Genetic algorithm-based method for selecting wavelengths and model size for use with partial least-squares regression: application to near-infrared spectroscopy

Application of spectroscopic techniques combined with chemometrics to the authenticity and quality attributes of rice

Rice is a staple food for two-thirds of the world's population and is grown in over a hundred countries around the world. Due to its large scale, it is vulnerable to adulteration. In addition, the quality attribute of rice is an important factor affecting the circulation and price, which is also paid more and more attention. The combination of spectroscopy and chemometrics enables rapid detection of authenticity and quality attributes in rice. This article described the application of seven spectroscopic techniques combined with chemometrics to the rice industry. For a long time, near-infrared spectroscopy and linear chemometric methods (e.g., PLSR and PLS-DA) have been widely used in the rice industry. Although some studies have achieved good accuracy, with models in many studies having greater than 90% accuracy. However, higher accuracy and stability were more likely to be obtained using multiple spectroscopic techniques, nonlinear chemometric methods, and key wavelength selection algorithms. Future research should develop larger rice databases to include more rice varieties and larger amounts of rice depending on the type of rice, and then combine various spectroscopic techniques, nonlinear chemometric methods, and key wavelength selection algorithms. This article provided a reference for a more efficient and accurate determination of rice quality and authenticity.

Ensemble calibration model of near-infrared spectroscopy based on functional data analysis

As a nondestructive detection technology, near-infrared spectroscopy has been widely applied in various fields. With the wide application of near-infrared spectroscopy, the research on data processing has attracted more attention. Different from the existing discrete data model and based on the functional data analysis method, an ensemble calibration model FDA-EM-PLS (functional data analysis-ensemble learning-partial least squares) of near-infrared spectroscopy is proposed in this paper. Firstly, the near-infrared spectroscopy of each sample is divided into several intervals, and the functional data analysis is carried out on each interval. Then, the samples are clustered according to the generated functions, which can not only reduce the influence of noise, but also provide a theoretical basis for selecting variables. Further, Monte Carlo sampling is used to generate training subsets from clustering samples for ensemble learning, which not only solves the problem of small samples, but also improves the robustness of the model. The relevant experimental results show that the absolute relative error of FDA-EM-PLS for the corn and soil data are both less than 10%.

Competitive adaptive reweighted sampling assisted partial least square analysis of excitation-emission matrix fluorescence spectroscopic data sets of certain polycyclic aromatic hydrocarbons

Excitation-emission matrix fluorescence (EEMF) spectroscopy is a simple and sensitive analytical technique. EEMF spectrum is essentially a collection of emission and excitation spectra acquired as increasing functions of excitation and emission wavelengths, respectively. EEMF spectral data sets produced per sample are highly correlated and larger in amount that need the assistance of chemometric techniques such partial least square (PLS) analysis if one desire to build robust calibration model. The objective of the PLS algorithm is to explain maximum variation of the spectral and concentration data matrices and to maximise the correlation between them. The application of a suitable variable selection technique can significantly improve the performance of PLS calibration model. Towards this, the present work proposes application of competitive adaptive reweighted sampling (CARS) as a variable selection approach prior to PLS analysis of EEMF spectral data sets. The utility of proposed approach was successfully demonstrated by analysing the significantly overlapped EEMF spectral data set of aqueous mixtures of Anthracene, Chrysene, Fluoranthene and Pyrene that are highly carcinogenic and mutagenic in nature. The developed procedure was also successfully used for the analysis of Chrysene and Pyrene mixtures in gasoline spiked ground water samples. The CARS assisted PLS model was also compared with full spectrum PLS, genetic algorithm assisted PLS, ant colony optimisation assisted PLS and N-way PLS models. The obtained results of the present work clearly indicated that application of PLS algorithm on CARS optimised EEMF spectral variables significantly improved the performance of the calibration models.

