SENSORS: Multi-Channel Wearable Biosensors for Continuous Cardiovascular Monitoring

目前，还没有足够的可穿戴式生物传感器系统(WBS)用于循环监测。在数周、数月或数年的观察期内实现对血流动力学的连续逐跳监测，将被证明是研究和管理慢性心血管(CV)疾病(如高血压、心力衰竭和外周血管疾病)的革命性工具。这项任务需要在身体上安装传感器，理想情况下，传感器可以一次佩戴几天，从而产生实时数据。能够进行有用的血流动力学和血管测量的现有传感器模式实际上是不可穿戴的，而现有的可穿戴传感器模式记录的生物信号由于运动伪影和变化的生理条件而在医疗应用中存在问题。为了克服这些问题，我们提出了一种新的基于多传感器融合的系统辨识方法：多通道盲系统辨识技术。MBSI通过利用在多个点同时观测到的传感器输出之间的相关性来估计未知输入和未知动态。在受试者的表面放置多个非侵入性生物传感器可以产生同时的循环信号。这些测量可以用MBSI算法系统地处理，以估计血管通道动力学和中枢血流动力学，如动脉血压(ABP)和心输出量(CO)。虽然在延长的监测期间，对象的生理条件经常变化，但MBSI方法可以实时识别变化，并使用识别的通道动力学将观察到的外围传感器信号正确地转换为仅通过有创导管测量的诸如CO的深部信号。在本项目中，我们将建立基于MBSI的心血管监测的理论基础，开发一个由非侵入性、低功耗、紧凑的生物传感器通过无线自组织网络连接的可穿戴生物传感器网络原型，并开发三个主要临床应用：1)非侵入性传感器的长期心输出量估计，2)可穿戴传感器的连续动脉血压监测，3)外周动脉病理诊断，如局灶性动脉粥样硬化和夹层。该项目的高度跨学科性质和国际合作部分为本科生和研究生提供了独特的教育机会。

Presently, there is no adequate wearable biosensor system (WBS) for circulation monitoring. Enabling the continuous beat-to-beat monitoring of hemodynamics for observation periods of weeks, months, or years, could prove revolutionary as a tool for the study and management of chronic cardiovascular (CV) diseases such as hypertension, heart failure, and peripheral vascular disease. This task requires sensors on the body, which ideally can be worn for days at a time, yielding real time data. Existing sensor modalities that are able to make useful hemodynamic and vascular measurements are not actually wearable, whereas existing wearable sensor modalities record biosignals that are problematic for medical applications due to motion artifact and varying physiological conditions. To overcome these problems, we propose a new system identification approach based on multiple sensor fusion: the multi-channel blind system identification (MBSI) technique. MBSI allows the estimation of both an unknown input and the unknown dynamics by exploiting the correlation among sensor outputs observed simultaneously at multiple points. Placing multiple non-invasive biosensors on the surface of a subject yields simultaneous circulatory signals. These measurements can be systematically processed with MBSI algorithms to estimate the vascular channel dynamics and central hemodynamics, such as arterial blood pressure (ABP) and cardiac output (CO). Although the physiological conditions of the subject frequently change during an extended period of monitoring, the MBSI method can identify the changes in real time, and use the identified channel dynamics for correctly converting the observed peripheral sensor signals to the deep-body signals, such as CO, which has been measured only by invasive catheterization. In this project we will establish the theoretical foundation of the MBSI-based CV monitoring, develop a prototype wearable biosensor network having non-invasive, low-power, compact biosensors connected by a wireless ad hoc network, and develop three major clinical applications; 1) long-term cardiac output estimation with non-invasive sensors, 2) continuous arterial blood pressure monitoring with wearable sensors, 3) diagnosis of peripheral arterial pathology such as focal atherosclerosis and dissection. The highly interdisciplinary nature and the international collaboration component of this project provide unique educational opportunities to undergraduate and graduate students.