Root Mean Square Error Of Cross-validation Research Articles

Analysis of Lard in Sausage Using Fourier Transform Infrared Spectrophotometer Combined with Chemometrics.

Sausage is one of foods which must be confirmed halal to consumers. Meat is commonly used in producing sausages, especially beef. However, due to high cost of meat producer usually mixes the ingredients with other cheaper meats, such as pork. This study aimed to analyze the differences in the spectral profile of lard and beef in the sausages using Fourier transform infrared (FTIR). Lard and beef tallow was extracted using Soxhlet apparatus at ±70°C for 6 h with n-hexane. After extraction, lard and beef tallow was evaporated. Then obtained fats were stored in eppendorf and analyzed using FTIR spectrophotometer. The results were then combined with chemometrics such as Partial least squares (PLS) for the quantitative analysis and principal component analysis (PCA) for classification. PLS and PCA analysis was performed on 1200–1000 cm–1. The results of the analyzed PLS provided the linear regression equation y = 0.921x + 4.623 with R2 = 0.985 and root-mean-square error of calibration (RMSEC) = 2.094%. External validation root-mean-square error of prediction (RMSEP) was 4.77% and internal validation root-mean-square-error cross-validation (RMSECV) was 5.12%. The results of the PCA analysis showed the classification of different quadrants between 100% pork sausage and 100% beef sausage. Thus, it can be concluded that FTIR spectroscopy method combined with chemometrics can be applied to identify the presence of pork in the sausage.

A variable importance criterion for variable selection in near-infrared spectral analysis

Variable selection is a universal problem in building multivariate calibration models, such as quantitative structure-activity relationship (QSAR) and quantitative relationships between quantity or property and spectral data. Significant improvement in the prediction ability of the models can be achieved by reducing the bias induced by the uninformative variables. A new criterion, named as C, is proposed in this study to evaluate the importance of the variables in a model. The value of C is defined as the average contribution of a variable to the model, which is calculated by the statistics of the models built with different combinations of the variables. In the calculation, a large number of partial least squares (PLS) models are built using a subset of variables selected by randomly re-sampling. Then, a vector of the prediction errors, in terms of root mean squared error of cross validation (RMSECV), and a matrix composed of 1 and 0 indicating the selected and unselected variables can be obtained. If multiple linear regression (MLR) is employed to model the relationship between the RMSECVs and the matrix, the coefficients of the MLR model can be used as a criterion to evaluate the contribution of a variable to the RMSECV. To enhance the efficiency of the method, a multi-step shrinkage strategy was used. Comparison with Monte Carlo-uninformative variables elimination (MC-UVE), randomization test (RT) and competitive adaptive reweighted sampling (CARS) was conducted using three NIR benchmark datasets. The results show that the proposed criterion is effective for selecting the informative variables from the spectra to improve the prediction ability of models.

Tuning CdTe quantum dots reactivity for multipoint detection of mercury(II), silver(I) and copper(II)

In this work we tuned the surface chemistry of distinctly-sized CdTe quantum dots (QDs), by using a variety of capping ligands with distinctive functional moieties, and combined them in multi-QDs probing solutions for Hg2+, Cu2+ and Ag+. Upon the individual interaction of the distinctly-capped QDs with each of the abovementioned metal ions, several parameters such as Stern-Volmer constants, equilibrium binding constants, the red-shift magnitude and photoluminescence (PL) lifetime values were thoroughly evaluated.Due to their great reactivity, selectivity could become a major issue in the application of semiconductor quantum dots in chemical analysis. Therefore, the combination in the same analysis of multiple QDs, capped with distinct ligands and emitting at different wavelengths, was exploited to guarantee a specific analyte-response profile. When coupled with chemometrics, for data analysis and processing, the developed approach provided enhanced selectivity and accuracy. In fact, the obtained results of all the partial least squares (PLS) models coefficient of determination of cross-validation (R2CV) close to 1, root mean square error of calibration (RMSEC) and root mean square error of cross-validation (RMSECV) values ranging 0.0094–0.15 and 0.013–0.19 mg L−1, respectively. demonstrated the possibility to perform accurate predictions of Hg2+ and Cu2+. Additionally, the sensitivity of the developed models was also calculated and allowed us to conclude which combination of distinctly-capped QDs assured a more sensitive quantification of Cu2+ (sensitivity = 653 L−1 mg for 3-mercaptopropionic acid (MPA)/MPA conjugated probe) and Hg2+ (sensitivity = 14,017 L−1 mg for reduced glutathione (GSH)/MPA-conjugated probe).