Development of a direct mapping model between hedonic rating and facial responses by dynamic facial expression representation

Consumer tests are one of the most important activities in product development. More evidence indicates that consumer emotions in real life are mostly driven by unconscious mechanisms, and implicit measurements are regarded as beneficial by an increasing number of sensory and consumer scientists. Nonverbal manner such as facial expression analysis is a supplement to the declarative method and brings very insightful results. Up until now, the facial expression analysis for consumers' acceptance identification is limited to investigate the relationship between hedonic rating and facial expression descriptors, such as facial coding system (FACS or MAX), discrete facial expressions (i.e. happiness, sadness, surprise, fear, anger, and disgust), and affective dimensional model (valence and activation). In this study, we attempt to develop a direct mapping model between the hedonic rating and facial responses evoked by various taste stimuli. Basic taste solutions (sourness, sweetness, bitterness, umami, and saltiness) with six levels, and five types of juice are used as stimuli. Firstly, the hedonic rating categories are defined based on the nine-point hedonic scale, with a coarse-to-fine division of scale levels based on two directions of like and dislike. Secondly, the facial dynamic optical flow method is employed to analyze facial characteristics of the subjects' facial responses evoked by taste stimuli. And the genetic algorithm is conducted to select facial regions that have high contribution to hedonic rating identification. It indicates that the texture changes of eye area, wrinkles at the nasal root, and mouth area can effectively reflect the facial reaction corresponding to hedonic rating. The research shows that it is feasible to establish a direct mapping model between hedonic rating and facial responses. The hedonic rating can be predicted through automatic facial reading technology, without extra transformation from predefined emotional models. In general, this is the first try to discuss the direct prediction of hedonic rating through facial expressions up to now, and it is a complex problem due to various influence factors.

High accuracy determination of copper in copper concentrate with double genetic algorithm and partial least square in laser-induced breakdown spectroscopy

There are many challenges in the determination of elements in complex matrix such as soil, coal and minerals by laser induced breakdown spectroscopy (LIBS) method. Due to the influence of matrix effect, instability of laser plasma and fluctuation of laser parameters, the repeatability and accuracy of quantitative results are always not satisfactory. In order to improve the accuracy, high-energy laser (30mJ-100mJ) with precise control was utilized in many laboratories. In this paper, quantitative analysis of copper in copper concentrate by low-energy (10µJ) LIBS is studied. In order to reduce the influence of matrix effect and other factors, a partial least square regression method based on double genetic algorithm (DGA-PLS) is proposed. The detail operations are as follow: the reference spectral lines are automatically selected by GA as the optimal internal standard for spectral normalization. Then the GA is used to select variables from the normalized spectra for PLS. The results showed that, for univariate model, the coefficient of determination (R²) was improved from 0.6 to 0.97 by the optimal internal standard normalization. Compared with tradition PLS, the root mean square error of cross validation (RMSECV) and root mean square error of prediction (RMSEP) of PLS trained by the normalized spectral data decreased from 1.4% and 0.42% to 0.9% and 0.29% respectively. Compared with the normalized PLS, the RMSECV and RMSEP of the DGA-PLS trained by the normalized and feature selected spectral data decreased from 0.9% and 0.29% to 0.26% and 0.21% respectively. The results show that DGA-PLS can significantly reduce matrix effect, improve prediction accuracy and reduce the risk of overfitting in determination of copper in copper concentrate.

A high-throughput quantification of resin and rubber contents in Parthenium argentatum using near-infrared (NIR) spectroscopy

BACKGROUND

Guayule (Parthenium argentatum A. Gray), a plant native to semi-arid regions of northern Mexico and southern Texas in the United States, is an alternative source for natural rubber (NR). Rapid screening tools are needed to replace the current labor-intensive and cost-inefficient method for quantifying rubber and resin contents. Near-infrared (NIR) spectroscopy is a promising technique that simplifies and speeds up the quantification procedure without losing precision. In this study, two spectral instruments were used to rapidly quantify resin and rubber contents in 315 ground samples harvested from a guayule germplasm collection grown under different irrigation conditions at Maricopa, AZ. The effects of eight different pretreatment approaches on improving prediction models using partial least squares regression (PLSR) were investigated and compared. Important characteristic wavelengths that contribute to prominent absorbance peaks were identified.

