Optimisation of the predictive ability of NIR models to estimate nutritional parameters in elephant grass through LOCAL algorithms

Elephant grass is a tropical forage widely used for livestock feed. The analytical techniques traditionally used for its nutritional evaluation are costly and time consuming. Alternatively, Near Infrared Spectroscopy (NIRS) technology has been used as a rapid analysis technique. However, in crops with high variability due to genetic improvement, predictive models quickly lose accuracy and must be recalibrated. The use of non-linear models such as LOCAL calibrations could mitigate these issues, although a number of parameters need to be optimized to obtain accurate results. The objective of this work was to compare the predictive results obtained with global NIRS calibrations and with LOCAL calibrations, paying special attention to the configuration parameters of the models. The results obtained showed that the prediction errors with the LOCAL models were between 1.6 and 17.5 % lower. The best results were obtained in most cases with a low number of selected samples (n = 100-250) and a high number of PLS terms (n = 20). This configuration allows a reduced computation time with high accuracy, becoming a valuable alternative for analytical determinations that require ruminal fluid, which would improve the welfare of the animals by avoiding the need to surgically prepare animals to estimate the nutritional value of the feeds.

Machine-Learning Approach for Predicting the Discharging Capacities of Doped Lithium Nickel-Cobalt-Manganese Cathode Materials in Li-Ion Batteries

Understanding the governing dopant feature for cyclic discharge capacity is vital for the design and discovery of new doped lithium nickel-cobalt-manganese (NCM) oxide cathodes for lithium-ion battery applications. We herein apply six machine-learning regression algorithms to study the correlations of the structural, elemental features of 168 distinct doped NCM systems with their respective initial discharge capacity (IC) and 50th cycle discharge capacity (EC). First, a Pearson correlation coefficient study suggests that the lithium content ratio is highly correlated to both discharge capacity variables. Among all six regression algorithms, gradient boosting models have demonstrated the best prediction power for both IC and EC, with the root-mean-square errors calculated to be 16.66 mAhg-1 and 18.59 mAhg-1, respectively, against a hold-out test set. Furthermore, a game-theory-based variable-importance analysis reveals that doped NCM materials with higher lithium content, smaller dopant content, and lower-electronegativity atoms as the dopant are more likely to possess higher IC and EC. This study has demonstrated the exciting potentials of applying cutting-edge machine-learning techniques to accurately capture the complex structure-property relationship of doped NCM systems, and the models can be used as fast screening tools for new doped NCM structures with more superior electrochemical discharging properties.

Probabilistic classification models for the in situ authentication of iberian pig carcasses using near infrared spectroscopy

Iberian pig ham is one of several high value European food products that are the subject of significant attempts at fraud because of the high price differences between commercial categories. Iberian pig products are classified by the Spanish regulations into different categories, mainly depending on the feeding regime during the fattening phase and the race involved, being of Premium quality those products obtained from the animals fed with acorns and other natural resources. Most of the previous NIRS studies related to the Iberian pig have involved the use of at-line instruments to predict quantitative quality parameters. This paper explores the use of the NIR spectra (369 for training and 199 for validation) to classify samples according to the categories Premium (animals fed with acorn) and Non Premium (animals fed with compound feeds), using a MicroNIR™ Pro1700 microspectrometer to analyse individual carcasses in situ at the slaughterhouse line. Four discriminant methods were explored: linear discriminant analysis (LDA), quadratic discriminant analysis (QDA), Kernel Bayes and Logistic Regression. These are all discriminant methods that naturally produce classification probabilities to quantify the uncertainty of the results. Rules were tuned and methods compared using both classification error rates and a probability scoring rule. LDA gave the best results, attaining an overall accuracy of 93% and providing well-calibrated classification probabilities.

Parsimonious test of dynamic interaction

In recent years, there have been significant advances in the technology used to collect data on the movement and activity patterns of humans and animals. GPS units, which form the primary source of location data, have become cheaper, more accurate, lighter and less power-hungry, and their accuracy has been further improved with the addition of inertial measurement units. The consequence is a glut of geospatial time series data, recorded at rates that range from one position fix every several hours (to maximize system lifetime) to ten fixes per second (in high dynamic situations). Since data of this quality and volume have only recently become available, the analytical methods to extract behavioral information from raw position data are at an early stage of development. An instance of this lies in the analysis of animal movement patterns. When investigating solitary animals, the timing and location of instances of avoidance and association are important behavioral markers. In this paper, a novel analytical method to detect avoidance and association between individuals is proposed; unlike existing methods, assumptions about the shape of the territories or the nature of individual movement are not needed. Simulations demonstrate that false positives (type I error) are rare (1%-3%), which means that the test rarely suggests that there is an association if there is none.

