Development of Lightweight Ensemble Network for Detecting Heart Disease Using ECG Signals
- Author(s)
- 신시호
- Issued Date
- 2022
- Abstract
- 심장병은 증상이 나타나면 신속히 치료해야 한다. 심장 질병을 검출하는 머신 러닝 기법은 데스크탑 컴퓨터가 필요하며 이것은 주기적으로 건강을 확인해야 하는 환자에게 치명적인 단점이 될 수 있다. 본 연구에서는 모바일 환경에서 쉽고 빠르게 작동할 수 있는 부정맥 진단을 위한 Federated learning 및 MobileNetV2-BiLSTM 기반 앙상블 알고리즘을 제안한다. 짧은 시간에 측정한 심전도(ECG) 신호는 Matching pursuit 알고리즘을 이용하여 데이터를 증강하며 이를 이용하여 향상된 부정맥 분류 정확도를 달성하였다. 부정맥 데이터는 MobileNetV2와 BiLSTM을 결합한 앙상블 분류기를 통해 분류되었다. 이 알고리즘을 사용하여 데이터를 분류함으로써 91.7%의 정확도를 달성했다. 알고리즘의 성능은 Confusion matrix와 ROC curve를 사용하여 평가하였다. 민감도, 특이도, 정밀도, F1 score는 각각 0.92, 0.91, 0.92, 0.92였다. 제안된 알고리즘은 오랜 시간동안 심전도 신호를 측정하지 않기 때문에 바쁜 현대인들의 건강관리를 용이하게 한다. 또한, Federated learning 방식을 도입하여 데이터 하이재킹 문제에 대한 해결 가능성을 확인하였으며, 이 알고리즘은 모바일 건강관리, 객체 감지, 텍스트 인식 및 인증에 적용될 수 있다.|Heart disease should be treated quickly when symptoms appear. Machine-learning methods for detecting heart disease require desktop computers, an obstacle that can have fatal consequences for patients who must check their health periodically. In this work, we propose Federated learning and MobileNetV2-BiLSTM-based ensemble algorithms for arrhythmia diagnosis that can operate easily and quickly in a mobile environment. The electrocardiogram (ECG) signal measured over a short period of time was augmented using the matching pursuit algorithm to achieve a high accuracy. The arrhythmia data were classified through an ensemble classifier combining MobileNetV2 and BiLSTM. By classifying the data using this algorithm, an accuracy of 91.7% was achieved. The performance of the algorithm was evaluated using a confusion matrix and a receiver operating characteristic curve. The sensitivity, specificity, precision, and F1 score were 0.92, 0.91, 0.92, and 0.92, respectively. Because the proposed algorithm does not require long-term ECG signal measurement, it facilitates health management for busy people. Moreover, parameters are exchanged when learning data, enhancing the security of the system. In addition, owing to the lightweight deep-learning model, the proposed algorithm can be applied to mobile healthcare, object detection, text recognition, and authentication.
- Alternative Title
- 심전도 기반 심장 질병 검출을 위한 앙상블 네트워크 개발
- Alternative Author(s)
- Siho Shin
- Affiliation
- 조선대학교 일반대학원
- Department
- 일반대학원 IT융합학과
- Advisor
- 김윤태
- Awarded Date
- 2022-08
- Table Of Contents
- Table of Contents i
List of Figures iv
List of Tables ix
Acronyms x
Abstract(Korean) xi
I. Introduction 1
1.1. Research background 1
1.2. Key technology 3
1.3. Previous study 6
II. MobileNetV2-BiLSTM algorithms 8
2.1. Preprocessing 8
2.2. Data augmentation 10
2.3. Wavelet Transform 12
2.4. Proposed ECG Signal Classification Method 14
2.5. Performance Evaluation 19
III. Federated learning algorithms 20
3.1. Federated learning based Arrhythmia detection 20
3.1.1 MobileNet 20
3.1.2 Artificial Neural Network V1 21
3.1.1 Artificial Neural Network V2 22
3.2. parameter aggregation 23
IV. Results and Discussion 24
4.1. Results and discussion 24
V. Conclusion 36
References 37
List of Publications 42
Abstract(English) 48
- Degree
- Doctor
- Publisher
- 조선대학교 대학원
- Citation
- 신시호. (2022). Development of Lightweight Ensemble Network for Detecting Heart Disease Using ECG Signals.
- Type
- Dissertation
- URI
- https://oak.chosun.ac.kr/handle/2020.oak/17407
http://chosun.dcollection.net/common/orgView/200000631063
-
Appears in Collections:
- General Graduate School > 4. Theses(Ph.D)
- Authorize & License
-
- AuthorizeOpen
- Embargo2022-08-26
- Files in This Item:
-
Items in Repository are protected by copyright, with all rights reserved, unless otherwise indicated.