RESULTS

Using two different NIR devices, ASD FieldSpec^®3 performed better than Polychromix Phazir™ in improving R² and residual predicative deviation (RPD) values of PLSR models. Compared to the models based on full-range spectra (750-2500 nm), using a subset of wavelengths (1100-2400 nm) with high sensitivity to guayule rubber and resin contents could lead to better prediction accuracy. The prediction power of the models for quantifying resin content was better than rubber content.

CONCLUSIONS

In summary, the calibrated PLSR models for resin and rubber contents were successfully developed for a diverse guayule germplasm collection and were applied to roughly screen samples in a low-cost and efficient way. This improved efficiency could enable breeders to rapidly screen large guayule populations to identify cultivars that are high in rubber and resin contents.

Phenotype Profiling for Forensic Purposes: Determining Donor Sex Based on Fourier Transform Infrared Spectroscopy of Urine Traces

Forensic science is an important field of analytical chemistry where vibrational spectroscopy, in particular Fourier transform infrared spectroscopy and Raman spectroscopy, present advantages as they have a nondestructive nature, high selectivity, and no need for sample preparation. Herein, we demonstrate a method for determination of donor sex, based on attenuated total reflection Fourier transform infrared (ATR FT-IR) spectroscopy of dry urine traces. Trace body fluid evidence is of special importance to the modern criminal investigation as a source of individualizing DNA evidence. However, individual identification of a urine donor is generally difficult because of the small amount of DNA. Therefore, the development of an innovative method to provide phenotype information about the urine donor-including sex-is highly desirable. In this study, we developed a multivariate discriminant model for the ATR FT-IR spectra of dry urine to identify the donor sex. Rigorous selection of significant wavenumbers on the spectrum using a genetic algorithm enabled superb discrimination performance for the model and conclusively indicated a chemical origin for donor sex differences, which was supported by physiological knowledge. Although further investigations need to be conducted, this proof-of-concept study demonstrates the great potential of the developed methodology for phenotype profiling based on the analysis of urine traces.

Cultivar Classification of Single Sweet Corn Seed Using Fourier Transform Near-Infrared Spectroscopy Combined with Discriminant Analysis

Seed purity is a key indicator of crop seed quality. The conventional methods for cultivar identification are time-consuming, expensive, and destructive. Fourier transform near-infrared (FT-NIR) spectroscopy combined with discriminant analyses, was studied as a rapid and nondestructive technique to classify the cultivars of sweet corn seeds. Spectra with a range of 1000–2500 nm collected from 760 seeds of two cultivars were used for the discriminant analyses. Thereafter, 126 feature wavelengths were identified from 1557 wavelengths using a genetic algorithm (GA) to build simplified classification models. Four classification algorithms, namely K-nearest neighbor (KNN), soft independent method of class analogy (SIMCA), partial least-squares discriminant analysis (PLS-DA), and support vector machine discriminant analysis (SVM-DA) were tested on full-range wavelengths and feature wavelengths, respectively. With the full-range wavelengths, all four algorithms achieved a high classification accuracy range from 97.56% to 99.59%, and the SVM-DA worked better than other models. From the feature wavelengths, no significant decline in accuracies was observed in most of the models and a high accuracy of 99.19% was still obtained by the PLS-DA model. This study demonstrated that using the FT-NIR technique with discriminant analyses could be a feasible way to classify sweet corn seed cultivars and the proper classification model could be embedded in seed sorting machinery to select high-purity seeds.

Discrete Wavelet Transform (DWT) Assisted Partial Least Square (PLS) Analysis of Excitation-Emission Matrix Fluorescence (EEMF) Spectroscopic Data Sets: Improving the Quantification Accuracy of EEMF Technique

In the present work, it is shown that quantitative estimation efficiency of the partial least square (PLS) calibration model can be significantly improved by pre-processing the EEMF with discrete wavelet transform (DWT) analysis. The application of DWT essentially reduces the volume of data sets retaining all the analytically relevant information that subsequently helps in establishing a better correlation between the spectral and concentration data matrices. The utility of the proposed approach is successfully validated by analyzing the dilute aqueous mixtures of four fluorophores having significant spectral overlap with each other. The analytical procedure developed in the present study could be useful for analyzing the environmental, agricultural, and biological samples containing the fluorescent molecules at low concentration levels.