Elastic scattering spectroscopy for early detection of breast cancer: partially supervised Bayesian image classification of scanned sentinel lymph nodes

Sentinel lymph node biopsy is a standard diagnosis procedure to determine whether breast cancer has spread to the lymph glands in the armpit (the axillary nodes). The metastatic status of the sentinel node (the first node in the axillary chain that drains the affected breast) is the determining factor in surgery between conservative lumpectomy and more radical mastectomy including axillary node excision. The traditional assessment of the node requires sample preparation and pathologist interpretation. An automated elastic scattering spectroscopy (ESS) scanning device was constructed to take measurements from the entire cut surface of the excised sentinel node and to produce ESS images for cancer diagnosis. Here, we report on a partially supervised image classification scheme employing a Bayesian multivariate, finite mixture model with a Markov random field (MRF) spatial prior. A reduced dimensional space was applied to represent the scanning data of the node by a statistical image, in which normal, metastatic, and nonnodal-tissue pixels are identified. Our results show that our model enables rapid imaging of lymph nodes. It can be used to recognize nonnodal areas automatically at the same time as diagnosing sentinel node metastases with sensitivity and specificity of 85% and 94%, respectively. ESS images can help surgeons by providing a reliable and rapid intraoperative determination of sentinel nodal metastases in breast cancer.

Classifying Degraded Modern Polymeric Museum Artefacts by Their Smell

The use of VOC analysis to diagnose degradation in modern polymeric museum artefacts is reported. Volatile organic compound (VOC) analysis is a successful method for diagnosing medical conditions but to date has found little application in museums. Modern polymers are increasingly found in museum collections but pose serious conservation difficulties owing to unstable and widely varying formulations. Solid‐phase microextraction gas chromatography/mass spectrometry and linear discriminant analysis were used to classify samples according to the length of time they had been artificially degraded. Accuracies in classification of 50–83 % were obtained after validation with separate test sets. The method was applied to three artefacts from collections at Tate to detect evidence of degradation. This approach could be used for any material in heritage collections and more widely in the field of polymer degradation.

Parent-dependent stressors and the onset of anxiety disorders in children: links with parental psychopathology

Exposure to stressors is associated with an increased risk for child anxiety. Investigating the family origins of stressors may provide promising avenues for identifying and intervening with children at risk for the onset of anxiety disorders and their families. The aim of this study was to compare the frequency of parent-dependent negative life events and chronic adversities experienced by children with an anxiety disorder (n = 34) in the 12 months prior to the onset of the child's most recent episode, compared to healthy controls (n = 34). Life events and chronic adversities were assessed using maternal report during an investigator-based interview, which provided independent panel ratings of the extent that reported experiences were related to parent behaviour. There were no group differences in the number of parent-dependent negative life events for anxious children compared to controls. However, significantly more parent-dependent chronic adversities were present for anxious children compared to controls. Findings suggest that parents contribute to an increased frequency of chronic adversities but not negative life events prior to their child's most recent onset of anxiety. Furthermore, increased child exposure to parent-dependent chronic adversities was related to parental history of mental disorder.

Near-Infrared Spectroscopy and Geostatistical Analysis for Modeling Spatial Distribution of Analytical Constituents in Bulk Animal By-Product Protein Meals

Control and inspection operations within the context of safety and quality assessment of bulk foods and feeds are not only of particular importance, they are also demanding challenges, given the complexity of food/feed production systems and the variability of product properties. Existing methodologies have a variety of limitations, such as high costs of implementation per sample or shortcomings in early detection of potential threats for human/animal health or quality deviations. Therefore, new proposals are required for the analysis of raw materials in situ in a more efficient and cost-effective manner. For this purpose, a pilot laboratory study was performed on a set of bulk lots of animal by-product protein meals to introduce and test an approach based on near-infrared (NIR) spectroscopy and geostatistical analysis. Spectral data, provided by a fiber optic probe connected to a Fourier transform (FT) NIR spectrometer, were used to predict moisture and crude protein content at each sampling point. Variographic analysis was carried out for spatial structure characterization, while ordinary Kriging achieved continuous maps for those parameters. The results indicated that the methodology could be a first approximation to an approach that, properly complemented with the Theory of Sampling and supported by experimental validation in real-life conditions, would enhance efficiency and the decision-making process regarding safety and adulteration issues.

