The full text of this item is not available at this time because the student has placed this item under an embargo for a period of time. The Libraries are not authorized to provide a copy of this work during the embargo period, even for Texas A&M users with NetID.
Advanced Sensors Coupled with Human Physiology for Cuffless Blood Pressure Monitoring
Abstract
We need advanced artificial intelligence (AI) algorithms in both engineering and medicine. These algorithms will enhance the traditional medical practice and lead towards a new level of personalized, effective, and accessible medical care. The use of these algorithms with unobtrusive advanced wearable sensors will ultimately enable continuous access to complex health parameters, improving the diagnostic and prognostic services, and augmenting our understanding of the risk factors to facilitate any necessary corrective actions for personalized care. This dissertation presents a set of innovative solutions that target two fundamental challenges facing the current state-of-the-art AI-based physiological monitoring approaches: the data-intensive requirements of AI training, and the inconsistent and low-fidelity sensing that conventional wearable sensor technologies provide. Traditionally, AI algorithms require copious amounts of personalized ground truth data during training to accurately model input-output relationships of a biological system that exhibits complex nature and significant heterogeneity across individuals. However, for many biomedical applications, collecting such large amounts of ground truth data, particularly at the personalized levels, is challenging, burdensome, and in some cases, infeasible. To address this critical shortcoming, we establish physics-informed neural network (PINN) models for physiological time series data that would reduce reliance on ground truth information. We achieve this by building Taylor's approximation for the gradually changing known cardiovascular relationships between input and output of the system and incorporating this approximation into neural network training. Moreover, we propose advanced wearable sensors that utilize novel bioimpedance modality to unobtrusively and seamlessly capture hemodynamic information addressing the challenges with the conventional wearable sensing technologies. We demonstrate the effectiveness of our techniques through a comprehensive case study on continuous cuffless blood pressure (BP) estimation. We use PINNs over the state-of-the-art AI and machine learning models to extract beat-to-beat BP from bioimpedance measurements collected with advanced wearable sensors. We show that PINNs retain a high correlation (systolic: 0.90, diastolic: 0.89) and low error (systolic: 1.3 ± 7.6 mmHg, diastolic: 0.6 ± 6.4 mmHg) in BP estimations, while reducing the amount of ground truth training data on average by a factor of 15.
Subject
precision medicinewearable sensors
blood pressure monitoring
remote health monitoring
physics informed neural networks
Citation
Sel, Kaan (2023). Advanced Sensors Coupled with Human Physiology for Cuffless Blood Pressure Monitoring. Doctoral dissertation, Texas A&M University. Available electronically from https : / /hdl .handle .net /1969 .1 /199019.