The use of Fourier-transform infrared spectroscopy to predict cheese yield and nutrient recovery or whey loss traits from unprocessed bovine milk samples

Cheese yield is an important technological trait in the dairy industry in many countries. The aim of this study was to evaluate the effectiveness of Fourier-transform infrared (FTIR) spectral analysis of fresh unprocessed milk samples for predicting cheese yield and nutrient recovery traits. A total of 1,264 model cheeses were obtained from 1,500-mL milk samples collected from individual Brown Swiss cows. Individual measurements of 7 new cheese yield-related traits were obtained from the laboratory cheese-making procedure, including the fresh cheese yield, total solid cheese yield, and the water retained in curd, all as a percentage of the processed milk, and nutrient recovery (fat, protein, total solids, and energy) in the curd as a percentage of the same nutrient contained in the milk. All individual milk samples were analyzed using a MilkoScan FT6000 over the spectral range from 5,000 to 900 wavenumber × cm⁻¹. Two spectral acquisitions were carried out for each sample and the results were averaged before data analysis. Different chemometric models were fitted and compared with the aim of improving the accuracy of the calibration equations for predicting these traits. The most accurate predictions were obtained for total solid cheese yield and fresh cheese yield, which exhibited coefficients of determination between the predicted and measured values in cross-validation (1-VR) of 0.95 and 0.83, respectively. A less favorable result was obtained for water retained in curd (1-VR = 0.65). Promising results were obtained for recovered protein (1-VR = 0.81), total solids (1-VR = 0.86), and energy (1-VR = 0.76), whereas recovered fat exhibited a low accuracy (1-VR = 0.41). As FTIR spectroscopy is a rapid, cheap, high-throughput technique that is already used to collect standard milk recording data, these FTIR calibrations for cheese yield and nutrient recovery highlight additional potential applications of the technique in the dairy industry, especially for monitoring cheese-making processes and milk payment systems. In addition, the prediction models can be used to provide breeding organizations with information on new phenotypes for cheese yield and milk nutrient recovery, potentially allowing these traits to be enhanced through selection.     

Slower proteolysis in Cheddar cheese made from high-protein cheese milk is due to an elevated whey protein content

A growing number of companies within the cheese-making industry are now using high-protein (e.g., 4–5%) milks to increase cheese yield. Previous studies have suggested that cheeses made from high-protein (both casein and whey protein; WP) milks may ripen more slowly; one suggested explanation is inhibition of residual rennet activity due to elevated WP levels. We explored the use of microfiltration (MF) to concentrate milk for cheese-making, as that would allow us to concentrate the casein while varying the WP content. Our objective was to determine if reducing the level of WP in concentrated cheese milk had any impact on cheese characteristics, including ripening, texture, and nutritional profile. Three types of 5% casein standardized and pasteurized cheese milks were prepared that had various casein:true protein (CN:TP) ratios: (a) control with CN:TP 83:100, (b) 35% WP reduced, 89:100 CN:TP, and (c) 70% WP reduced, 95:100 CN:TP. Standardized milks were preacidified to pH 6.2 with dilute lactic acid during cheese-making. Composition, proteolysis, textural, rheological, and sensory properties of cheeses were monitored over a 9-mo ripening period. The lactose, total solids, total protein, and WP contents in the 5% casein concentrated milks were reduced with increasing levels of WP removal. All milks had similar casein and total calcium levels. Cheeses had similar compositions, but, as expected, lower WP levels were observed in the cheeses where WP depletion by MF was performed on the cheese milks. Cheese yield and nitrogen recoveries were highest in cheese made with the 95:100 CN:TP milk. These enhanced recoveries were due to the higher fraction of nitrogen being casein-based solids. Microfiltration depletion of WP did not affect pH, sensory attributes, or insoluble calcium content of cheese. Proteolysis (the amount of pH 4.6 soluble nitrogen) was lower in control cheeses compared with WP-reduced cheeses. During ripening, the hardness values and the temperature of the crossover point, an indicator of the melting point of the cheese, were higher in the control cheese. It was thus likely that the higher residual WP content in the control cheese inhibited proteolysis during ripening, and the lower breakdown rate resulted in its higher hardness and melting point. There were no major differences in the concentrations of key nutrients with this WP depletion method. Cheese milk concentration by MF provides the benefit of more typical ripening rates.     