Evaluation of local approaches to obtain accurate near-infrared (NIR) equations for prediction of ingredient composition of compound feeds

This research work investigated new methods to improve the accuracy of intact feed calibrations for the near-infrared (NIR) prediction of the ingredient composition. When NIR reflection spectroscopy, together with linear models, was used for the prediction of the ingredient composition, the results were not always acceptable. Therefore, other methods have been investigated. Three different local methods (comparison analysis using restructured near-infrared and constituent data [CARNAC]), locally weighed regression [LWR], and LOCAL) were applied to a large (N = 20 320) and heterogeneous population of non-milled feed compounds for the NIR prediction of the inclusion percentage of wheat and sunflower meal, as representative of two different classes of ingredients. Compared with partial least-squares regression, results showed considerable reductions of standard error of prediction values for all methods and ingredients: reductions of 59, 47, and 50% with CARNAC, LWR, and LOCAL, respectively, for wheat, and reductions of 49, 45, and 43% with CARNAC, LWR, and LOCAL, respectively, for sunflower meal. These results are a valuable achievement in coping with legislation and manufacture requirements concerning the labeling of intact feedstuffs.

Scanning elastic scattering spectroscopy detects metastatic breast cancer in sentinel lymph nodes

A novel method for rapidly detecting metastatic breast cancer within excised sentinel lymph node(s) of the axilla is presented. Elastic scattering spectroscopy (ESS) is a point-contact technique that collects broadband optical spectra sensitive to absorption and scattering within the tissue. A statistical discrimination algorithm was generated from a training set of nearly 3000 clinical spectra and used to test clinical spectra collected from an independent set of nodes. Freshly excised nodes were bivalved and mounted under a fiber-optic plate. Stepper motors raster-scanned a fiber-optic probe over the plate to interrogate the node's cut surface, creating a 20x20 grid of spectra. These spectra were analyzed to create a map of cancer risk across the node surface. Rules were developed to convert these maps to a prediction for the presence of cancer in the node. Using these analyses, a leave-one-out cross-validation to optimize discrimination parameters on 128 scanned nodes gave a sensitivity of 69% for detection of clinically relevant metastases (71% for macrometastases) and a specificity of 96%, comparable to literature results for touch imprint cytology, a standard technique for intraoperative diagnosis. ESS has the advantage of not requiring a pathologist to review the tissue sample.

Improving the discriminatory power of a near-infrared microscopy spectral library with a support vector machine classifier

A multi-group classifier based on the support vector machine (SVM) has been developed for use with a library of 48,456 spectra measured by near-infrared reflection microscopy (NIRM) on 227 samples representing 26 animal feed ingredients and 4 possible contaminants of animal origin. The performance of the classifier was assessed by a five-fold cross-validation, dividing at the sample level. Although the overall proportion of misclassifications was 27%, almost all of these involved the confusion of pairs of similar ingredients of vegetable origin. Such confusions are unimportant in the context of the intended use of the library, which is the detection of banned ingredients in animal feed. The error rate in discrimination between permitted and banned ingredients was just 0.17%. The performance of the SVM classifier was substantially better than that of the K-nearest-neighbors method employed in previous work with the same library, for which the comparable error rates are 36% overall and 0.39% for permitted versus banned ingredients.

