Kemeny–Snell Distance in Nuclear Magnetic Resonance Metabolomics

We shall introduce Kemeny–Snell distance (KSD) metric on metabolomics and validate results with partial least squares discriminant analysis (PLS-DA). KSD metric allows principally to identify the most relevant chemical shift ranges directly from spectra without metabolite library-limited signature search and quantitation. The one-dimensional proton nuclear magnetic resonance spectra of the serum of ischemic heart disease (IHD) patients ( n = 19) and controls ( n = 19) showed the statistical significance by the first latent variable (LV1; 35.18%) of PLS-DA. The significance between ischemic heart disease (IHD) patients and controls was tested and confirmed by KSD metric. We used PLS-DA and KSD metric for the interpretation of serum NMR spectra of IHD patients and healthy controls and both methods show a significant deviation from the controls. KSD is a robust to spectral artifacts and potentially useful as a diagnostic tool to assess the likelihood of many pathologies simultaneously.