Metabonomic analysis of hepatitis B virus-induced liver failure: identification of potential diagnostic biomarkers by fuzzy support vector machine *

摘要: Hepatitis B virus (HBV)-induced liver failure is an emergent disease leading to high mortality. The severity of may be reflected by the profile some metabolites. This study assessed potential using metabolites as biomarkers for identifying with good discriminative performance its phenotype. serum samples from 24 HBV-induced patients and 23 healthy volunteers were collected analyzed gas chromatography-mass spectrometry (GC-MS) generate metabolite profiles. further grouped into two classes according failure. Twenty-five commensal peaks in all profiles extracted, relative area values these used features each sample. Three algorithms, F-test, k-nearest neighbor (KNN) fuzzy support vector machine (FSVM) combined exhaustive search (ES), employed identify a subset (biomarkers) that best predict Based on achieved experimental dataset, 93.62% predictive accuracy 6 was selected FSVM-ES three key metabolites, glyceric acid, cis-aconitic acid citric are identified diagnostic biomarkers.