Early Detection of Heart Valve Disease Employing Multiclass Classifier

Cardiac disorder can prove to be fatal for a person's life. Therefore, these disorders must be detected precisely in the preliminary stages. By the use of cardiac auscultation examination, one can examine the heart sounds. Cardiovascular auscultation is the most widely used technique to listen and analyze the cardiac sound in the form of phonocardiogram using an electronic stethoscope. Useful information can be derived from the PCG signal to derive the accurate functioning and status of the heart. Based on information derived, the heart sound signal can be classified into multiple categories. This method proposes an automatic, real-time and modified classification over previous methods to detect cardiac disorder by PCG heart sound signal and was tested over a database containing 5 categories of heart sound signal (PCG signals) which contains signals of one normal and 4 are abnormal categories. The method achieved an accuracy of 97.50 % during the classification process. Features are extracted from the phonocardiogram signal and then those signals are processed using machine learning classification techniques. The experimental observations suggest that the proposed model is efficient for classification of the multi-class heart sounds.