Browse

Unobtrusive Daily ECG Monitoring Sensors and Middleware for Mobile Healthcare Applications
모바일 헬스케어 어플리케이션을 위한 일상 중 무구속, 무자각적 심전도 모니터링 센서 및 스마트폰 미들웨어

Cited 0 time in Web of Science Cited 0 time in Scopus
Authors
권성준
Advisor
박광석
Major
공과대학 협동과정 바이오엔지니어링전공
Issue Date
2016-08
Publisher
서울대학교 대학원
Keywords
CardioGuardSinabrobra sensorphone-case-type sensorphysiological monitoringelectrocardiographywearable sensoropportunistic sensingunobtrusive sensingsmartphone middleware
Description
학위논문 (박사)-- 서울대학교 대학원 : 바이오엔지니어링전공, 2016. 8. 박광석.
Abstract
Electrocardiography is a non-invasive procedure of recording the electrical impulses of the heart with electrodes placed on the skin. Remarkable advances in sensing and communication technology are expanding healthcare services to our everyday lives. Daily monitoring of electrocardiogram (ECG) will open up new opportunities for developing useful healthcare applications because ECG is a basic indicator to infer the structure of the heart and the function of its electrical conduction system. It also sensitively represents the changes in cardiac activity and the activation in autonomous nervous system (ANS). Despite the high potential of daily ECG monitoring, its obtrusiveness makes it challenging for it to become widely deployed and available. The users of existing ECG monitoring systems need to intentionally touch the sensor continuously on the body for monitoring or to wear an uncomfortable sensing device such as a chest belt, which is not yet widely accepted to most of common users.
This study is motivated to address the obtrusiveness of daily ECG monitoring sensors. The purposes of the study are summarized as the following two aims: 1) enhancing the usability and wearability of a body-attached sensor, 2) reducing a burden of intentional cooperation and carrying a sensor for daily monitoring. First, I propose CardioGuard, a nonintrusive sensor with high usability and wearability for daily ECG monitoring for women, which is designed by leveraging the structure of women's functional underwear. By utilizing the existing functional components of a bra (i.e., bra wires) as sensor electrodes, the bra sensor avoids making users feel uncomfortable by attaching additional sensors. The sensing reliability of CardioGuard was evaluated with 10 participants during 12 representative daily activities. CardioGuard sensor reliably captured about 90% of ECG during those activities. While almost 100% of ECG was reliably captured by CardioGuard during the activities excepting walking, relatively less ECG (72.2%) was reliably captured during the walking and running activities. However, the result excepting three abnormal cases which has unexpectedly very low sensing reliability (21.4% to 27.8%) showed more than 92% of reliable ECG on average and it presents the feasibility of the CardioGuard for reliable sensing during the walking activities. The user study was also conducted to evaluate the usability of the sensor. The results showed that the sensor reliably captures ECG with high usability and wearability during natural daily activities. Most of the participants strongly agreed that the bra sensor achieves high comfort and it is fully satisfied with its own functions as the underwear. The signal quality test and the washing durability test also showed excellent results. The signal quality test showed that the CardioGuard sensor measures high quality of ECG whose power is 12 times stronger than the noise power. The washing durability test showed that the sensing capacity of the bra sensor is expected to be maintained during its typical lifetime.
Second, I propose Sinabro, a smartphone-integrated unobtrusive mobile ECG monitoring sensor which captures a users ECG opportunistically during daily smartphone use. I first investigated a potential opportunity to capture ECG during daily usage of smartphones. Based on this opportunity, I designed a smartphone-case-type ECG sensor that can be integrated with a smartphone for unobtrusive sensing. Multiple metal-based dry electrodes are placed on the outer frame of the case to come into contact with the users two hands and ear during natural smartphone usage such as texting and calling. Sinabro was evaluated in aspect of the sensing reliability with 14 participants during natural smartphone use
i.e., holding, typing, gaming, and calling. The results showed Sinabro opportunistically captured reliable ECG in diverse smartphone use cases. 100% of ECG was reliably captured during holding. During texting in portrait orientation and calling, over 96% of ECG was reliably captured on average. In contrast, there were relatively unreliable sensing in the other three cases, i.e., texting and playing highly- and low-interactive gaming in landscape orientation. The ratios of reliable sensing varied from 81.4% to 90.8%, and their SDs were relatively large (11.8%–22.6%). However, the result excepting four abnormally unreliable ECG data, the ratio was increasing to 96% and it shows the high feasibility to reliable sense during the use cases in landscape orientation.
Finally, I propose a smartphone middleware for supporting ECG-based mobile healthcare applications using CardioGuard and Sinabro. The middleware handles unstable ECG signals and extracts diverse heart-related information, such as heart rate (HR), heart rate variability (HRV). The middleware exposes the extracted health information to mobile healthcare applications via a set of APIs and it enables the applications to be developed and executed without a burden for sensing and extracting high-level health information issues. The middleware includes the algorithms for extracting ECG-derived features
i.e., HR and HRV. The feature extraction algorithms were optimized by empirical analyses for determining the parameter values. The performance of the feature extraction was evaluated and the results showed the middleware extracted highly reliable ECG-derived features opportunistically in serious and frequent noisy data. Overall, the average error of the extracted parameters was only 5.6%. The error in the low-interaction gaming was slightly high, 8.3%. In the average errors of different HRV parameters, most of the parameters showed less than an average of 10% error, except LF/HF. The average error of SDNN was only 1.9%. The average error of mean HR was especially low, at only 0.1%, which is almost completely accurate heart rate estimation. LF and HF showed relatively high error rates (7.4%, 9.5%, respectively), and they caused the high error of LF/HF. However, TF, normalized LF and normalized HF showed low average error (< 5.3%). The results also showed the high feasibility of the middleware to provide reliable variation trends of HRV parameters. The variation trends of HRV parameters showed significantly high Pearson correlations (0.97) with those from the reference data.
Although recent significant progress in daily ECG monitoring has shown the potential of pervasive healthcare applications, its obtrusiveness has obstructed them to be widely accepted to common users. The proposed unobtrusive daily ECG monitoring sensors will help these applications come to our reality by addressing the obtrusiveness. The proposed software framework will also promote the development of ECG-based smartphone healthcare applications by reducing the burden of the sensing, signal processing, and the feature extraction issues in development procedures.
Language
English
URI
https://hdl.handle.net/10371/119897
Files in This Item:
Appears in Collections:
College of Engineering/Engineering Practice School (공과대학/대학원)Program in Bioengineering (협동과정-바이오엔지니어링전공)Theses (Ph.D. / Sc.D._협동과정-바이오엔지니어링전공)
  • mendeley

Items in S-Space are protected by copyright, with all rights reserved, unless otherwise indicated.

Browse