Unfolded partial least squares/residual bilinearization combined with the successive projections algorithm for interval selection: enhanced excitation-emission fluorescence data modeling in presence of inner filter effect

ARAUJO GOMES A; SCHENONE A V; GOICOECHEA H C; UGULINO DE ARAUJO M

ANALYTICAL AND BIOANALYTICAL CHEMISTRY

SPRINGER HEIDELBERG

Lugar: HEIDELBERG; Año: 2015 vol. 407 p. 5649 - 5659

1618-2642

The use of interval selection by the successive projections algorithm (SPA) for elimination of uninformative variables and unfolded partial least squares regression (U-PLS) modeling of excitation-emission matrices (EEM), when inner filter effect (IFE) is present, is reported for first time. The post-calibration residual bilinearization (RBL) step is employed in the event of unknown components in test samples. IFE can originate changes in both shape and intensity of the analyte spectra leading to trilinearity loss in both modes, invalidating most of the available methods of multiway calibration. The algorithm presented in this paper was named iSPA-U-PLS/RBL, and both simulated and experimental data sets were used to compare the prediction ability: a) simulated EEM, and b) quantitation of phenylephrine (PHE) in the presence of paracetamol (PAR) in water samples; test sets were built in both systems containing unexpected components (a single interference was taken into account in the simulated data set, while water samples were added with variable amounts of ibuprofen (IBU), and acetyl salicylic acid (ASA)). The prediction results and figures of merit obtained with the novel algorithm were compared with those obtained with U-PLS/RBL without intervals selection and the well-known parallel factors analysis (PARAFAC). In all cases U-PLS/RBL has shown capability of handling EEM with presence of IFE compared to PARAFAC. In addition iSPA was able to improve the results obtained by U-PLS/RBL, showing the potential of variable selection in this particular case.