الفهرس 
AbstractOnly 14 pages are availabe for public view
 |  | from | 132 |  |  |
| | | from | 132 | | |
Abstract
Sensor networks constitute a scientific revolution in the field of communication and embedded systems. They continuously record a massive amount of data that contains uncertain data because of low battery power, network transmission problems, and noise. For example, Body sensor networks (BSNs) are used to monitor the electrical activity of the human heart by continuously recording Electrocardiogram (ECG) signals to diagnose early abnormalities. Still, they can be impaired due to the reasons mentioned above in addition to the kinds of activities practiced by the patient during the monitoring time. Sensor networks raise the challenges related to analyzing the streaming data to handle unreliable and incomplete data. Sensor data can be annotated by the semantic technologies to enrich the contextual information. Cardiac arrhythmia is a perturbation in the rhythm of the heart, evident by abnormally fast rates or abnormally slow rates or irregular rates. The need increases to analyze that streaming ECG signals, despite the existence of unreliable and incomplete data, for detecting early abnormal heartbeats and classifying its arrhythmia type. In this thesis, we meet the previous challenge in two directions. The first direction is to represent the knowledge about arrhythmias concepts, properties, and types to describe that highly sensitive medical domain by building our original fuzzy ontology for cardiac arrhythmias. As ordinary ontologies are insufficient to represent uncertainty and vagueness in real-world critical information. Therefore, fuzzy ontologies are the ideal solution to represent this knowledge. The fuzzy logic can capture the uncertainty problem as it replaces the crisp recordings of sensors by fuzzy values. Fuzzy ontology is a suitable formalism for representing uncertainty knowledge for cardiac arrhythmias domain, including their concepts, relations, and properties. Our fuzzy ontology contains diseases, symptoms, diagnosis, and treatments, using the standard Web Ontology Language (OWL2) and FuzzyOWL2 plug-in for protégé. The cardiac arrhythmias diseases hierarchy and terms are determined upon the standard Disease Ontology (DO). The second direction is to propose a framework to analyze the streaming ECG signals from Holter monitors, according to ECG features, and classify the type of cardiac arrhythmias for a heartbeat. Our framework is relying on our constructed OWL2 fuzzy ontology for cardiac arrhythmias and fuzzy ontology-based inference engine. from our experimental results, the accuracy of the proposed framework is 97.33% using the collected dataset from Physio Bank. We compared our experimental results with other suggested frameworks that depend on ordinary ontologies to analyze streaming ECG signals and classify each heartbeat. We have achieved more accurate results using a dataset from Physio Bank.