Introducing 'Simple Variable Selection (SVS) Approach' for Improving the Quantitative Accuracy of Chemometric Assisted Fluorimetric Estimations of Dilute Aqueous Mixtures

Excitation emission matrix fluorescence (EEMF) spectroscopy is a multiparametric fluorescence technique where the fluorescence intensity of a fluorophore is a function of excitation wavelength, emission wavelength and its concentration. The manual analysis of large volume of highly correlated EEMF data sets towards developing a calibration model for quantifying each fluorophores present in multifluorophoric mixtures is a difficult and time-consuming task. Over the years, Partial least square (PLS) algorithm has found its application towards providing swift and efficient analyses of large volumes of highly correlated spectral data sets. The PLS assisted EEMF spectroscopy has been successfully used towards quantifying the fluorophores in multifluorophoric mixtures without involving any pre-separation. However, the accuracy and robustness of developed calibration model can be significantly improved provided PLS analysis is carried out on the analytically relevant EEMF spectral variables. In the present work, a variable selection method baptized as simple variable selection (SVS) approach is introduced that provides a simple and computationally economical means of identifying the useful spectral variables for subsequent PLS analysis. The proposed SVS approach is successfully validated by analyzing the complex EEMF data sets of multifluorophoric mixtures of consisting of multifluorophoric mixtures of biological relevance. The proposed approach is found to provide a simple, swift and efficient means for developing a robust PLS assisted EEMF spectroscopy based calibration model for simultaneous quantification of various fluorophores present in multifluorophoric mixtures.

Detection of Nitrogen Content in Rubber Leaves Using Near-Infrared (NIR) Spectroscopy with Correlation-Based Successive Projections Algorithm (SPA)

Near-infrared spectroscopy is an efficient, low-cost technology that has potential as an accurate method in detecting the nitrogen content of natural rubber leaves. Successive projections algorithm (SPA) is a widely used variable selection method for multivariate calibration, which uses projection operations to select a variable subset with minimum multi-collinearity. However, due to the fluctuation of correlation between variables, high collinearity may still exist in non-adjacent variables of subset obtained by basic SPA. Based on analysis to the correlation matrix of the spectra data, this paper proposed a correlation-based SPA (CB-SPA) to apply the successive projections algorithm in regions with consistent correlation. The result shows that CB-SPA can select variable subsets with more valuable variables and less multi-collinearity. Meanwhile, models established by the CB-SPA subset outperform basic SPA subsets in predicting nitrogen content in terms of both cross-validation and external prediction. Moreover, CB-SPA is assured to be more efficient, for the time cost in its selection procedure is one-twelfth that of the basic SPA.

Application of Genetic Algorithm (GA) Assisted Partial Least Square (PLS) Analysis on Trilinear and Non-trilinear Fluorescence Data Sets to Quantify the Fluorophores in Multifluorophoric Mixtures: Improving Quantification Accuracy of Fluorimetric Estimations of Dilute Aqueous Mixtures

Excitation-emission matrix fluorescence (EEMF) and total synchronous fluorescence spectroscopy (TSFS) are the 2 fluorescence techniques that are commonly used for the analysis of multifluorophoric mixtures. These 2 fluorescence techniques are conceptually different and provide certain advantages over each other. The manual analysis of such highly correlated large volume of EEMF and TSFS towards developing a calibration model is difficult. Partial least square (PLS) analysis can analyze the large volume of EEMF and TSFS data sets by finding important factors that maximize the correlation between the spectral and concentration information for each fluorophore. However, often the application of PLS analysis on entire data sets does not provide a robust calibration model and requires application of suitable pre-processing step. The present work evaluates the application of genetic algorithm (GA) analysis prior to PLS analysis on EEMF and TSFS data sets towards improving the precision and accuracy of the calibration model. The GA algorithm essentially combines the advantages provided by stochastic methods with those provided by deterministic approaches and can find the set of EEMF and TSFS variables that perfectly correlate well with the concentration of each of the fluorophores present in the multifluorophoric mixtures. The utility of the GA assisted PLS analysis is successfully validated using (i) EEMF data sets acquired for dilute aqueous mixture of four biomolecules and (ii) TSFS data sets acquired for dilute aqueous mixtures of four carcinogenic polycyclic aromatic hydrocarbons (PAHs) mixtures. In the present work, it is shown that by using the GA it is possible to significantly improve the accuracy and precision of the PLS calibration model developed for both EEMF and TSFS data set. Hence, GA must be considered as a useful pre-processing technique while developing an EEMF and TSFS calibration model.

