Design and development of a system for prophylactic coronary artery disease prediction and prevention

Abstract

Recent advancements in the field of computer science and information technology have contributed tremendous enhancements in designing and developing Medical Decision Support Systems (MDSS). MDSS play an important role in diagnosing the disease by interpreting medical information. This system serves the purpose of an electronic doctor to detect abnormalities, disease progress and diagnosis. It makes use of data processing techniques by accepting textual data from medical records and medical images for predicting the prevalence of disease automatically. On positive prognosis of disease, the patient can undertake prophylactic treatment to prevent its severity. The system is also helpful to follow up the current status of the disease occurred in the patient. Coronary Artery Disease (CAD) is one of the major rampant causes of death in humans worldwide. The purpose of this thesis is to design algorithms and develop computer aided diagnosis system for predicting the presence or absence of coronary artery disease. The heart disease severity level is determined and confirmed using cardiac image that is obtained from invasive clinical procedure. We designed MDSS to predict the risk of CAD using Multi Criteria Decision Making (MCDM) methods, computational intelligence techniques and medical image processing. Initially, CAD risk prediction is carried out by using MCDM methods namely Analytic Hierarchy Processing (AHP), Technique for Order Preference by Similarity to Ideal Solution (TOPSIS) and Simple Additive Weighting (SAW) methods. AHP is employed to analyze the hierarchy structure of CAD diagnosis and determine the attribute weights based on the individual contribution of risk attributes for disease diagnosis as prescribed by a medical expert. The extended TOPSIS and SAW methods are designed for predicting CAD risk using attribute weights obtained from AHP. Performance of the proposed system is evaluated using Cleveland heart disease dataset as benchmark which is available online in University of California at Irvine (UCI) machine learning repository. The proposed MCDM methods provided accurate prediction results on the diagnosis of CAD risk similar to the medical experts. newline