Determination of Stevioside and Rebaudioside A in <i>Stevia rebaudiana</i> Bertoni Leaves Using near Infrared Spectroscopy and Multivariate Data Analysis

Rebaudioside A and stevioside are abundant steviol glycoside contained in Stevia rebaudiana leaves. These components are widely used as a natural sweetener. The objective of this study was to develop rapid determination method of stevioside, and rebaudioside A in S. rebaudiana leaves using near infrared trans-reflectance spectroscopy (NIRS) combined with multivariate analysis. The reference method used was HPLC. A prediction model was developed using partial least square (PLS) regression. Calibration parameters were calculated based on a calibration set of various stevioside, rebaudioside A from 23 samples. Performance of PLS model was assessed in term of optimum determination coefficient (R2), and minimum root mean square error of cross-validation (RMSEV). Validation of PLS model was performed using cross-validation and leave one out calibration of PLS component. Rebaudioside A has well PLS model in wavenumber region of 4100–5100 cm-1, and stevioside determination using difference wavenumber region of 4760-5016 cm-1. PLS model for total (sum of stevioside and rebaudioside A content) was exploited in wavenumber region of 4568-4928 cm-1. NIRS in combination with multivariate data analysis of PLSR can be applied as a rapid method for determining rebaudioside A and the total amount of steviol glycosides in S. rebaudiana leaves.

Non-destructive evaluation of wood stiffness and fiber coarseness, derived from SilviScan data, via near infrared hyperspectral imaging

Near infrared hyperspectral imaging combined with partial least squares regression analysis was used to evaluate wood stiffness (modulus of elasticity) and fiber coarseness. Five samples with normal wood and compression wood collected from two Japanese Cedar ( Cryptomeria japonica) trees were analyzed. To achieve high reliability of the prediction values, a SilviScan system (X-ray densitometry, X-ray diffractometry, and optical microscopy) with the high spatial resolution was used for measuring reference data. The measurement interval for modulus of elasticity and fiber coarseness was 1 mm and 25 µm, respectively. After spectral pre-treatment and key wavelengths selection, partial least squares analysis was applied to calibrate near infrared data to reference values. The determination coefficient ( RCV2) of modulus of elasticity was 0.66 with a root mean square error of cross validation (RMSECV) of 1.80 GPa. For the constructed fiber coarseness calibration model, RCV2 and RMSECV were 0.62 and 35.02 µm/g, respectively. Finally, modulus of elasticity and fiber coarseness mapping results show detailed information (156 µm/pixel) at the grown ring level. The differences among earlywood, latewood, and compression wood were all well identifiable.

Representative splitting cross validation

Cross-validation (CV) is widely used to estimate model complexity or the number of significant latent variables (LVs) for multivariate calibration methods like partial least squares (PLS). A basic consideration when developing and validating multivariate calibration models is that both the training and validation sets should be representative and distributed in the experimental space as uniformly as possible. Motivated by this idea, we proposed a new CV method called representative splitting cross-validation (RSCV). In RSCV, firstly, the DUPLEX algorithm was used to sequentially divide the original training set into k (in this work, k = 2, 4, 8 and 16) equal parts. Secondly, a series of k-fold (k = 2, 4, 8 and 16) CVs were performed based on the above data splitting. Finally, the pooled root mean squared error of CV (RMSECV) was used to estimate model complexity. Five real multivariate calibration data sets were investigated and RSCV was compared with leave-one-out CV (LOOCV), 10-fold CV and Monte Carlo CV (MCCV). With a maximum k of 16, RSCV was shown to be a useful and stable method to select PLS LVs, and can obtain simpler models with acceptable computational burden.