Boosting the Performance of Genetic Algorithms for Variable Selection in Partial Least Squares Spectral Calibrations

A genetic algorithm (GA) for variable selection in partial least squares (PLS) regression that incorporates adaptive boosting to identify informative wavelengths in near-infrared (NIR) spectra has been developed. Three studies demonstrating the advantages of incorporating an adaptive boosting routine into a GA that employs the root mean square error of calibration as its fitness function are highlighted: (1) prediction of hydroxyl number of terpolymers from NIR diffuse reflectance spectra; (2) calibration of acetone from NIR transmission spectra of mixtures of water, acetone, t-butyl alcohol and isopropyl alcohol; and (3) determination of the active pharmaceutical ingredients in drug tablets from NIR diffuse reflectance spectra. The performance of the GA with adaptive boosting to select wavelengths was compared with one without adaptive boosting. For all three NIR data sets, variable selected PLS models developed by a GA with adaptive boosting performed better. Analysis of the wavelengths selected by the GA with adaptive boosting also demonstrate that chemical information indicative of the analyte was captured by the selected wavelengths.

Effect of genetic algorithm as a variable selection method on different chemometric models applied for the analysis of binary mixture of amoxicillin and flucloxacillin: A comparative study

Different chemometric models were applied for the quantitative analysis of amoxicillin (AMX), and flucloxacillin (FLX) in their binary mixtures, namely, partial least squares (PLS), spectral residual augmented classical least squares (SRACLS), concentration residual augmented classical least squares (CRACLS) and artificial neural networks (ANNs). All methods were applied with and without variable selection procedure (genetic algorithm GA). The methods were used for the quantitative analysis of the drugs in laboratory prepared mixtures and real market sample via handling the UV spectral data. Robust and simpler models were obtained by applying GA. The proposed methods were found to be rapid, simple and required no preliminary separation steps.

Variable selection in multivariate calibration based on clustering of variable concept

Recently we have proposed a new variable selection algorithm, based on clustering of variable concept (CLoVA) in classification problem. With the same idea, this new concept has been applied to a regression problem and then the obtained results have been compared with conventional variable selection strategies for PLS. The basic idea behind the clustering of variable is that, the instrument channels are clustered into different clusters via clustering algorithms. Then, the spectral data of each cluster are subjected to PLS regression. Different real data sets (Cargill corn, Biscuit dough, ACE QSAR, Soy, and Tablet) have been used to evaluate the influence of the clustering of variables on the prediction performances of PLS. Almost in the all cases, the statistical parameter especially in prediction error shows the superiority of CLoVA-PLS respect to other variable selection strategies. Finally the synergy clustering of variable (sCLoVA-PLS), which is used the combination of cluster, has been proposed as an efficient and modification of CLoVA algorithm. The obtained statistical parameter indicates that variable clustering can split useful part from redundant ones, and then based on informative cluster; stable model can be reached.

