Article Preview
Top1. Introduction
The basic defects of the heart are often gathered via sounds produced by heart. Heart auscultation (the observing of sounds created by the heart and blood flow) is a fundamental tool in the diagnosis of heart disease. Doctors use the stethoscope as a device to monitoring patient’s heart and make a diagnosis accordingly. However, there are several limitations on the method of listening via stethoscope. Listening via stethoscope is dependent on the physician’s ability to understand different heart sounds, on his/her experience, hearing capability, and skill (Inbarani et al., 2013; White et al., 1996). Experience and skill are acquired through investigations, being lead over long years. There are experience and skill problems being come upon particularly with respect to intern physicians and freshly graduated, unsuitable environmental conditions, and unconventionality of patient as well may lead to lacking diagnoses (Azar, 2014; Nakamitsu et al., 1996). Doctors especially concerned in abnormal sounds, it may propose the presence of a cardiac pathology and also offer diagnostic information (Azar et al., 2013). A very significant type of abnormal sound is the murmur sound produced by the turbulent flow of blood in the cardiovascular organism. The timing and pitch of a murmur are of significant importance in the diagnosis of a heart condition, for example, murmurs through diastole are signs of malfunctioning of heart valves but murmurs through systole may relate to any a pathological or healthy heart, depending on the acoustic features of the murmurs. Development of decision support systems will be useful for the doctors in identifying the heart sounds against the possibility of meeting such insufficiencies. Decision support systems, being developed via pattern recognition, classification methods and signal processing, will be helpful for physicians in understanding the heart sounds, and in diagnosing heart diseases consequently. Besides, those developed systems are easy-to-use, cost-efficient, and accurate. In case that heart sounds may be identified, or diagnosed via computer software’s, the abovementioned difficulties are significantly solved.
Heart valves disease data set has recorded only continuous data. The continuous data may reduce classification accuracy; therefore, it is essential to apply discretization as a pre-processing step to convert numerical data into nominal data for such models. Discretization is a technique to partition continuous attributes into a finite set of adjacent intervals in order to produce attributes with a small number of distinct values. Feature selection method should be applied after the discretization. The most significant preprocessing step is the feature selection of the input data set. The data set hold features that are considered as noisy or irrelevant features, these features could have an undesirable impact on the classification accuracy of the data (Azar & Hassanien 2014). Therefore, a good feature selection approach provides better predictive accuracy, processing rate, and avoids incomprehensibility.
The rest of the paper is structured as Sections 2–7. Section 2 describes the related work of feature selection and classification of heart valves diseases diagnosis system using numerous methods. Section 3 describes the research motivation behind this work. Section 4, introduces some preliminaries in rough sets and Bijective soft set. Section 5 explains various steps of the proposed methodology such as discretization, feature selection and Bijective soft set based classification. The experimental evaluation is addressed in section 6. Finally, Section 7 addresses the conclusions and points out instructions for further research.