KINETIC FLUORESCENCE FOUR-WAY DATA ANALYZED BY PARAFAC AND RADIAL BASIS FUNCTIONS/RESIDUAL TRILINEARIZATION: MALONALDEHYDE ANALYSIS IN OILS

GARCÍA-REIRIZ, ALEJANDRO G.; CAÑADA-CAÑADA, FLORENTINA; DAMIANI, PATRICIA C.; MUÑOZ DE LA PEÑA, ARSENIO; OLIVIERI, ALEJANDRO C.

Bologna - Italia

Congreso; XIII International Symposium on Luminescence Spectrometry; 2008

Four-way data were obtained by recording the kinetic evolution of excitation-emission fluorescence matrices for the highly fluorescent product of the Hantzsch reaction between the analyte malonaldehyde and methylamine: 1,4-disubstituted-1,4-dihydropyridine-3,5-di carbaldehyde, which was previously employed for the linear relationship between the kinetic fluorescence data and nominal analyte concentrations has been demonstrated, and therefore the four-way data were processed whit PARAFAC followed by non-linear pseudo-univariate regression and also by a recently introduced neural network methodology, based on radial basis functions (RBF). This algorithm is able to perform a variety of nonlinear multivariate function estimation and nonlinear regression task. The analyte malonaldehyde is a product of stress and bad storage of lipids. It is the best know lipid peroxidation product, and the onethat has been most widely used as marker for various studies associated with this phenomenon. Its analysis in the presence of the fluorescent olive oil background is possible with the presently studied data, because they permit the achievement of the second-order advantage. This was accomplished by resorting to PARAFAC with non-linear pseudo-univariate modeling of the score-concentration relationship, and also bye the useful combination of unfolded principal component analysis and residual trilinearization prior to application of RBF neural networks. This allows for an adequate modeling of the non-linear data and separation of the contribution from the interfering background, leading to the successful determination of malonaldehyde in olive oli samples.