Discrimination of organic and conventional rice by chemometric analysis of NIR spectra: a pilot study

A pilot study was conducted to develop nondestructive calibration models to discriminate organic and conventional rice from selected field trials in Heilongjiang Province, China using near-infrared (NIR) spectroscopy with the absorption mode in the wave number range of 12000–4000 cm−1. Multivariate methods such as principal component analysis (PCA) and partial least squares (PLS) regression were used to interpret the NIR spectral data. PLS regression was used for discrimination between organic and conventional rice samples after several pretreatments of the spectra. The coefficient of determination (R2) value for the PLS regression model was 0.8430 with a standard error for cross validation (SECV) of 0.1992 and a root mean square error for cross validation (RMSECV) of 0.1982. Overall, the results indicated good performance of the prediction models and supported the capability of NIR spectroscopy to classify between discriminate organic and conventional rice. This study further supports the utilization of NIR in the discriminative analysis of foods and as a noteworthy method for the authentication of organic rice at the industrial level.

In-situ monitoring of saccharides removal of alcohol precipitation using near-infrared spectroscopy

As unsafe components in herbal medicine (HM), saccharides can affect not only the drug appearance and stabilization, but also the drug efficacy and safety. The present study focuses on the in-line monitoring of batch alcohol precipitation processes for saccharide removal using near-infrared (NIR) spectroscopy. NIR spectra in the 4000–10,000-cm[Formula: see text] wavelength range are acquired in situ using a transflectance probe. These directly acquired spectra allow characterization of the dynamic variation tendency of saccharides during alcohol precipitation. Calibration models based on partial least squares (PLS) regression have been developed for the three saccharide impurities, namely glucose, fructose, and sucrose. Model errors are estimated as the root-mean-square errors of cross-validation (RMSECVs) of internal validation and root-mean-square errors of prediction (RMSEPs) of external validation. The RMSECV values of glucose, fructose, and sucrose were 1.150, 1.535, and 3.067[Formula: see text]mg[Formula: see text]mL[Formula: see text], and the RMSEP values were 0.711, 1.547, and 3.740[Formula: see text][Formula: see text], respectively. The correlation coefficients [Formula: see text] between the NIR predictive and the reference measurement values were all above 0.94. Furthermore, NIR predictions based on the constructed models improved our understanding of sugar removal and helped develop a control strategy for alcohol precipitation. The results demonstrate that, as an alternative process analytical technology (PAT) tool for monitoring batch alcohol precipitation processes, NIR spectroscopy is advantageous for both efficient determination of quality characteristics (fast, in situ, and requiring no toxic reagents) and process stability, and evaluating the repeatability.

Noninvasive Determination of Firmness and Dry Matter Content of Stored Onion Bulbs Using Shortwave Infrared Imaging with Whole Spectra and Selected Wavelengths.

A firm texture of dry onions is important for consumer acceptance. Both the texture and dry matter content decline during storage, influencing the market value of onions. The main goal of this study was to develop predictive models that in future might form the basis for automated sorting of onions for firmness and dry matter content in the industry. Hyperspectral scanning was conducted in reflectance mode for six commercial batches of onions that were monitored three times during storage. Mean spectra from the region of interest were extracted and partial least squares regression (PLSR) models were constructed. Feature wavelengths were identified using variable selection techniques resulting from interval partial least squares and recursive partial least squares analyses. The PLSR model for firmness gave a root mean square error of cross-validation (RMSECV) of 0.84 N, and a root mean square error of prediction (RMSEP) of 0.73 N, with coefficients of determination ( R2) of 0.72 and 0.83, respectively. The RMSECV and RMSEP of the PLSR model for dry matter content were 0.10% and 0.08%, respectively, with a R2 of 0.58 and 0.79, respectively. The whole wavelength range and selected wavelengths showed nearly similar results for both dry matter content and firmness. The results obtained from this study clearly reveal that hyperspectral imaging of onion bulbs with selected wavelengths, coupled with chemometric modeling, can be used for the noninvasive determination of the firmness and dry matter content of stored onion bulbs.

