Classification and differential metabolite discovery of liver diseases based on plasma metabolic profiling and support vector machines

Abstract

Discovery of differential metabolites is the focus of metabonomics study. It has very important applications in pathogenesis and disease classification. The aim of this work is to identify differential metabolites for classifying the patients with hepatocellular carcinoma, cirrhosis and hepatitis based on metabolic profiling data analyzed by gas chromatography-time of flight mass spectrometry. A two-stage feature selection algorithm, F-SVM, combining F-score in analysis of variance and support vector machine (SVM), was applied in discovering discriminative metabolites for three different types of liver diseases. The results show that the accuracy rate of the double cross-validation was 73.68±2.98%. 22 important differential metabolites selected by F-SVM were identified and related pathophysiological process of liver diseases was set forth. We conclude that F-SVM is quite feasible to be applied in the selection of biologically relevant features in metabonomics.