Search

The least squares support vector machines (LS-SVM) was used to model infrared spectral data for TSH hormone secreted by thyroid, which regulates the basal metabolic rate. This model was used for direct estimation of the content of TSH in blood serum samples, and the results were comparable with those obtained with the conventional analytical method based on chemoluminescence methodology. Excellent agreement was observed between the conventional method and the newly developed calibration model based in analysis of spectral data with LS-SVM. The latter has clear advantages, because it is fast and requires no reagent once the measurements were done directly in the serum by using a simple mid-infrared spectrometer in the ATR mode. An important advantage observed in this calibration method based on LS-SVM is the remarkable capacity to avoid overfitting in the model-building step, that is, the developed method is highly robust.

The discrimination of the bacteria that cause pharyngitis through classical–microbiological methods is very efficient in the great majority of the cases. However, the high cost of chemicals and the time spent for such identifications, about four days, could generate serious consequences for the patients. Thus, the search for low cost spectroscopic methods which would allow a fast and reagentless discrimination of these microorganisms is extremely relevant. In this work, the main microorganisms that cause pharyngitis: S. aureus, S. pyogenes and Neisseria gonorrhoeae were studied. For each of the microorganisms 60 different dispersions were prepared using physiological solution as solvent and its Raman spectra were recorded. The 1D spectra obtained were similar, making it very difficult to differentiate the microorganisms. However, applying the 2D correlation method, it was possible to identify the microorganisms evaluated using the synchronous spectrum as “whole-organism fingerprinting” in a reduced time interval (∼10 h).

This paper reports the use of least-squares support vector machines (LS-SVM) for non-linear multivariate calibration in the determination of the alcohol content in the Brazilian spirit “cachaça” using near infrared spectroscopy. Fifty cachaça samples, with alcohol contents in the range of 20.9% to 46.5% v/v were used and the spectra were obtained at five different temperatures: 15°C, 20°C, 25°C, 30°C and 35°C. Two models were proposed: in the first, a single model was built, using the spectra from all five temperatures. In the second, the calibration set was composed of the spectra taken at four temperatures and the validation set was composed of spectra of the other temperature. All the combinations were made. In four of them, LS-SVM produced better predictions than PLS and in the other, the results were the same. These results indicate that LS-SVM can be an alternative when there is an influence of some physical variations, such as temperature, on near-infrared spectra.

The isomeric triterpenes oleanoic acid and ursolic acid are compounds exhibiting a variety of biological activities. Structurally, they differ only in the position of the methyl group (C-29) at ring E. The differentiation of these two compounds requires a detailed analysis of their 13C and 1H NMR spectra which is often tedious and time-consuming, besides the need of using deuterated solvents. In this work, we report the use of bidimensional Raman spectroscopy as a fast technique to distinguish these two bioactive isomeric compounds.

This paper proposes the use of the least-squares support vector machine (LS-SVM) as an alternative multivariate calibration method for the simultaneous quantification of some common adulterants (starch, whey or sucrose) found in powdered milk samples, using near-infrared spectroscopy with direct measurements by diffuse reflectance. Due to the spectral differences of the three adulterants a nonlinear behavior is present when all groups of adulterants are in the same data set, making the use of linear methods such as partial least squares regression (PLSR) difficult. Excellent models were built using LS-SVM, with low prediction errors and superior performance in relation to PLSR. These results show it possible to built robust models to quantify some common adulterants in powdered milk using near-infrared spectroscopy and LS-SVM as a nonlinear multivariate calibration procedure.

In this article, a new approach called partial least squares (PLS) pruning is described for variable selection in PLS modeling. The aim of the method is the deletion of unimportant PLS coefficients of regression by using information from all second derivatives of the error function. The proposed approach was applied to Brix determination in sugar cane juice by near infrared spectroscopy. The results obtained were promising, leading to a meaningful variable reduction of 96% without loss of model prediction capability.

A neuro-genetic approach was developed to optimise an automated procedure for the spectrophotometric determination of cobalt in bovine liver, fish and mussel samples by using a monosegmented flow system. The method exploited the Co(II) catalysed oxidation of Tiron by hydrogen peroxide. The values related to reagent concentrations and mean residence time were optimised in terms of slope and correlation coefficient of the calibration curve. The neural networks were used to model the system, relating the parameters of the monosegmented flow system with those of the calibration curve. The genetic algorithm was applied to attain the optimum values from the model developed by the neural networks. The detection and quantification limits were 1.66 and 5.33 ng l 1 Co, respectively. © 2001 Elsevier Science B.V. All rights reserved.