Canopy Hyperspectral Sensing of Paddy Fields at the Booting Stage and PLS Regression can Assess Grain Yield

Canopy hyperspectral (HS) sensing is a promising tool for estimating rice (Oryza sativa L.) yield. However, the timing of HS measurements is crucial for assessing grain yield prior to harvest because rice growth stages strongly influence the sensitivity to different wavelengths and the evaluation performance. To clarify the optimum growth stage for HS sensing-based yield assessments, the grain yield of paddy fields during the reproductive phase to the ripening phase was evaluated from field HS data in conjunction with iterative stepwise elimination partial least squares (ISE-PLS) regression. The field experiments involved three different transplanting dates (12 July, 26 July, and 9 August) in 2017 for six cultivars with three replicates (n = 3 × 6 × 3 = 54). Field HS measurements were performed on 2 October 2017, during the panicle initiation, booting, and ripening growth stages. The predictive accuracy of ISE-PLS was compared with that of the standard full-spectrum PLS (FS-PLS) via coefficient of determination (R2) values and root mean squared errors of cross-validation (RMSECV), and the robustness was evaluated by the residual predictive deviation (RPD). Compared with the FS-PLS models, the ISE-PLS models exhibited higher R2 values and lower RMSECV values for all data sets. Overall, the highest R2 values and the lowest RMSECV values were obtained from the ISE-PLS model at the booting stage (R2 = 0.873, RMSECV = 22.903); the RPD was >2.4. Selected HS wavebands in the ISE-PLS model were identified in the red-edge (710–740 nm) and near-infrared (830 nm) regions. Overall, these results suggest that the booting stage might be the best time for in-season rice grain assessment and that rice yield could be evaluated accurately from the HS sensing data via the ISE-PLS model.

Rapid Prediction of Low (

FDA ruled that partially hydrogenated oils were no longer "generally recognized as safe (GRAS)" for any use in human food. Consequently, we have proposed a rapid screening procedure, based on infrared spectroscopy and chemometrics, to rapidly and accurately predict low concentrations of trans fatty acids (<1% of total fatty acids) in edible oils and food lipid extracts.

Examination of Dimension Reduction Performances of PLSR and PCR Techniques in Data with Multicollinearity

One of the common ways to cope with the multicollinearity problem in multiple regression analysis is to use dimension reduction techniques. Among these techniques, the present study focuses on the Partial Least Square Regression (PLSR) and the Principle Component Regression (PCR) techniques. The study tries to determine in which cases the two techniques give similar results and in which cases and to what extent they are different in terms of dimension reduction. For this purpose, the performance of the techniques is examined on two real dataset. In addition, a Monte Carlo simulation is made to evaluate the performances of these techniques based on the criterion of Root Mean Square Error of Cross Validation (RMSECV) under different conditions.

Determination of salvianolic acid B and borneol in compound Danshen tablet by near-infrared spectroscopy and establishment of dependency model

To determine the contents of salvianolic acid B and borneol in compound Danshen tablet (composed of Salviae Miltiorrhizae Radix et Rhizome, Notoginseng Radix et Rhizome and Borneolum Syntheticum) by near-infrared spectroscopy (NIR) and to establish a dependency model for rapid quantitative analysis. NIR data of 74 batches of compound Danshen tablet from different companies were collected; the contents of salvianolic acid B and borneol were determined by using high performanceliquid chromatography (HPLC) and gas chromatography (GC) respectively to establish the dependency model for salvianolic acid B and borneol. The results showed that the best waveband for salvianolic acid B was 10 846.2-10 013, 9 195.3-8 362.2, 6 719.1-4 242.8 cm⁻¹; root-mean-squares error of cross-validation (RMSECV), coefficient of determination R² and regression point displacement (RPD) of salvianolic acid B were 1.72 mg·g⁻¹, 91.05% and 7.93 mg·g⁻¹, respectively. While the best wavebands, RMSECV, R² and RPD of borneol were 10 846.2-5 060.5 cm⁻¹, 1.2 mg·g⁻¹, 96.11% and 6.71 mg·g⁻¹，respectively. The relative error of the established model as validation results for validation set samples was 2.67% and 4.64%， respectively.With the good predictability, the models can be applied to the real time monitoring and quality control of compound Danshen tablet.

Multivariate Validated Models for Simultaneous Determination of Ibuprofen and Famotidine in the Presence of Related Substances.