Spectral library validation to identify ingredients of compound feedingstuffs by near infrared reflectance microscopy

To guarantee feed quality and safety the development and improvement of analytical methods for feed authentication and detection of contaminants is fundamental. Near infrared reflectance microscopy (NIRM) has been investigated as an alternative method to contribute to control systems for feed materials. The major task is the need to build NIRM reference spectral libraries that must represent the variability in feed ingredients. The aim of the present work was to evaluate the performance of a NIRM reference spectral library on animal feed, with external samples of animal feed ingredients and possible contaminants such as processed animal proteins, and in particular to assess its ability to identify ingredients in mixtures. Three external sample sets were used: (A) artificial mixtures, (B) synthetic mixtures and (C) synthetic binary mixtures. The prediction and repeatability results for set A, in which the spectra are from pure ingredients, were very good for both animal and vegetable ingredients and confirm that the spectral library is very good at identifying spectra from pure ingredients. For sets B and C, in which the spectra were measured on mixtures, the prediction results were very disappointing compared with the artificial samples. This means that a strategy that tries to match the spectra taken from a mixture with those of pure ingredients is unlikely to meet with much success. It is possible that an interpolation between pure ingredients for suitably chosen spectral ranges may provide a way to extend this system to mixtures, including mixtures of several ingredients.

Elastic scattering spectroscopy for detection of cancer risk in Barrett's esophagus: experimental and clinical validation of error removal by orthogonal subtraction for increasing accuracy

Elastic scattering spectroscopy (ESS) may be used to detect high-grade dysplasia (HGD) or cancer in Barrett's esophagus (BE). When spectra are measured in vivo by a hand-held optical probe, variability among replicated spectra from the same site can hinder the development of a diagnostic model for cancer risk. An experiment was carried out on excised tissue to investigate how two potential sources of this variability, pressure and angle, influence spectral variability, and the results were compared with the variations observed in spectra collected in vivo from patients with Barrett's esophagus. A statistical method called error removal by orthogonal subtraction (EROS) was applied to model and remove this measurement variability, which accounted for 96.6% of the variation in the spectra, from the in vivo data. Its removal allowed the construction of a diagnostic model with specificity improved from 67% to 82% (with sensitivity fixed at 90%). The improvement was maintained in predictions on an independent in vivo data set. EROS works well as an effective pretreatment for Barrett's in vivo data by identifying measurement variability and ameliorating its effect. The procedure reduces the complexity and increases the accuracy and interpretability of the model for classification and detection of cancer risk in Barrett's esophagus.

Taking NIR calibrations of feed compounds from the laboratory to the process: calibration transfer between predispersive and postdispersive instruments

In the context of current demands in the animal feed industry for controls and analyses, the use of instruments that may be applied on the process line has acquired a significant interest. A key aspect is that the calibrations developed for quality control with instruments sited in the laboratory (at-line) must be transferred to instruments that will be used in the plant itself (online). This study evaluates the standardization and the calibration transfer between a grating monochromator instrument (predispersive) designed for laboratory analysis and a diode array instrument (postdispersive) more adapted to process conditions. Two procedures that correct differences between spectra of two instruments were tested: the patented algorithm by Shenk and Westerhaus and piecewise direct standardization (PDS). Although results were slightly better with PDS, both methods achieved good spectral matching between the two instruments, with levels of repeatability similar to that of the grating instrument itself. The calibration transfer was evaluated in terms of the standard error of prediction (SEP), which was considerably reduced after standardization. However, final calibration models to be used in the diode array instrument must contain spectra from both types of instruments to give acceptable prediction accuracy.

Measurement error in the explanatory variable of a binary regression: regression calibration and integrated conditional likelihood in studies of residential radon and lung cancer

In epidemiology, one approach to investigating the dependence of disease risk on an explanatory variable in the presence of several confounding variables is by fitting a binary regression using a conditional likelihood, thus eliminating the nuisance parameters. When the explanatory variable is measured with error, the estimated regression coefficient is biased usually towards zero. Motivated by the need to correct for this bias in analyses that combine data from a number of case-control studies of lung cancer risk associated with exposure to residential radon, two approaches are investigated. Both employ the conditional distribution of the true explanatory variable given the measured one. The method of regression calibration uses the expected value of the true given measured variable as the covariate. The second approach integrates the conditional likelihood numerically by sampling from the distribution of the true given measured explanatory variable. The two approaches give very similar point estimates and confidence intervals not only for the motivating example but also for an artificial data set with known properties. These results and some further simulations that demonstrate correct coverage for the confidence intervals suggest that for studies of residential radon and lung cancer the regression calibration approach will perform very well, so that nothing more sophisticated is needed to correct for measurement error.