Effects of protein and fat levels in milk on cheese and whey compositions

Bulk tank milk was standardised to six levels of fat (3·0, 3·2, 3·4, 3·6, 3·8, 4·0%) and similarly to six levels of protein, thus giving a total of 36 combinations in composition. Milk was analyzed for total solids, fat, protein, casein, lactose and somatic cell count and was used to make laboratory-scale cheese. Cheese samples from each batch were assayed for total solids, fat, protein and salt. Losses of milk components in the whey were also determined. Least squares analysis of data indicated that higher protein level in milk was associated with higher protein and lower fat contents in cheese. This was accompanied by lower total solids (higher moisture) in cheese. Inversely, higher fat level in milk gave higher fat and lower protein and moisture contents in cheese. Higher fat level in milk resulted in lower retention of fat in cheese and more fat losses in the whey. Higher protein level in milk gave higher fat retention in cheese and less fat losses in the whey. Regression analysis showed that cheese fat increased by 4·22%, while cheese protein decreased by 2·61% for every percentage increase in milk fat. Cheese protein increased by 2·35%, while cheese fat decreased by 6·14% per percentage increase in milk protein. Milk with protein to fat ratio close to 0·9 would produce a minimum of 50% fat in the dry matter of cheese.     

Short communication: utilization of sheep's milk cheese whey in the manufacture of an alkylphenol flavor concentrate

The recovery of species-related conjugated sheep-like flavored alkylphenols from Manchego-type cheese whey by ultrafiltration was investigated. Concentrations of conjugated alkylphenols were similar in the various fractions of whey permeate collected during ultrafiltration, and this was interpreted as a reflection of their high water solubility. About 49 and 62% of conjugated 3- and 4-ethylphenols and p- and m-cresols in sheep's milk cheese whey, respectively, were recovered in the permeate after ultrafiltration with a volume concentration factor of 5.4. Cheese whey retentate correspondingly contained 38 and 28% of conjugated 3- and 4-ethylphenols and p- and m-cresols from the original whey, respectively. Permeate fractions from sheep's milk cheese whey were combined, concentrated by vacuum evaporation, and lactose was partially removed by crystallization and filtration to obtain an aqueous sheep-like flavor precursor concentrate.     

Thermocalcic aggregation of milk fat globule membrane fragments from acid buttermilk cheese whey

Fragments originating from the milk fat globule membrane (MFGM), which is rich in polar lipids and membrane-specific proteins, are gaining interest for their functional and nutritional properties. Acid buttermilk cheese whey was used as a source for MFGM purification, because its MFGM content is more than 5 times higher than that of standard rennet whey. Because polar lipids are the main constituent of the MFGM and only occur in membranous structures, the polar lipid content was taken as a parameter for the total MFGM fragment content. The process of thermocalcic aggregation was evaluated on its recovery of MFGM fragments in the pellet. This method, originally intended for whey clarification and defatting, is a combination of calcium addition, a pH increase, and a thermal treatment. The influence of pH (6.5 to 8), temperature (40 to 70°C), and calcium concentration (0.1 to 0.24 g/100 g) on the pellet mass and dry matter (DM) content and on recovery of protein and polar lipids (and thus indirectly on MFGM fragments) was investigated by means of a response surface Box-Behnken orthogonal design. Reduced quadratic models were fit to the experimental data and were found to be highly significant. No outliers were observed. The recovery of MFGM fragments was found to be highly dependent on the pH, and less dependent on temperature and calcium addition. Next to MFGM proteins, whey proteins were also found to be involved in the formation of aggregates. Optimal conditions were found at 55°C, pH 7.7, and 0.205 g of calcium/L of whey. Under these conditions, 91.0% of the whey polar lipids were recovered in a firm and compact pellet of only 7.86% of the original whey mass, with a polar lipid concentration of 8.34% on pellet DM. Washing with water and centrifugation of the pellet was successful because after one washing step, virtually all sugars were removed, whereas 75.9% of the whey polar lipids could still be recovered. As such, the polar lipid content of the washed pellet increased to 10.70% on a DM basis. However, a second washing step resulted in serious losses of MFGM material.     

