On the assessment of cardiomechanical function via wearable sensing systems: harnessing population-level patterns and dynamics for robust physiological monitoring

Abstract

The objective of this research is to provide a mathematical and conceptual foundation for the processing and analysis of cardiomechanical signals. We begin by exploring a potential clinical application of this technology, using a multi-modal wearable system to accurately track the progression toward hypovolemic shock in an animal model of hemorrhage. In this manner, we demonstrate the potential for cardiomechanical sensing to enable data-driven triage and management of trauma injury. Capturing these signals from wearable systems, however, is a difficult task, creating a barrier to widespread application. To enable more robust analysis of these signals, we begin by presenting a unified method of determining signal quality and localizing the position of the cardiomechanical sensors on the chest wall by analyzing population-level patterns in signal morphology. Next, we develop and explore the idea that observed cardiomechanical signals – while noisy and complex in the time domain – derive from a simple low-dimensional dynamic process. By understanding and modeling these dynamics, we may perform more robust extraction of physiological data from these signals, as well as enabling higher-level tasks such as algorithmic compensation for sensor misplacement.