Wearable Sensing of Cardio-Pulmonary Function: Non-invasive Sensor Design and Statistical approaches to Signal Compression and Analysis

Wearable Sensing of Cardio-Pulmonary Function: Non-invasive Sensor Design and Statistical approaches to Signal Compression and Analysis

Bibliographic Details
Main Author: Gao, Ju
Language:English
Published: The Ohio State University / OhioLINK 2018
Subjects:
Electrical Engineering
Online Access:http://rave.ohiolink.edu/etdc/view?acc_num=osu1534463839758362
id ndltd-OhioLink-oai-etd.ohiolink.edu-osu1534463839758362
record_format oai_dc
collection NDLTD
language English
sources NDLTD
topic Electrical Engineering
spellingShingle Electrical Engineering
Gao, Ju
Wearable Sensing of Cardio-Pulmonary Function: Non-invasive Sensor Design and Statistical approaches to Signal Compression and Analysis
author Gao, Ju
author_facet Gao, Ju
author_sort Gao, Ju
title Wearable Sensing of Cardio-Pulmonary Function: Non-invasive Sensor Design and Statistical approaches to Signal Compression and Analysis
title_short Wearable Sensing of Cardio-Pulmonary Function: Non-invasive Sensor Design and Statistical approaches to Signal Compression and Analysis
title_full Wearable Sensing of Cardio-Pulmonary Function: Non-invasive Sensor Design and Statistical approaches to Signal Compression and Analysis
title_fullStr Wearable Sensing of Cardio-Pulmonary Function: Non-invasive Sensor Design and Statistical approaches to Signal Compression and Analysis
title_full_unstemmed Wearable Sensing of Cardio-Pulmonary Function: Non-invasive Sensor Design and Statistical approaches to Signal Compression and Analysis
title_sort wearable sensing of cardio-pulmonary function: non-invasive sensor design and statistical approaches to signal compression and analysis
publisher The Ohio State University / OhioLINK
publishDate 2018
url http://rave.ohiolink.edu/etdc/view?acc_num=osu1534463839758362
work_keys_str_mv AT gaoju wearablesensingofcardiopulmonaryfunctionnoninvasivesensordesignandstatisticalapproachestosignalcompressionandanalysis
_version_ 1719454527666520064
spelling ndltd-OhioLink-oai-etd.ohiolink.edu-osu15344638397583622021-08-03T07:08:18Z Wearable Sensing of Cardio-Pulmonary Function: Non-invasive Sensor Design and Statistical approaches to Signal Compression and Analysis Gao, Ju Electrical Engineering Chronic diseases have become the leading cause of mortality in recent years. Compared with acute diseases, chronic diseases pose significant challenges to public health due to many reasons. First, chronic diseases are usually developed slowly over a long period of time as a result of unhealthy living habits and their diagnosis is delayed until they cause serious symptoms. Second, it is often difficult to return chronic disease patients to a healthy state, unlike the case of acute diseases which can be treated over a relatively short time period with proper care. Third, the cost for caring chronic disease patients constitute a major potion of the healthcare spending. Among chronic conditions, cardiovascular diseases remain to be the number one cause of mortality worldwide. A heart attack can often be lethal if the patient is not rescued immediately. Moreover, it is usually difficult to predict when the person will fall into a critical state leading to an imminent cardiac event. Consequently, diagnosing and treating cardiovascular diseases have become a challenging task for health systems across the world. One promising approach to prevent development of cardiovascular diseases is to use bio-markers as early indicators of disease progression, based on cardiac measures obtained on a daily basis using mobile wearable devices. However, unlike the measures collected in the clinic, field measurements is full of motion artifacts and corrupted with noise. This thesis focuses on the problem of mobile cardiac sensing with application to continuous assessment of cardiovascular state. The challenges associated with mobile cardiac sensing include developing unobtrusive sensor modalities, extracting robust measures from noisy data, and improving sensor battery life. In this thesis, we address these challenges through a combination of sensor and algorithm development. Specifically, in the first part of the thesis we review the design of a contactless ultra-wideband (UWB) sensing platform called EasySense for monitoring heart and lung motion. This work aims at providing a novel cardiac sensing modality that directly measures physical motions of heart and lung. We explore both time and frequency domain methods to recover heart rate and respiratory effort from the UWB radar backscatter measurements. Moreover, by further exploiting spatial diversity of multiple-input and multiple-output (MIMO) antenna arrays, we map EasySense measurements to two-dimensional images of reflector locations corresponding to moving tissue boundaries. Measured data experiments show that the contactless EasySense sensor measurements can be used to compute heart rate indices matching the performance of traditional ECG measures. In the second part of the thesis, we consider the problem of minimizing transmission bandwidth of wireless cardiac monitors to enable low power operation. Low power mobile health sensors combine a power efficient front end for signal conditioning and digitization with a wireless transceiver to communicate the measurements to a mobile terminal for analysis and storage. We propose a random projection based data compression and feature detection algorithm suitable for transmission of heart beat signal over low rate wireless links, while maintaining heart rate variability (HRV) information. We apply compressive sensing theory to construct block diagonal structured compression matrices. We provide both experimental results and theoretical guarantees for high temporal resolution accurate heart beat detections.Finally, mobile cardiac sensors have to operate in frequent episodes of low signal-to-noise ratio (SNR) conditions due to motion artifacts. To incorporate data from mobile sensors into clinic research robust beat detection algorithms and confidence measures associated with the estimated cardiac indices are required. Thus, in the final part of the thesis, we propose a Bayesian inference framework for analyzing cardiac sensor measurements to compute cardiac indices of interest, while simultaneously proving confidence intervals. Experimental results with benchmark datasets show that the proposed Bayesian inference framework outperforms state-of-the-art algorithms in continuous heart rate estimation. 2018 English text The Ohio State University / OhioLINK http://rave.ohiolink.edu/etdc/view?acc_num=osu1534463839758362 http://rave.ohiolink.edu/etdc/view?acc_num=osu1534463839758362 unrestricted This thesis or dissertation is protected by copyright: all rights reserved. It may not be copied or redistributed beyond the terms of applicable copyright laws.

Similar Items