Classification and Class-Modelling

What the fish? Tracing the geographical origin of fish using NIR spectroscopy

Food authentication is a growing concern with rising complexities of the food supply network, with fish being an easy target of food fraud. In this regard, NIR spectroscopy has been used as an efficient tool for food authentication. This article reviews the latest research advances on NIR based fish authentication. The process from sampling/sample preparation to data analysis has been covered. Special attention was given to NIR spectra pre-processing and its unsupervised and supervised analysis. Sampling is an important aspect of traceability study and samples chosen ought to be a true representative of the population. NIR spectra acquired is often laden with overlapping bands, scattering and highly multicollinear. It needs adequate pre-processing to remove all undesirable features. The pre-processing technique can make or break a model and thus need a trial-and-error approach to find the best fit. As for spectral analysis and modelling, multicollinear nature of NIR spectra demands unsupervised analysis (PCA) to compact the features before application of supervised multivariate techniques such as LDA, PLS-DA, QDA etc. Machine learning approach of modelling has shown promising result in food authentication modelling and negates the need for unsupervised analysis before modelling.

An evaluation of IR spectroscopy for authentication of adulterated turmeric powder using pattern recognition

Turmeric powder is a widely consumed spice, making it an attractive target for adulteration, which is not easily detected. The study examined the simultaneous use of IR spectroscopy in combination with controlled (PCA) and uncontrolled (PLS-DA and CMCA) pattern recognition techniques to detect and classify Sudan Red, starch and metanil yellow fraud in turmeric powder nondestructively. The results showed that the two major peaks in turmeric powder at 1625 cm^-1 and 1600 cm^-1 are not present in Sudan Red, starch and metanil yellow because these materials lack this functional group. Data distribution at the two PC locations showed clearly scattered clusters according to the four mixing studied models (turmeric powder, turmeric powder-Sudan Red mixture, turmeric powder-starch mixture and turmeric powder-metanil yellow mixture), but there was a clear overlap between turmeric powder and turmeric powder - Sudan red mixture. Both PLS-DA and SIMCA supervised methods showed satisfactory discrimination. The results also showed that in all the sample groups, when the samples were classified by PLS-DA, the values were higher compared to the SIMCA model. The overall precision of the SIMCA and PLS-DA classifier were 82% and 92%, respectively. However, when considering only two main categories adulterated (the samples at the groups 2, 3 and 4) and pure (the samples at the group 1), an acceptable degree of separation between the resulting classes was obtained. Consequently, IR spectroscopy with pattern recognition methods was found to be a promising tool for nondestructive grouping of turmeric powder samples with different types of adulteration in turmeric powder.

Critical evaluation of ambient mass spectrometry coupled with chemometrics for the early detection of adulteration scenarios in Origanum vulgare L

Nowadays, most of the screening methods in food manufacturing are based on spectroscopic techniques. Ambient Mass Spectrometry is a relatively new field of analytical chemistry which has proven to offer similar speed and ease-of-use when compared to other fingerprinting techniques, alongside the advantages of good selectivity, sensitivity and chemical information. Numerous applications have been explored in food authenticity, based either on the target detection of adulteration markers or, less frequently, on the development of multivariate classification models. The aim of the present work was to evaluate and compare the capabilities of Direct Analysis in Real Time (DART) and Atmospheric Solid Analysis Probe (ASAP) Mass Spectrometry (MS) for the high-throughput authenticity screening of commercial herbs and spices products. The gross addition of bulking material to dried Mediterranean oregano was taken as case study. First, a pilot sample set, constituted by authentic dried oregano, olive leaves (a frequently reported adulterant) and mixtures thereof at different levels (i.e. 10, 20, 30 and 50% w/w) was used. Each sample was fingerprinted by both ambient-MS techniques. After appropriate pre-processing, the whole mass spectra were used for the subsequent multivariate data analysis. Soft Independent Modelling of Class Analogy was adopted as classification algorithm and the model was challenged with both new authentic oregano and in-house prepared blends. To the best of our knowledge, this is the first report of DART-MS and ASAP-MS used in full scan mode and coupled to chemometric modelling as rapid fingerprinting approach for food authentication. Although both the techniques provided satisfactory results, ASAP-MS clearly showed greater potential, leading to reproducible, diagnostic feature-rich mass spectra. For this reason, ASAP-MS was further tested under a more convoluted scenario, where the training and validation sets were enlarged with additional authentic oregano samples and a wider range of adulterant species, respectively. Overall good results were achieved, with 93% model predictive accuracy, and screening detection capability estimated between 5-20% (w/w) addition, depending on the adulterant considered with the only exception of majorana. Investigation of Q residuals could highlight the statistically-relevant chemical markers which could be tentatively annotated by coupling the ASAP probe with a high resolution mass analyser. The results from the validation study confirmed the great potential of ASAP-MS in combination with chemometrics as fast MS-based screening solution and demonstrated its feasibility for classification model building.

