Diagnosis of liver disease using correlation distance metric based K-nearest neighbor approach

Abstract

Mining meaningful information from huge medical datasets is a key aspect of automated disease diagnosis. In recent years, liver disease has emerged as one of the commonly occurring disease worldwide. In this study, a correlation distance metric and nearest rule based k-nearest neighbor approach is presented as an effective prediction model for liver disease. Intelligent classification algorithms employed on liver patient dataset are linear discriminant analysis (LDA), diagonal linear discriminant analysis (DLDA), quadratic discriminant analysis (QDA), diagonal quadratic discriminant analysis (DQDA), least squares support vector machine (LSSVM) and k-nearest neighbor (KNN) based approaches. K-fold cross validation method is used to validate the performance of mentioned classifiers. It is observed that KNN based approaches are superior to all classifiers in terms of attained accuracy, sensitivity, specificity, positive predictive value (PPV) and negative predictive (NPV) value rates. Furthermore, KNN with correlation distance metric and nearest rule based machine learning approach emerged as the best predictive model with highest diagnostic accuracy. Especially, the proposed model attained remarkable sensitivity by reducing the false negative rates.