Two multivariate validated spectrophotometric methods, namely partial least-squares (PLS) and principal component regression (PCR), were developed and validated for the determination of ibuprofen and famotidine in presence of famotidine degradation products and ibuprofen impurity (4-isobutylacetophenone). A calibration set was prepared in which the two drugs together with the degradation products and impurity were modeled using a multilevel multifactor design. This calibration set was used to build the PLS and PCR models. The proposed models successfully predicted the concentrations of both drugs in validation samples, with low root mean square error of cross validation (RMSECV) percentage. The method was validated by the estimate of the figures of merit depending on the net analyte signal. The results of the two models showed that the simultaneous determination of both drugs could be performed in the concentration ranges of 100-500 µg/mL for ibuprofen and 5-25 µg/mL for famotidine. The proposed multivariate calibration methods were applied for the determination of ibuprofen and famotidine in their pharmaceutical formulation, and the results were verified by the standard addition technique.

Evaluation of Quality Parameters of Apple Juices Using Near-Infrared Spectroscopy and Chemometrics

Near-infrared (NIR) spectra were recorded for commercial apple juices. Analysis of these spectra using partial least squares (PLS) regression revealed quantitative relations between the spectra and quality- and taste-related properties of juices: soluble solids content (SSC), titratable acidity (TA), and the ratio of soluble solids content to titratable acidity (SSC/TA). Various spectral preprocessing methods were used for model optimization. The optimal spectral variables were chosen using the jack-knife-based method and different variants of the interval PLS (iPLS) method. The models were cross-validated and evaluated based on the determination coefficients (R2), root-mean-square error of cross-validation (RMSECV), and relative error (RE). The best model for the prediction of SSC (R2 = 0.881, RMSECV = 0.277 °Brix, and RE = 2.37%) was obtained for the first-derivative preprocessed spectra and jack-knife variable selection. The optimal model for TA (R2 = 0.761, RMSECV = 0.239 g/L, and RE = 4.55%) was obtained for smoothed spectra in the range of 6224–5350 cm−1. The best model for the SSC/TA (R2 = 0.843, RMSECV = 0.113, and RE = 5.04%) was obtained for the spectra without preprocessing in the range of 6224–5350 cm−1. The present results show the potential of the NIR spectroscopy for screening the important quality parameters of apple juices.

Laser-Induced Breakdown Spectroscopy (LIBS) Measurement of Uranium in Molten Salt.

In this current study, the molten salt aerosol-laser-induced breakdown spectroscopy (LIBS) system was used to measure the uranium (U) content in a ternary UCl3-LiCl-KCl salt to investigate and assess a near real-time analytical approach for material safeguards and accountability. Experiments were conducted using five different U concentrations to determine the analytical figures of merit for the system with respect to U. In the analysis, three U lines were used to develop univariate calibration curves at the 367.01 nm, 385.96 nm, and 387.10 nm lines. The 367.01 nm line had the lowest limit of detection (LOD) of 0.065 wt% U. The 385.96 nm line had the best root mean square error of cross-validation (RMSECV) of 0.20 wt% U. In addition to the univariate calibration approach, a multivariate partial least squares (PLS) model was developed to further analyze the data. Using partial least squares (PLS) modeling, an RMSECV of 0.085 wt% U was determined. The RMSECV from the multivariate approach was significantly better than the univariate case and the PLS model is recommended for future LIBS analysis. Overall, the aerosol-LIBS system performed well in monitoring the U concentration and it is expected that the system could be used to quantitatively determine the U compositions within the normal operational concentrations of U in pyroprocessing molten salts.

Rapid and nondestructive measurement of glucose in a skin tissue phantom by near-infrared spectroscopy

Diabetes is one of the most serious metabolic diseases worldwide, and frequent monitoring of blood glucose is an essential part of diabetic management. However, a significant drawback of current monitoring methods was destructive and time-consuming. To meet this need, this study was to develop a method for rapid and noninvasive blood glucose assay in a skin tissue phantom by Near-Infrared spectroscopy (NIRS) and Raman spectroscopy. With partial least-squares (PLS) regression method, the multivariate calibration models of NIRS were generated and optimized individually by considering spectral region, spectral pretreatment methods and latent variables (LVs). The optimal NIR model was established with root mean square error of cross-validation (RMSECV) of 0.114, root mean square error of validation (RMSEP) of 0.061, correlation coefficient (R) of 0.9933, and residual predictive deviation (RPD) of 12.2, respectively. The validation results demonstrated that NIRS could be applied for rapid and noninvasive blood glucose assay.

