The impact of methodological choices when developing predictive models using urinary metabolite data

• DOI: 10.1002/sim.9431
• Publication Date: 2022-05-14T07:19:02Z
• Authors: Nikolas Krstic; Kevin Multani; David S. Wishart; Tom Blydt‐Hansen; Gabriela V. Cohen...
• Keywords: Database normalization; Normalization

Abstract The continuous evolution of metabolomics over the past two decades has stimulated search for metabolic biomarkers many diseases. Metabolomic data measured from urinary samples can provide rich information biological events triggered by organ rejection in pediatric kidney transplant recipients. With additional validation, markers be used to build clinically useful diagnostic tools. However, there are methodological steps ranging processing modeling that influence performance resulting metabolomic classifiers. In this study we focus on comparison various classification methods handle complex structure data, including regularized classifiers, partial least squares discriminant analysis, and nonlinear models. We also examine effectiveness a physiological normalization technique widely clinical biochemical literature but not extensively analyzed compared urine studies. While main objective work is interrogate recipients improve diagnosis T cell‐mediated (TCMR), analyze three independent datasets other disease conditions investigate generalizability our findings.