transport from the heart to the pulmonary system, and 6) a gas exchange system that describes O² and CO² transport between blood in the pulmonary capillaries and gas in the alveoli. Each subsystem was developed based on the latest knowledge of lung physiology and was validated using patient data when available or published and validated physiology-based models. To our knowledge, the combined six-module model would be the most rigorous and expansive lung dynamic model in the literature. This dynamic respiratory system can be used to describe human breathing under healthy and diseased conditions. The model can readily be used to test different what-if scenarios to find the optimal therapy for the patients.
Further, I tailor the proposed lung model and adopt system identification techniques for noninvasive assessment of the lung mechanical properties (resistance and compliance) and the patient breathing effort. Pulmonary syndromes or diseases, such as ARDS and COPD (Chronic Obstructive Pulmonary Disease) evoke alterations in lung resistance and compliance. These two parameters reflect, by and large, the state of health and functionality of the respiratory system. Hence tracking these two parameters can lead to better disease diagnosis and easier monitoring of the respiratory disease progression. For spontaneously breathing patients on ventilatory support, the estimation of the lung parameters is challenging due to the added patient’s breathing effort. This dissertation presents a model-based nonlinear constrained optimization algorithm to estimate, breath-by-breath, the lung resistance, the lung compliance, as well as the patient breathing effort due to the respiratory muscle activity, using readily available non-invasive measurements (airway opening pressure and airflow).
Identifer | oai:union.ndltd.org:columbia.edu/oai:academiccommons.columbia.edu:10.7916/d8-t170-t155 |
Date | January 2021 |
Creators | Yuan, Jiayao |
Source Sets | Columbia University |
Language | English |
Detected Language | English |
Type | Theses |
Page generated in 0.2378 seconds