Rapid authentication of Pseudostellaria heterophylla (Taizishen) from different regions by Raman spectroscopy coupled with chemometric methods

The quality of Pseudostellaria heterophylla depends on the growing area of the plants greatly. Compared with near infrared spectroscopy employed to discriminate the geographic regions before, the Raman signal is more obvious. So far, discrimination of P. heterophylla from different regions by Raman spectroscopy has not yet been realized. Hence, Raman spectroscopy coupled with chemometric methods to rapidly and effectively discriminate P. heterophylla from different regions was studied. Original spectra of P. heterophylla in wavenumber range of 4000–100 cm−1 were acquired. Then, a steady and exact model, partial least squares discriminant analysis (PLS-DA), was constructed. And competitive adaptive reweighted sampling (CARS) was further used to extract effective wavelength spectral characteristic variables. Results show that CARS-PLS-DA model is an appropriate model to discriminate the P. heterophylla, with determination coefficient of calibration (R2C), root mean square error of cross validation (RMSECV), determination coefficient of prediction (R2P), and root mean squared error of prediction (RMSEP) are 0.9468, 0.1979, 0.9665, and 0.1120, respectively. These results demonstrated that the built method is useful and effective to discriminate P. heterophylla from different regions, which provide a new idea in the application of rapid field test.

Quantitative analysis of glycated albumin in serum based on ATR-FTIR spectrum combined with SiPLS and SVM

A rapid quantitative analysis model for determining the glycated albumin (GA) content based on Attenuated total reflectance (ATR)-Fourier transform infrared spectroscopy (FTIR) combining with linear SiPLS and nonlinear SVM has been developed. Firstly, the real GA content in human serum was determined by GA enzymatic method, meanwhile, the ATR-FTIR spectra of serum samples from the population of health examination were obtained. The spectral data of the whole spectra mid-infrared region (4000–600 cm−1) and GA's characteristic region (1800–800 cm−1) were used as the research object of quantitative analysis. Secondly, several preprocessing steps including first derivative, second derivative, variable standardization and spectral normalization, were performed. Lastly, quantitative analysis regression models were established by using SiPLS and SVM respectively. The SiPLS modeling results are as follows: root mean square error of cross validation (RMSECVT) = 0.523 g/L, calibration coefficient (RC) = 0.937, Root Mean Square Error of Prediction (RMSEPT) = 0.787 g/L, and prediction coefficient (RP) = 0.938. The SVM modeling results are as follows: RMSECVT = 0.0048 g/L, RC = 0.998, RMSEPT = 0.442 g/L, and Rp = 0.916. The results indicated that the model performance was improved significantly after preprocessing and optimization of characteristic regions. While modeling performance of nonlinear SVM was considerably better than that of linear SiPLS. Hence, the quantitative analysis model for GA in human serum based on ATR-FTIR combined with SiPLS and SVM is effective. And it does not need sample preprocessing while being characterized by simple operations and high time efficiency, providing a rapid and accurate method for GA content determination.

PH value monitoring during human albumin purification with near infrared spectroscopy and chemometrics

Human albumin (HA) is a very important blood product which requires strict quality control strategy. Acid precipitation is a key step which has a great effect on the quality of final product. Therefore, a new method based on quality by design (QbD) was proposed to investigate the feasibility of realizing online quality control with the help of near infrared spectroscopy (NIRS) and chemometrics. The pH value is the critical process parameter (CPP) in acid precipitation process, which is used as the end-point indicator. Six batches, a total of 74 samples of acid precipitation process, were simulated in our lab. Four batches were selected randomly as calibration set and remaining two batches as validation set. Then, the analysis based on material information and three different variable selection methods, including interval partial least squares regression (iPLS), competitive adaptive reweighted sampling (CARS) and correlation coefficient (CC) were compared for eliminating irrelevant variables. Finally, iPLS was used for variables selection. The quantitative model was built up by partial least squares regression (PLSR). The values of determination coefficients ([Formula: see text] and [Formula: see text]), root mean squares error of prediction (RMSEP), root mean squares error of calibration (RMSEC) and root mean squared error of cross validation (RMSECV) were 0.969, 0.953, 0.0496, 0.0695 and 0.0826, respectively. The paired [Formula: see text] test and repeatability test showed that the model had good prediction ability and stability. The results indicated that PLSR model could give accurate measurement of the pH value.