A partial least squares (PLS) calibration model was developed for the simultaneous spectrophotometric determination of Al and Fe in plant extracts by using an excess of xylenol orange as the cromogenic reagent. Experimental conditions were established to reduce interferences, decrease system complexity and produce a robust procedure that could be used for routine analysis. The spectra should be recorded from 2 to 4 h after mixing the reagents. Ethanolic solutions were used to improve the formation of the Al-XO complexes. The best calibration model was obtained using the PLS-2 algorithm after mean centering the data with five latent variables. Under the experimental conditions adopted, only Zn was found to be a potential interfering species with the plant extracts used. Validation of the proposed spectrophotometric method was performed by running nine plant samples from the 1997 International Plant-Analytical Exchange (IPE) which were also analyzed by ICP-AES for further correlation. The proposed procedure showed to be useful for prediction of Al and Fe values from 12.5 to 2500 mg kg −1 in plant dry weight basis.

Neural networks with pruning were applied to model overlapped peaks obtained in differential pulse voltammetry (DPV) with modified carbon fibre electrode with TiO 2 of binary mixtures of catechol and hydroquinone. The best condition for electrochemical response was obtained with 0.05 mol l 1 Tris‐HCl buffer at pH 6.0 and T-800 sized carbon fibre electrode. Initially the voltammograms were processed using Fourier transform filter and principal component analysis (PCA) to noise reduction and data compression, respectively. The scores of these principal components were the input into the neural network and the optimal brain surgeon (OBS) was the procedure employed for pruning the neural network. The results obtained with pruning procedure were slightly better in relation to hydroquinone in comparison to the PLS1 and PLS2. However, the similar errors were obtained to catechol when using PLS or neural networks models. Using neural networks with pruning was possible to determine catechol and hydroquinone by DPV using carbon fibre electrode, in concentration range of 1.0 10 4 up to 6.010 4 mol l 1 with root mean square errors of predictions (%RMSEP) of 7.42 and 8.02, respectively. The good results show that the proposed methodology is a good alternative to simultaneous determination of catechol and hydroquinone in binary mixtures. © 2000 Elsevier Science B.V. All rights reserved.

The need for mathematical methods to model data in energy-dispersive x-ray fluorescence (EDXRF) spectrometry is common owing to the overlapping of intense spectral lines in complex samples. This overlapping generally produces a large amount of scatter in the analytical curve, preventing simultaneous direct determinations of some elements without data treatment. This work demonstrates the performance of artificial neural networks (ANN) and other methods of multivariate calibration (linear or not) for the simultaneous determination of sulfur and lead, when overlapping of the sulfur Kα spectral line (2.308 keV) and the lead Mα line (2.346 keV) is observed. The performance of neural networks was compared by the f-test with five other data treatment methods: PLS (partial least squares), POLYPLS (polynomial partial least squares), NNPLS (partial least square neural networks), LR (linear regression) and CI (corrected intensity). It was verified that the ANN produces better predictions than the other methods, for both sulfur and lead, allowing their simultaneous determination in solid samples with good accuracy. Copyright © 1999 John Wiley & Sons, Ltd.

The pruning neural network, based on the algorithm called optimum brain surgeon, was used for network architecture optimization. This network pruning procedure was applied for estimating the nitrogen contents in wheat leaves, using near-infrared diffuse reflectance spectroscopy. The results obtained with pruning were compared with those obtained by using ordinary procedures with neural networks, partial least squares, polynomial partial least squares and neural networks/partial least squares methodologies. Comparison of the results with those obtained by the conventional Kjeldahl method showed that the results with pruning neural networks were as good as those with ordinary neural networks and with PLS/neural networks, but better than those with the other methodologies. Although the comparison was performed for one data set, the pruning procedure has the advantage of introducing an automatic architecture optimization, which is cumbersome when performed by the other neural network procedures used in this work, generating a simplified model with better generalization abilities.

This paper presents the use of least-squares support vector machine (LS-SVM) for quantitative determination of hydroxyl value (OHV) of hydroxylated soybean oils by horizontal attenuated total reflection Fourier transform infrared (HATR/FT-IR) spectroscopy. A least-squares support vector machine (LS-SVM) calibration model for the prediction of hydroxyl value (OHV) was developed using the range 1805.1-649.9 cm(-1). Validation of the method was carried out by comparing the OHV of a series of hydroxylated soybean oil predicted by the LS-SVM model to the values obtained by the AOCS standard method. A correlation coefficient equal to 0.989 and RMSEP = 4.96 mg of KOH/g was obtained. This study demonstrates a better prediction ability of the LS-SVM technique to determine OHV in hydroxylated soybean oil samples by HATR/FT-IR spectra.