Filtration of milk fat globule membrane fragments from acid buttermilk cheese whey

The proteins and polar lipids present in milk fat globule membrane (MFGM) fragments are gaining attention for their technological and nutritional properties. These MFGM fragments are preferentially enriched in side streams of the dairy industry, like butter serum, buttermilk, and whey. The objective of this study was to recover MFGM fragments from whey by tangential filtration techniques. Acid buttermilk cheese whey was chosen as a source for purification by tangential membrane filtration because it is relatively rich in MFGM-fragments and because casein micelles are absent. Polyethersulfone and cellulose acetate membranes of different pore sizes were evaluated on polar lipid and MFGM-protein retention upon filtration at 40°C. All fractions were analyzed for dry matter, ash, lipids, proteins, reducing sugars, polar lipid content by HPLC, and for the presence of MFGM proteins by sodium dodecyl sulfate-PAGE. A fouling coefficient was calculated. It was found that a thermocalcic aggregation whey pretreatment was very effective in the clarification of the whey, but resulted in low permeate fluxes and high retention of ash and whey proteins. By means of an experimental design, the influence of pH and temperature on the fouling and the retention of polar lipids (and thus MFGM fragments), proteins, and total lipids upon microfiltration with 0.15 μM cellulose acetate membrane was investigated. All models were highly significant, and no outliers were observed. By increasing the pH from 4.6 to 7.5, polar lipid retention at 50°C increased from 64 to 98%, whereas fouling of the filtration membrane was minimized. A 3-step diafiltration of acid whey under these conditions resulted in a polar lipid concentration of 6.79 g/100 g of dry matter. As such, this study shows that tangential filtration techniques are suited for the purification of MFGM fragments.     

Compositional and physical properties of yogurts manufactured from milk and whey cheese concentrated by ultrafiltration

Yogurts made with 80% milk retentate (MR) [Volume Reduction Factor (VRF) = 1.5] and 20% cheese whey retentate (WR; VRF = 8.0) (yogurt 1) and yogurts made with 100% MR through ultrafiltration have been evaluated as to flow, texture profile analysis (TPA) and syneresis index. As with MR and WR, their physico‐chemical composition was also determined. The yogurt to which WR had been added showed; less apparent viscosity and greater tixotrophya; less firmness and adhesiveness and greater cohesiveness; higher syneresis index, less protein and mineral content, and greater lipid content in comparison with the yogurt made only with MR.     

Comparison between genetic parameters of cheese yield and nutrient recovery or whey loss traits measured from individual model cheese-making methods or predicted from unprocessed bovine milk samples using Fourier-transform infrared spectroscopy

