A Novel, Low-Cost, Machine Learning based and Wearable Technology for Monitoring Heart Failure Parameters - Heart failure (HF) is a leading cause of mortality, with a low five-year survival rate, often requiring frequent hospitalizations. Remote cardiac monitoring has improved HF management, yet a crucial gap remains in monitoring the heart's pumping function, specifically Left Ventricular Function (LVF) parameters such as Left Ventricular Ejection Fraction (LVEF), Stroke Volume (SV), and Cardiac Output (CO), which are vital for understanding HF's underlying pathophysiology. Our long-term objective is to address this gap by developing a cost-effective, non-invasive, machine learning based and wearable solution for the continuous assessment of LVF parameters. We base our approach on the measurement of 3D precordial vibrations, which correlate with myocardial movements. Our central hypothesis, which was formulated based on previous studies and preliminary data, is that the precordial vibrations associated with cardiac movements contain features closely linked to LV hemodynamic changes. Precise extraction of these features allows us to monitor LVF parameters through a wearable, low-cost approach. This strategy involves placing accelerometers and gyroscopes on the precordial region to achieve non-invasive and compact monitoring. The objective of this project will be accomplished by three specific aims: Aim 1 focuses on further investigating the correlations between precordial vibrations and LVF measures by employing cardiac image processing and computational modeling. The goal is to identify optimal sensor placements and feature extraction methods for precise estimation of LV function parameters. Aim 2 involves evaluating the feasibility of using precordial vibrations to accurately estimate LV function measures in human subjects. To ensure effective feature extraction, we will optimize sensor placements and feature selection and employ cutting-edge artificial intelligence (AI) techniques to accurately estimate LVF parameters. This aim also examines the applicability of our method in patients with varying precordial vibrations. Aim 3 is dedicated to improving our technology to monitor subtle respiratory-related changes in LVF, particularly stroke volume variation (SVV), by capturing respiratory-related feature changes in precordial vibrations, which is crucial for assessing fluid responsiveness in patients. This innovative approach, which combines image processing, patient data-driven computational modeling, and machine learning with wearable sensing for the first time, has the potential to enhance the lives of heart failure patients by reducing healthcare costs and improving their prognosis and quality of life. This contribution will be significant because of its ability to enable continuous monitoring of LV function parameters from a remote setting. Moreover, this proposal aims to substantially broaden student engagement with biomedical research at the Florida Institute of Technology, while enhancing the university's research infrastructure, rendering it an ideal candidate for an R15 AREA grant.