The main effects of elevated CO2 and soil-water deficiency on 1H NMR-based metabolic fingerprints of Coffea arabica beans by factorial and mixture design

The metabolic response of Coffea arabica trees in the face of the rising atmospheric concentration of carbon dioxide (CO₂) combined with the reduction in soil-water availability is complex due to the various (bio)chemical feedbacks. Modern analytical tools and the experimental advance of agronomic science tend to advance in the understanding of the metabolic complexity of plants. In this work, Coffea arabica trees were grown in a Free-Air Carbon Dioxide Enrichment dispositive under factorial design (2²) conditions considering two CO₂ levels and two soil-water availabilities. The ¹H NMR mixture design-fingerprinting effects of CO₂ and soil-water levels on beans were strategically investigated using the principal component analysis (PCA), analysis of variance (ANOVA) - simultaneous component analysis (ASCA) and partial least squares-discriminant analysis (PLS-DA). From the ASCA, the CO₂ factor had a significant effect on changing the ¹H NMR profile of fingerprints. The soil-water factor and interaction (CO₂ × soil-water) were not significant. ¹H NMR fingerprints with PCA, ASCA and PLS-DA analysis determined spectral profiles for fatty acids, caffeine, trigonelline and glucose increases in beans from current CO₂, while quinic acid/chlorogenic acids, malic acid and kahweol/cafestol increased in coffee beans from elevated CO₂. PLS-DA results revealed a good classification performance between the significant effect of the atmospheric CO₂ levels on the fingerprints, regardless of the soil-water availabilities. Finally, the PLS-DA model showed good prediction ability, successfully classifying validation data-set of coffee beans collected over the vertical profile of the plants and included several fingerprints of different extracting solvents. The results of this investigation suggest that the association of experimental design, mixture design, PCA, ASCA and PLS-DA can provide accurate information on a series of metabolic changes provoked by climate changes in products of commercial importance, in addition to minimizing the extra work necessary in classic analytical approaches, encouraging the development of similar strategies.

Modeling CO2 exchange and meteorological factors of an apple orchard using partial least square regression

The eddy covariance (EC) technique was used to measure variations of orchard-atmosphere CO₂ exchange, as a function of meteorological variables in an apple orchard in 2016-2017. The annual average CO₂ exchange rate was 2.295 kg m^-2. Excavations and biomass assessments demonstrated that the orchard stored close to 20.6 tC ha^-1 as plant C over a 15-year period. Seasonally, high rates of CO₂ uptake and low CO₂ emissions occurred between May and August and December and March, respectively. The maximum rates of monthly CO₂ exchange were 144.44 and 153.61 gC m^-2 month^-1 in August 2016 and June 2017, respectively. Partial least squares (PLS) regressions were used to analyze the influence of meteorological factors to on CO₂ exchange rates. Temperature and photosynthetic active radiation (PAR) were observed to exert the largest influence on driving variation in CO₂ exchange. The main meteorological factors affecting CO₂ exchange on daily and monthly time scales were soil temperature (T_soil), air temperature (T_a), PAR, below canopy CO₂ concentration (BCC), vapor pressure deficit (VPD), and soil water content at 50 cm (SWC_50cm). The regression model equation describing CO₂ exchange included T_a, VPD, precipitation (PPT), and sunshine duration (SD), as significant variables. This model curve fitting explains over 80% of the variation in CO₂ exchange. This study provides CO₂ exchange characteristics and a model equation capable of predicting CO₂ exchange of an apple orchard. Graphical Abstract.

