Diagnosis of breast cancer using HPLC metabonomics fingerprints coupled with computational methods

Abstract

The present study was focused on developing a computational procedure for analysis of the HPLC metabonomics fingerprints of human urine to distinguish between patients with breast cancer from healthy people. The predictive rate of support vector machine (SVM) based diagnosis model is 100% for training set and 93.2% for test set, respectively. Current work might have important reference values to explore the methodology of metabonomics.