Smart Wearable Continuous Cuffless Blood Pressure, Cardiac and Respiratory Monitoring System Agnostic to Skin Tone - Project Summary/Abstract Deaths by cardiovascular diseases (CVDs) have increased in the US for over a decade. Elevated blood pressure (BP), i.e., hypertension, is a major driver of critical CVD events. Hypertension is the leading cause of noncommunicable disease deaths, affecting over 1 billion people worldwide, 121 million adults in the USA, with only about 1 in 4 (24%) of US adults having the problem under control. Additionally, hospitalized patients can suffer from dangerously low BP. All currently FDA- approved BP monitors use conventional oscillometric (i.e., cuff-based) methods, which can only provide episodic measurements. Further, cuffs are inconvenient and cause discomfort, which limits their widespread use. Though several commercial cBP wearable devices are in development, they routinely rely on photoplethysmography (PPG), an optical approach which is subject to skin tone, high BMI, and motion artifacts. Thus, there is a large gap between the capabilities of these devices and the accuracy/sensitivity requirements for clinical applications This R01 application seeks to develop an innovative, patent-pending, wearable device that integrates multiple non-invasive physiological sensors on the wrist, includes array-processing deep-learning algorithms to provide health-grade continuous cuffless BP (cBP) monitoring across a diverse population of skin tones and body mass index (BMI) levels, and provides simultaneous estimation of heart rate (HR) and respiratory rate (RR) at no additional cost or burden to the patient. To achieve our research goals, the following three aims will be pursued in parallel: (1) Develop a wearable device that integrates a unique quad design for strain gauge sensors, dual temperature sensors, altimetry and inertial measurements, and wireless communication via Bluetooth. (2) Develop deep learning DL algorithms to estimate cBP, HR, and RR using a hybrid approach that combines (i) a physics-based (mechanistic) model of the interaction between pressure sensors (e.g., strain gauges) and the tissue, and (ii) DL models that learn model parameters from raw sensor signals, compensate for hydrostatic effects and provide semi- automated calibration to account for differences in BMI. (3) Validate the system on a patent- pending phantom system and three cohorts: (i) normal subjects using a unique leg-press protocol to reliably and safely induce BP changes, (ii) patients in an outpatient setting, and (iii) patients in hospital telemetry units. This project brings together a multidisciplinary team of bioengineers, computer scientists, and clinical researchers with decades of experience and a track record of collaboration for more than a decade.