Rapid detection and quantification of sucrose adulteration in cow milk using Attenuated total reflectance-Fourier transform infrared spectroscopy coupled with multivariate analysis

Adulteration of milk to gain economic benefit has become a common practice in recent years. Sucrose is illegally added in milk to reconstitute its compositional requirement by improving the total solid contents. The present study is aimed to use FTIR spectroscopy in combination with multivariate chemometric modelling for the differentiation and quantification of sucrose in cow milk. Pure milk and adulterated milk spectra (0.5-7.5% w/v) were observed in the spectral region 4000-400 cm^-1. Principal component analysis (PCA) was used for the discrimination of pure milk and adulterated milk. Soft independent modelling of class analogy (SIMCA) was able to classify test samples with a classification efficiency of 100%. Partial least square regression (PLS-R) and principle component regression (PCR) models were established for normal spectra, 1st derivative and 2nd derivative for the quantification of sucrose in milk. PLS-R model (normal spectra) in the combined wavenumber range of 1070-980 cm^-1 showed the best prediction based on parameters like coefficient of determination (R²) (Cal: 0.996; Val: 0.993), RMSE (Cal: 0.15% w/v; Val: 0.20% w/v), RE% (Cal: 4.9% w/v; Val: 5.1% w/v) and RPD (13.40). This method has a detection level of 0.5% w/v sucrose adulteration.

A metabolomic data fusion approach to support gliomas grading

Magnetic resonance imaging (MRI) is the current gold standard for the diagnosis of brain tumors. However, despite the development of MRI techniques, the differential diagnosis of central nervous system (CNS) primary pathologies, such as lymphoma and glioblastoma or tumor-like brain lesions and glioma, is often challenging. MRI can be supported by in vivo magnetic resonance spectroscopy (MRS) to enhance its diagnostic power and multiproject-multicenter evaluations of classification of brain tumors have shown that an accuracy around 90% can be achieved for most of the pairwise discrimination problems. However, the survival rate for patients affected by gliomas is still low. The High-Resolution Magic-Angle-Spinning Nuclear Magnetic Resonance (HR-MAS NMR) metabolomics studies may be helpful for the discrimination of gliomas grades and the development of new strategies for clinical intervention. Here, we propose to use T₂ -filtered, diffusion-filtered and conventional water-presaturated spectra to try to extract as much information as possible, fusing the data gathered by these different NMR experiments and applying a chemometric approach based on Multivariate Curve Resolution (MCR). Biomarkers important for glioma's discrimination were found. In particular, we focused our attention on cystathionine (Cyst) that shows promise as a biomarker for the better prognosis of glioma tumors.

Reconstitution followed by non-targeted mid-infrared analysis as a workable and cost-effective solution to overcome the blending duality in milk powder adulteration detection

Mid-infrared analysis of reconstituted milk is proposed as a feasible solution for the detection of milk powder adulteration regardless of the blending practice. To challenge the concept, skim milk powders were spiked with three of the most reactive/unstable of potential milk adulterants: semicarbazide hydrochloride, ammonium sulfate and cornstarch. To create the wet-blended set, a fraction of each sample was reconstituted and re-spray dried at laboratory scale with a benchtop spray dryer. Dry and wet-blended adulterated samples were reconstituted prior to mid-infrared measurement and projected onto a one-class classifier SIMCA model for reconstituted skim milk. Quantitative sensitivities, determined from the normalized orthogonal distances, were compared. Although the non-industrial spray drying introduced a spectroscopic bias, as revealed by the control samples, the non-targeted mid-infrared model showed comparable sensitivities for both blending practices once the main bias-rich spectral regions were removed, validating thereby the proposed concept.

