Non-Negative Matrix Factorization for Diabetes II Metabolic Profiling Analysis

¹H nuclear magnetic resonance (NMR) spectra of biological fluids contain rich biochemical information about the metabolic status of an organism. However, spectra interpretation is still an open problem. In this paper, a novel approach, non- negative matrix factorization (NMF), was introduced to reveal the spectral difference between healthy and diabetes II human plasma. The biomarker candidates which respond for discriminating individuals of diabetes II from individuals of healthy were identified by the loading plot and the score plot of NMF. Detail comparisons were also made between NMF and another conventional method principal component analysis (PCA), and the results have shown that NMF would be feasible as a effective and accurate feature extraction tool for characterizing the metabolic status of people as belonging to the healthy group or to the diabetes II group, and provide more information for disease detection as well as the potential therapeutics.