A colloidal quantum dot spectrometer

Spectroscopy is carried out in almost every field of science, whenever light interacts with matter. Although sophisticated instruments with impressive performance characteristics are available, much effort continues to be invested in the development of miniaturized, cheap and easy-to-use systems. Current microspectrometer designs mostly use interference filters and interferometric optics that limit their photon efficiency, resolution and spectral range. Here we show that many of these limitations can be overcome by replacing interferometric optics with a two-dimensional absorptive filter array composed of colloidal quantum dots. Instead of measuring different bands of a spectrum individually after introducing temporal or spatial separations with gratings or interference-based narrowband filters, a colloidal quantum dot spectrometer measures a light spectrum based on the wavelength multiplexing principle: multiple spectral bands are encoded and detected simultaneously with one filter and one detector, respectively, with the array format allowing the process to be efficiently repeated many times using different filters with different encoding so that sufficient information is obtained to enable computational reconstruction of the target spectrum. We illustrate the performance of such a quantum dot microspectrometer, made from 195 different types of quantum dots with absorption features that cover a spectral range of 300 nanometres, by measuring shifts in spectral peak positions as small as one nanometre. Given this performance, demonstrable avenues for further improvement, the ease with which quantum dots can be processed and integrated, and their numerous finely tuneable bandgaps that cover a broad spectral range, we expect that quantum dot microspectrometers will be useful in applications where minimizing size, weight, cost and complexity of the spectrometer are critical.

Near infrared spectroscopic (NIRS) analysis of drug-loading rate and particle size of risperidone microspheres by improved chemometric model

Microspheres have been developed as drug carriers in controlled drug delivery systems for years. In our present study, near infrared spectroscopy (NIRS) is applied to analyze the particle size and drug loading rate in risperidone poly(d,l-lactide-co-glycolide) (PLGA) microspheres. Various batches of risperidone PLGA microspheres were designed and prepared successfully. The particle size and drug-loading rate of all the samples were determined by a laser diffraction particle size analyzer and high performance liquid chromatography (HPLC) system. Monte Carlo algorithm combined with partial least squares (MCPLS) method was applied to identify the outliers and choose the numbers of calibration set. Furthermore, a series of preprocessing methods were performed to remove signal noise in NIR spectra. Moving window PLS and radical basis function neural network (RBFNN) methods were employed to establish calibration model. Our data demonstrated that PLS-developed model was only suitable for drug loading analysis in risperidone PLGA microspheres. Comparatively, RBFNN-based predictive models possess better fitting quality, predictive effect, and stability for both drug loading rate and particle size analysis. The correlation coefficients of calibration set (Rc(2)) were 0.935 and 0.880, respectively. The performance of optimum RBFNN models was confirmed by independent verification test with 15 samples. Collectively, our method is successfully performed to monitor drug-loading rate and particle size during risperidone PLGA microspheres preparation.

Improvement of near infrared spectroscopic (NIRS) analysis of caffeine in roasted Arabica coffee by variable selection method of stability competitive adaptive reweighted sampling (SCARS)

Coffee is the most heavily consumed beverage in the world after water, for which quality is a key consideration in commercial trade. Therefore, caffeine content which has a significant effect on the final quality of the coffee products requires to be determined fast and reliably by new analytical techniques. The main purpose of this work was to establish a powerful and practical analytical method based on near infrared spectroscopy (NIRS) and chemometrics for quantitative determination of caffeine content in roasted Arabica coffees. Ground coffee samples within a wide range of roasted levels were analyzed by NIR, meanwhile, in which the caffeine contents were quantitative determined by the most commonly used HPLC-UV method as the reference values. Then calibration models based on chemometric analyses of the NIR spectral data and reference concentrations of coffee samples were developed. Partial least squares (PLS) regression was used to construct the models. Furthermore, diverse spectra pretreatment and variable selection techniques were applied in order to obtain robust and reliable reduced-spectrum regression models. Comparing the respective quality of the different models constructed, the application of second derivative pretreatment and stability competitive adaptive reweighted sampling (SCARS) variable selection provided a notably improved regression model, with root mean square error of cross validation (RMSECV) of 0.375 mg/g and correlation coefficient (R) of 0.918 at PLS factor of 7. An independent test set was used to assess the model, with the root mean square error of prediction (RMSEP) of 0.378 mg/g, mean relative error of 1.976% and mean relative standard deviation (RSD) of 1.707%. Thus, the results provided by the high-quality calibration model revealed the feasibility of NIR spectroscopy for at-line application to predict the caffeine content of unknown roasted coffee samples, thanks to the short analysis time of a few seconds and non-destructive advantages of NIRS.