Potential biomonitoring of atmospheric carbon dioxide in Coffea arabica leaves using near-infrared spectroscopy and partial least squares discriminant analysis

The potencial of Coffea arabica leaves as bioindicators of atmospheric carbon dioxide (CO₂) was evaluated in a free-air carbon dioxide enrichment (FACE) experiment by using near-infrared reflectance (NIR) spectroscopy for direct analysis and partial least squares discriminant analysis (PLS-DA). A supervised classification model was built and validated from the spectra of coffee leaves grown under elevated and current CO₂ levels. PLS-DA allowed correct test set classification of 92% of the elevated-CO₂ level leaves and 100% of the current-CO₂ level leaves. The spectral bands accounting for the discrimination of the elevated-CO₂ leaves were at 1657 and 1698 nm, as indicated by the variable importance in the projection (VIP) score together with the regression coefficients. Seven months after suspension of enriched CO₂, returning to current-CO₂ levels, new spectral measurements were made and subjected to PLS-DA analysis. The predictive model correctly classified all leaves as grown under current-CO₂ levels. The fingerprints suggest that after suspension of elevated-CO₂, the spectral changes observed previously disappeared. The recovery could be triggered by two reasons: the relief of the stress stimulus or the perception of a return of favorable conditions. In addition, the results demonstrate that NIR spectroscopy can provide a rapid, nondestructive, and environmentally friendly method for biomonitoring leaves suffering environmental modification. Finally, C. arabica leaves associated with NIR and mathematical models have the potential to become a good biomonitoring system.

Raw material quality assessment approaches comparison in pectin production

Different opportunities are explored to evaluate quality variation in raw materials from biological origin. Assessment of raw materials attributes is an important step in a bio-based production as fluctuations in quality are a major source of process disturbance. This can be due to a variety of biological, seasonal, and supply scarcity reasons. The final properties of a product are invariably linked with the initial properties of the raw material. Thus, the operational conditions of a process can be tuned to drive the product to the required specification based on the quality assessment of the raw material being processed. Process analytical technology tools which enable this assessment in a far more informative and rapid manner than current industrial practices that rely on rule-of-thumb decisions are assessed. An example with citrus peels is used to demonstrate the conceptual and performance differences of distinct quality assessment approaches. The analysis demonstrates the advantage of characterization through multivariate data analysis coupled with a complementary spectroscopic technique, near-infrared spectroscopy. The quantitative comparative analysis of three different approaches, discriminant classification based on expert-knowledge, unsupervised classification, and spectroscopic correlation with reference physicochemical variables, is performed in the same dataset context. © 2018 Her Majesty the Queen in Right of Canada © 2018 American Institute of Chemical Engineers Biotechnol. Prog., 35: e2762, 2019.

Chemometric Methods for Spectroscopy-Based Pharmaceutical Analysis

Spectroscopy is widely used to characterize pharmaceutical products or processes, especially due to its desirable characteristics of being rapid, cheap, non-invasive/non-destructive and applicable both off-line and in-/at-/on-line. Spectroscopic techniques produce profiles containing a high amount of information, which can profitably be exploited through the use of multivariate mathematic and statistic (chemometric) techniques. The present paper aims at providing a brief overview of the different chemometric approaches applicable in the context of spectroscopy-based pharmaceutical analysis, discussing both the unsupervised exploration of the collected data and the possibility of building predictive models for both quantitative (calibration) and qualitative (classification) responses.

