B.G.M. Vandeginste

Elimination of uninformative variables 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.

Multicomponent self-modelling curve resolution in high-performance liquid chromatography by iterative target transformation analysis

Abstract Iterative target transformation factor analysis can provide a method for resolving elution profiles consisting of any number of compounds. The results obtained for 3-component resolution are consistent with the results obtained with conventional methods of curve resolution. The same restrictions with regard to overlap and relative signal heights of the compounds seem to apply to the conventional method of curve resolution and the proposed method. The method is tested on data from high-performance liquid chromatography with a diode-array detector obtained for polynuclear aromatic hydrocarbons and for proteins.

Analysis of phospholipids in lecithins comparison between micellar electrokinetic chromatography and high-performance liquid chromatography

The determination of phospholipids in lecithins by HPLC and MEKC has been compared. MEKC conditions have been optimized to provide a robust analytical method. MEKC offers the advantage of a much higher peak capacity which results in an improved resolution, especially for phosphatidylserine. Although the repeatability of the MEKC data is somewhat lower than for HPLC, good quantitative results are obtained.

Evaluation of peak-recognition techniques in liquid chromatography with photodiode array detection

Abstract Four mathematical techniques concerned with peak recognition in liquid chromatography by means of photodiode array detection are evaluated for application in the iterative regression design optimization of the mobile phase. The techniques are multicomponent analysis, multicomponent analysis with a non-negativity constraint, target factor analysis and iterative target testing. The results obtained for two-component clusters depend to a different degree on chromatographic resolution and changes in the UV spectra caused by a variation of the eluent composition. Multicomponent analysis is independent of resolution but highly sensitive to minute changes in the spectra. Iterative target testing needs no spectral information, but requires a reasonable resolution. Provided adequate spectra are available, target factor analysis allows excellent recognition down to very low resolution.

Realistic simulations of high-performance liquid chromatographic—ultraviolet data for the evaluation of multivariate techniques

Abstract In this paper it will be shown that real high-performance liquid chromatographic—ultraviolet (HPLC-UV) data do not perfectly obey the exact linear model assumed in most multivariate techniques. These small model errors may have an influence on the performance of multivariate techniques, especially on the principal component analysis step in which a reduced multivariate space is defined. In order to obtain a more realistic picture of the performance of multivariate techniques in practice, simulations of peak clusters are proposed that make use of real peak systems of pure analytes. It will be shown that these realistically simulated data contain the same small model errors that can be observed for real HPLC-UV data.

STRATEGY FOR PEAK TRACKING IN LIQUID CHROMATOGRAPHY ON THE BASIS OF A MULTIVARIATE ANALYSIS OF SPECTRAL DATA

Abstract Peak tracking, required for the interpretive optimization of the mobile phase in reversed-phase high-performance liquid chromatography, is performed by means of multiwavelength detection followed by a multivariate analysis of the resulting data. A combination of techniques (target factor analysis and iterative target transformation factor analysis) is necessary to circumvent problems caused by a limited resolution or lack of knowledge regarding spectral characteristics and to determine the number of relevant components. A procedure to perform the actual peak tracking, which combines data related to the spectral characteristics and relative peak areas of separated and coeluting components, is presented.

An expert system for designing an intelligent spreadsheet for evaluation of precision of liquid chromatographic methods

Abstract Method validation is increasing in importance as higher standards of precision and accuracy are required. The validation of liquid chromatographic methods requires the examination of selectivity, sensitivity, accuracy, precision and limitations, depending on the purpose of the method. The precision is affected by the repeatability of procedural steps (e.g., injection), within the method and by its overall reproducibility (e.g., interlaboratory reproducibility). The application of spreadsheets with an expert system to evaluate these effects is described. The spreadsheets are programmed to produce the necessary statistics. Rules are defined for the expert system to interact with these to select suitable tests for different applications, to set up spreadsheets into which the analyst inputs the data, and to interpret the data processed in the spreadsheet. The system is tested with a repeatability study of the separation of aspirin and salicylic acid. A combination of the Goldworks software for development of expert systems and the Lotus 1-2-3 spreadsheet package was used.

Strategies in molecular spectroscopic analysis with application of queueing theory and digital simulation

Abstract Waiting lines in a laboratory for molecular spectroscopie analysis are simulated by digital methods. The influence of several parameters, such as the mean analysis time, the mean inter-arrival time of the samples and the routing procedure are evaluated by application of queueing theory. The queues were treated as a sample realization from a stationary first-order autoregressive system. The structure of the time series of the number of waiting samples in the model was identical to the time series observed in a real laboratory. The possibilities of digital simulation for decision-making in analytical laboratories.are demonstrated.

Data handling in HPLC-diode array UV-spectrometry

SummaryThe mathematical resolution of overlapped chromatographic peaks obtained in HPLC with a diode array UV detector has received considerable attention recently. The purpose of the proposed methods is the quantitation and identification of unknown solutes in complex mixtures by an efficient use of all available data.Basically two approaches have been proposed: the first one resolves the concentration profiles and calculates the pure spectra by applying a minimal number of assumptions, which is denoted self-modeling-curve resolution. The second one is based on a match of pure spectra available in a spectral library with the measured mixture spectra.In this paper both approaches are evaluated with respect to their performance to provide the pure spectra and an accurate quantitation of the concentrations.

Expert systems and analytical chemistry

Les systemes experts. Quelques exemples. Fonctionnement. Revue des systemes experts en chimie. 22 references