Self-coach: an intelligent WBAN system for heart disease prediction using non-dominated sorting genetic algorithm

Emami-Abarghouei, Babak (2016) Self-coach: an intelligent WBAN system for heart disease prediction using non-dominated sorting genetic algorithm. Masters thesis, Memorial University of Newfoundland.

[img] [English] PDF - Accepted Version
Available under License - The author retains copyright ownership and moral rights in this thesis. Neither the thesis nor substantial extracts from it may be printed or otherwise reproduced without the author's permission.

Download (5MB)

Abstract

Wireless Body Area Network (WBAN) is a new technology based on an advanced healthcare system and Wireless Sensor Network (WSN). This domain has been designed for monitoring patients using their physical signals by providing lowcost, wearable, unobtrusive solutions for the continuous monitoring of cardiovascular health and physical activity status. Recent studies have addressed the use of WBAN, by means of real-time comprehensive health monitoring systems such as the Personal Health Monitoring system (PHM). The aim of utilizing these systems is to provide a fast and early diagnosis; however, WBAN has not fulfilled its potential while using the existing methods of machine learning and data mining techniques. The proposed method is a new framework for early heart failure prediction systems based on an intelligent WBAN named Self-Coach. Self-Coach is an intelligent monitoring system due to the detection/prediction method it uses, which provides real-time health status monitoring using the Non-Dominated Sorting Genetic Algorithm. In this study, Self-Coach is proposed as a new medical diagnosis method based on the hybrid approach. This approach applies the Support Vector Machine (SVM) to classification, where its parameter values are optimized and visualized by the Non- Dominated Sorting Genetic Algorithm-II (NSGA-II). To predict a potential health tolerance threshold for a particular patient, the optimal boundary curve between both the healthy and unhealthy class and the patient's possible health positions based on new conditions (blood pressure and heart rate) have been explored with NSGA-II. The optimal boundary is a set of non-dominated offspring from the unhealthy class that has been generated with NSGA-II and verified with the SVM. These members are plotted as a Pareto Front curve known as the optimal boundary curve. Based on the SVM classification results, this is the most fitted curve which can separate the healthy class from the unhealthy. To explore the potential health position of a particular patient, new possible positions will be generated with NSGA-II. To generate these points, the patient's dynamic data (blood pressure and heart rate) will be increased and simulated utilizing NSGAII. The potential tolerance threshold for each patient is the new health position for that particular patient based on the new health conditions which will cross the optimal boundary or be dominated by at least one of these members. To evaluate the Self-Coach's performance, its experimental results (optimal boundary members and explored tolerance thresholds for each individual) have been verified using SVM and compared with the raw dataset classification results. Based on the simulated results, this method can provide 24-hour health monitoring care for elderly people and those who might have coronary heart disease. It is a new technology for this domain. Real-time data analysis is a significant part of Self-Coach, which makes it a good candidate for supporting a broad array of high-impact applications in the domain of the healthcare system, for training, rehabilitation, surgical recovery and as a home-based monitoring healthcare system.

Item Type: Thesis (Masters)
URI: http://research.library.mun.ca/id/eprint/12706
Item ID: 12706
Additional Information: Includes bibliographical references (pages 142-151).
Keywords: Wireless Body Area Network, Non-Dominated Sorting Genetic Algorithm, Optimisation, Heart Attck Prediction, Support Vector Machine, Classification, WBAN
Department(s): Science, Faculty of > Computer Science
Date: September 2016
Date Type: Submission

Actions (login required)

View Item View Item

Downloads

Downloads per month over the past year

View more statistics