Search

A difficulty when applying partial least squares (PLS) in multivariate calibration is that overfitting may occur. This study proposes a novel approach by combining PLS and boosting. The latter is said to be resistant to overfitting. The proposed method, called boosting PLS (BPLS), combines a set of shrunken PLS models, each with only one PLS component. The method is iterative: the models are constructed on the basis of the residuals of the responses that are not explained by previous models. Unlike classical PLS, BPLS does not need to select an adequate number of PLS components to be included in the model. On the other hand, two parameters must be determined: the shrinkage value and the iteration number. Criteria are proposed for these two purposes. BPLS was applied to seven real data sets, and the results demonstrate that it is more resistant than classical PLS to overfitting without loosing accuracy.

Several methods estimating the partitioning over biological membranes and thus the biological activity of potential oral drug molecules have been developed and are described in the literature. A previous study suggested that fast micellar liquid chromatography on a monolithic column could be one of them. For a set of diverse pharmaceuticals, retention by this fast chromatographic method was determined, besides other parameters also thought or established to describe oral permeability or absorption, e.g., from the Caco-2 permeability method. In view of a high-throughput determination of membrane permeability, a study was made of which information fast micellar liquid chromatography is providing and to what degree this system can replace other methods, i.e., deliver similar information. The retention with this fast method, which is mainly based on hydrophobic interactions, proved useful to sort substances into classes of Caco-2 and percent intestinal absorption.

The recently introduced monolithic silica columns were tested for their use in micellar liquid chromatography. Micellar methods are utilized in high-throughput quantitative structure-retention relationships to estimate an indicator of the membrane permeability of drugs, namely, the octanol-water partition coefficient, log P. The monolithic column's ability to function at higher flow rates might be useful to speed up these chromatographic methods estimating the log P. Therefore, the elution behavior of diverse basic pharmaceutical substances was determined on a classical particle-based and a monolithic column, both with and without a micellar medium in the mobile phase. Utilizing among others principal component analysis, the extent to which these methods differ in retention characteristics was examined in the context of high-throughput determination of log P. Results indicate that combining monolithic columns with micellar media leads to faster log P and possibly even better permeability predictions.

The discrete wavelet transform was applied to denoise electropherograms in capillary electrophoresis (CE). The use of the Haar wavelet and translation invariant denoising were found to be very efficient for this purpose. An important improvement was obtained, as compared with Savitzky-Golay and Fourier, which are the most commonly used techniques for denoising in the instrumentation software packages. A better removal of the noise and, especially, a better preservation of the shapes of very sharp peaks was achieved. Removal of the baseline variations was also investigated.

Sequential projection pursuit (SPP) is proposed to detect inhomogeneities (clusters) in high-dimensional analytical data. Such inhomogeneities indicate that there are groups of objects (samples) with different chemical characteristics. The method is compared with principal component analysis (PCA). PCA is generally applied to visually explore structure in high-dimensional data, but is not specifically used to find clustering tendency. Projection pursuit (PP) is specifically designed to find inhomogeneities, but the original method is computationally very intensive. SPP combines the advantages of both methods and overcomes most of their weak points. In this method, latent variables are obtained sequentially according to their importance measured by the entropy index. This involves an optimization step, which is achieved by using a genetic algorithm. The performance of the method is demonstrated and evaluated, first on simulated data sets, and then on near-infrared and gas chromatography data sets. It is shown that SPP indeed reveals more easily information about inhomogeneities than PCA.

Supersaturated designs are factorial designs in which the number of factors examined exceeds the number of experiments performed. They do not allow an estimation of individual effects, since even the main effects are confounded. However, the total variance of a response estimated from such a design can in principle be used as a measure for the robustness of the method. A number of case studies were examined to determine whether the variance estimated from a supersaturated design is similar to the one from a Plackett-Burman design. This was found to be the case, which means that the estimated variance describes well the variation in the response caused by the variation in the factors.

A comparative study involving a global linear method (partial least squares), a local linear method (locally weighted regression), and a nonlinear method (neural networks) has been performed in order to implement a calibration model on an industrial process. The models were designed to predict the water content in a reactor during a distillation process, using in-line measurements from a near-infrared analyzer. Curved effects due to changes in temperature and variations between the different batches make the problem particularly challenging. The influence of spectral range selection and data preprocessing has been studied. With each calibration method, specific procedures have been applied to promote model robustness. In particular, the use of a monitoring set with neural networks does not always prevent overfitting. Therefore, we developed a model selection criterion based on the determination of the median of monitoring error over replicate trials. The back-propagation neural network models selected were found to outperform the other methods on independent test data.

The application of locally weighted regression (LWR) to nonlinear calibration problems and strongly clustered calibration data often yields more reliable predictions than global linear calibration models. This study compares the performance of LWR that uses PCR and PLS regression, the Euclidean and Mahalanobis distance as a distance measure, and the uniform and cubic weighting of calibration objects in local models. Recommendations are given on how to apply LWR to near-infrared data sets without spending too much time in the optimization phase.

An approach aiming at extracting the relevant component for multivariate calibration is introduced, and its performance is compared with the "uninformative variable elimination" approach and with the standard PLS method for the modeling of near-infrared data. The extraction of the relevant component is carried out in the wavelet domain. The PLS results on these relevant features are better, and therefore, it seems that this approach can successfully be used to remove noise and irrelevant information from spectra for multivariate calibration.

A new method for the elimination of uninformative variables in multivariate data sets is proposed. To achieve this, artificial (noise) variables are added and a closed form of the PLS or PCR model is obtained for the data set containing the experimental and the artificial variables. The experimental variables that do not have more importance than the artificial variables, as judged from a criterion based on the b coefficients, are eliminated. The performance of the method is evaluated on simulated data. Practical aspects are discussed on experimentally obtained near-IR data sets. It is concluded that the elimination of uninformative variables can improve predictive ability.

The application of eigenstructure tracking analysis (ETA) and SIMPLISMA for the investigation of the protonation equilibria of a monomer and several polynucleotides is proposed. Both approaches have been applied in the pH and in the wavelength direction to the spectroscopic data matrices obtained in the study of each equilibrum. ETA provides information about the number of components in the system, their evolution along the titration, and the local rank. SIMPLISMA is also used to obtain the number of compounds in the system, the concentration profiles, and the unit spectrum of each compound. The results obtained with SIMPLISMA and those obtained previously with the alternating least-squares approach are compared.

A strategy to select reference or other test materials on which method validation experiments should be performed is developed for the AAs analysis of foodstuffs. Representative materials are selected for each category of food (meat, cereals, vegetables, etc.) and for the complete food spectrum. A principal component analysis and two cluster analysis procedures are applied to accomplish this.

The orthogonal projection approach (OPA), a stepwise approach based on an orthogonalization algorithm, is proposed. The performance of OPA for the assessment of peak purity in HPLC-DAD is described and compared with that of SIMPLISMA. The occurrence of artifacts in both approaches under nonideal situations is discussed.