Design and evaluation of linear prediction models for lipidic families based on 1H-NMR LED spectra

Identifier:  TFG:7016

Handle: https://hdl.handle.net/20.500.11797/TFG7016

Authors:  Torné Charlez, Pol

Abstract:
This study focuses on the optimization of the NMR lipid profiling process by supressing the hitherto indispensable step of serum lipid extraction by designing predictive models that can quantify lipid families directly from native serum’s 1H-NMR LED spectrum. For a set of twelve lipidic families an exhaustive process for the development of a linear regression prediction model based on 1H-NMR LED spectra has been successfully conducted. Furthermore, an automatization of the entire linear regression predictive modelling process through the software MATLAB (MathWorks Inc.) and the Partial Least Squares (PLS) Toolbox has been performed and projected into a novel interactive software.