RESUMEN
Proton nuclear magnetic resonance (1H NMR) combined with partial least squares (PLS) was developed for the rapid determination of squalene and sterols (brassicasterol, campesterol, stigmasterol and ß-sitosterol) in 119 vegetable oils from 7 different species. The 1H NMR spectra of these oil samples were correlated to the reference value determined by gas chromatography-mass spectrometry (GC-MS) method by PLS regression, using outlier removal, selection of input X-variables and data pretreatments. Auto-scaling (UV) was chosen as the best pre-processing for the PLS models of stigmasterol, ß-sitosterol, and squalene. Pareto variance (Par) was more suitable for the PLS model of brassicasterol. The model for squalene was further improved by a reduced number of variables with variable importance for the projection (VIP) scores technique. The study demonstrated the potential application of NMR coupled with PLS as a rapid and nondestructive technique for the routine analysis of squalene and sterols in vegetable oils.