An efficient method of wavelength interval selection based on random frog for multivariate spectral calibration

Wavelength selection is a critical step for producing better prediction performance when applied to spectral data. Considering the fact that the vibrational and rotational spectra have continuous features of spectral bands, we propose a novel method of wavelength interval selection based on random frog, called interval random frog (iRF). To obtain all the possible continuous intervals, spectra are first divided into intervals by moving window of a fix width over the whole spectra. These overlapping intervals are ranked applying random frog coupled with PLS and the optimal ones are chosen. This method has been applied to two near-infrared spectral datasets displaying higher efficiency in wavelength interval selection than others. The source code of iRF can be freely downloaded for academy research at the website: http://code.google.com/p/multivariate-calibration/downloads/list.

Predictive-property-ranked variable reduction with final complexity adapted models in partial least squares modeling for multiple responses

For partial least-squares regression with one response (PLS1), many variable-reduction methods have been developed. However, only a few address the case of multiple-response partial-least-squares (PLS2) modeling. The calibration performance of PLS1 can be improved by elimination of uninformative variables. Many variable-reduction methods are based on various PLS-model-related parameters, called predictor-variable properties. Recently, an important adaptation, in which the model complexity is optimized, was introduced in these methods. This method was called Predictive-Property-Ranked Variable Reduction with Final Complexity Adapted Models, denoted as PPRVR-FCAM or simply FCAM. In this study, variable reduction for PLS2 models, using an adapted FCAM method, FCAM-PLS2, is investigated. The utility and effectiveness of four new predictor-variable properties, derived from the multiple response PLS2 regression coefficients, are studied for six data sets consisting of ultraviolet-visible (UV-vis) spectra, near-infrared (NIR) spectra, NMR spectra, and two simulated sets, one with correlated and one with uncorrelated responses. The four properties include the mean of the absolute values as well as the norm of the PLS2 regression coefficients and their significances. The four properties were found to be applicable by the FCAM-PLS2 method for variable reduction. The predictive abilities of models resulting from the four properties are similar. The norm of the PLS2 regression coefficients has the best selective abilities, low numbers of variables with an informative meaning to the responses are retained. The significance of the mean of the PLS2 regression coefficients is found to be the least-selective property.

Study of a new electromagnetic sensor for glycaemia measurement:in vitroresults on blood pig

In this paper, we develop a non-invasive method to measure glycaemia levels. The method is based on an electromagnetic sensor associated with a specific electronic card. Results obtained on whole pig blood show the influence of the temperature and the linearity between the glucose concentration and the process card signal output. Measurements are reproducible and the resolution obtained permits determination of the glucose level around 0.1 g l(-1).

Digital filtering and model updating methods for improving the robustness of near-infrared multivariate calibrations

Fourier transform near-infrared (NIR) transmission spectra are used for quantitative analysis of glucose for 17 sets of prediction data sampled as much as six months outside the timeframe of the corresponding calibration data. Aqueous samples containing physiological levels of glucose in a matrix of bovine serum albumin and triacetin are used to simulate clinical samples such as blood plasma. Background spectra of a single analyte-free matrix sample acquired during the instrumental warm-up period on the prediction day are used for calibration updating and for determining the optimal frequency response of a preprocessing infinite impulse response time-domain digital filter. By tuning the filter and the calibration model to the specific instrumental response associated with the prediction day, the calibration model is given enhanced ability to operate over time. This methodology is demonstrated in conjunction with partial least squares calibration models built with a spectral range of 4700-4300 cm(-1). By using a subset of the background spectra to evaluate the prediction performance of the updated model, projections can be made regarding the success of subsequent glucose predictions. If a threshold standard error of prediction (SEP) of 1.5 mM is used to establish successful model performance with the glucose samples, the corresponding threshold for the SEP of the background spectra is found to be 1.3 mM. For calibration updating in conjunction with digital filtering, SEP values of all 17 prediction sets collected over 3-178 days displaced from the calibration data are below 1.5 mM. In addition, the diagnostic based on the background spectra correctly assesses the prediction performance in 16 of the 17 cases.