Physicochemical characterization and antioxidant activity of Palestinian honey samples

Physicochemical characteristics, main minerals, and antioxidant activity were determined for Palestinian honey samples belonging to different floral and geographical origins. The features of the analyzed samples were within the established international standards for honey quality control. One clear exception was the hydroxymethylfurfural (HMF) of the Ziziphus sample purchased from the Jericho region, which is the lowest city in the word characterized by a hot desert climate. The observed HMF value was 81.86 ± 2.64 mg/kg being two folds the maximum allowed in honey samples (40 mg/kg). As a second objective of the present work, the parameters were divided into two groups with different discriminatory power. The assessed physicochemical parameters, and the antioxidant activities, specific to the botanical origin discrimination, were used to run the first PCA. A strong correlation could be seen between the bioactive compounds and the antioxidant activities despite the geographical origin of the samples. Thyme and Ziziphus samples were the best samples, while citrus sample presented the lowest activity. Regarding the geographical discrimination, Ash and mineral contents in addition to the electrical conductivity were used. The output PCA conserved high represent ability of the data in the two-first components being 82.72% and 9.60%. A little discrimination between the samples produced in the north and those produced in the south of the country, but it was not perfect. The intervention of the botanical variability could be the reason.

Multivariate approaches for stability control of the olive oil reference materials for sensory analysis - part I: framework and fundamentals

BACKGROUND

Virgin olive oil is the only food product for which sensory analysis is regulated to classify it in different quality categories. To harmonize the results of the sensorial method, the use of standards or reference materials is crucial. The stability of sensory reference materials is required to enable their suitable control, aiming to confirm that their specific target values are maintained on an ongoing basis. Currently, such stability is monitored by means of sensory analysis and the sensory panels are in the paradoxical situation of controlling the standards that are devoted to controlling the panels.

RESULTS

In the present study, several approaches based on similarity analysis are exploited. For each approach, the specific methodology to build a proper multivariate control chart to monitor the stability of the sensory properties is explained and discussed.

CONCLUSION

The normalized Euclidean and Mahalanobis distances, the so-called nearness and hardiness indices respectively, have been defined as new similarity indices to range the values from 0 to 1. Also, the squared mean from Hotelling's T² -statistic and Q² -statistic has been proposed as another similarity index. © 2018 Society of Chemical Industry.

Biomarkers for physical frailty and sarcopenia

Physical frailty (PF) and sarcopenia are major health issues in geriatric populations, given their high prevalence and association with several adverse outcomes. Nevertheless, the lack of an univocal operational definition for the two conditions has so far hampered their clinical implementation. Existing definitional ambiguities of PF and sarcopenia, together with their complex underlying pathophysiology, also account for the absence of robust biomarkers that can be used for screening, diagnostic and/or prognostication purposes. This review provides an overview of currently available biological markers for PF and sarcopenia, as well as a critical appraisal of strengths and weaknesses of traditional procedures for biomarker development in the field. A novel approach for biomarker identification and validation, based on multivariate methodologies, is also discussed. This strategy relies on the multidimensional modeling of complementary biomarkers to cope with the phenotypical and pathophysiological complexity of PF and sarcopenia. Biomarkers identified through the implementation of multivariate strategies may be used to support the detection of the two conditions, track their progression over time or in response to interventions, and reveal the onset of complications (e.g., mobility disability) at a very early stage.

Biomarkers for physical frailty and sarcopenia: state of the science and future developments

Physical frailty and sarcopenia are two common and largely overlapping geriatric conditions upstream of the disabling cascade. The lack of a unique operational definition for physical frailty and sarcopenia and the complex underlying pathophysiology make the development of biomarkers for these conditions extremely challenging. Indeed, the current definitional ambiguities of physical frailty and sarcopenia, together with their heterogeneous clinical manifestations, impact the accuracy, specificity, and sensitivity of individual biomarkers proposed so far. In this review, the current state of the art in the development of biomarkers for physical frailty and sarcopenia is presented. A novel approach for biomarker identification and validation is also introduced that moves from the 'one fits all' paradigm to a multivariate methodology.