The importance of handling multivariate attributes in the identification of heart valve diseases using heart signals

Automated detection of heart valve disease through heart sound has a great requirement due to its inexpensive and non-invasive availability. Extensive research has been conducted recently on applying different classification and features selection techniques. Heart sound data sets represent a real life data that contains continuous attributes and a large number of features that could be hardly classified by most of classification techniques. Data mining techniques including the feature evaluation and classification techniques that ignore the important characteristics that may exist in the heart sound data set may not be applicable on this case. In this context, the present paper initially surveys the research that has been conducted concerning the exploitation of heart sound signals for automated detection of heart conditions. Then, A comparative study is applied to determine the most effective data mining techniques that are capable for the detection of heart valve disease with a high accuracy. The results shows that the techniques that are capable of the handling the multivariate data sets that has continuous nature show the highest classification accuracy.