Cheese yield is an important technological trait in the dairy industry. The aim of this study was to infer the genetic parameters of some cheese yield-related traits predicted using Fourier-transform infrared (FTIR) spectral analysis and compare the results with those obtained using an individual model cheese-producing procedure. A total of 1,264 model cheeses were produced using 1,500-mL milk samples collected from individual Brown Swiss cows, and individual measurements were taken for 10 traits: 3 cheese yield traits (fresh curd, curd total solids, and curd water as a percent of the weight of the processed milk), 4 milk nutrient recovery traits (fat, protein, total solids, and energy of the curd as a percent of the same nutrient in the processed milk), and 3 daily cheese production traits per cow (fresh curd, total solids, and water weight of the curd). Each unprocessed milk sample was analyzed using a MilkoScan FT6000 (Foss, Hillerød, Denmark) over the spectral range, from 5,000 to 900 wavenumber × cm⁻¹. The FTIR spectrum-based prediction models for the previously mentioned traits were developed using modified partial least-square regression. Cross-validation of the whole data set yielded coefficients of determination between the predicted and measured values in cross-validation of 0.65 to 0.95 for all traits, except for the recovery of fat (0.41). A 3-fold external validation was also used, in which the available data were partitioned into 2 subsets: a training set (one-third of the herds) and a testing set (two-thirds). The training set was used to develop calibration equations, whereas the testing subsets were used for external validation of the calibration equations and to estimate the heritabilities and genetic correlations of the measured and FTIR-predicted phenotypes. The coefficients of determination between the predicted and measured values in cross-validation results obtained from the training sets were very similar to those obtained from the whole data set, but the coefficient of determination of validation values for the external validation sets were much lower for all traits (0.30 to 0.73), and particularly for fat recovery (0.05 to 0.18), for the training sets compared with the full data set. For each testing subset, the (co)variance components for the measured and FTIR-predicted phenotypes were estimated using bivariate Bayesian analyses and linear models. The intraherd heritabilities for the predicted traits obtained from our internal cross-validation using the whole data set ranged from 0.085 for daily yield of curd solids to 0.576 for protein recovery, and were similar to those obtained from the measured traits (0.079 to 0.586, respectively). The heritabilities estimated from the testing data set used for external validation were more variable but similar (on average) to the corresponding values obtained from the whole data set. Moreover, the genetic correlations between the predicted and measured traits were high in general (0.791 to 0.996), and they were always higher than the corresponding phenotypic correlations (0.383 to 0.995), especially for the external validation subset. In conclusion, we herein report that application of the cross-validation technique to the whole data set tended to overestimate the predictive ability of FTIR spectra, give more precise phenotypic predictions than the calibrations obtained using smaller data sets, and yield genetic correlations similar to those obtained from the measured traits. Collectively, our findings indicate that FTIR predictions have the potential to be used as indicator traits for the rapid and inexpensive selection of dairy populations for improvement of cheese yield, milk nutrient recovery in curd, and daily cheese production per cow.     

Predicting cow milk quality traits from routinely available milk spectra using statistical machine learning methods

Numerous statistical machine learning methods suitable for application to highly correlated features, as those that exist for spectral data, could potentially improve prediction performance over the commonly used partial least squares approach. Milk samples from 622 individual cows with known detailed protein composition and technological trait data accompanied by mid-infrared spectra were available to assess the predictive ability of different regression and classification algorithms. The regression-based approaches were partial least squares regression (PLSR), ridge regression (RR), least absolute shrinkage and selection operator (LASSO), elastic net, principal component regression, projection pursuit regression, spike and slab regression, random forests, boosting decision trees, neural networks (NN), and a post-hoc approach of model averaging (MA). Several classification methods (i.e., partial least squares discriminant analysis (PLSDA), random forests, boosting decision trees, and support vector machines (SVM)) were also used after stratifying the traits of interest into categories. In the regression analyses, MA was the best prediction method for 6 of the 14 traits investigated [curd firmness at 60 min, α_S1-casein (CN), α_S2-CN, κ-CN, α-lactalbumin, and β-lactoglobulin B], whereas NN and RR were the best algorithms for 3 traits each (rennet coagulation time, curd-firming time, and heat stability, and curd firmness at 30 min, β-CN, and β-lactoglobulin A, respectively), PLSR was best for pH, and LASSO was best for CN micelle size. When traits were divided into 2 classes, SVM had the greatest accuracy for the majority of the traits investigated. Although the well-established PLSR-based method performed competitively, the application of statistical machine learning methods for regression analyses reduced the root mean square error compared with PLSR from between 0.18% (κ-CN) to 3.67% (heat stability). The use of modern statistical machine learning methods for trait prediction from mid-infrared spectroscopy may improve the prediction accuracy for some traits.     

Short communication: Determination of lactoferrin in Feta cheese whey with reversed-phase high-performance liquid chromatography

In the current paper, a method is introduced to determine lactoferrin in sweet whey using reversed-phase HPLC without any pretreatment of the samples or use of a separation technique. As a starting point, the most common HPLC protocols for acid whey, which included pretreatment of the whey along with a sodium dodecyl sulfate-PAGE step, were tested. By skipping the pretreatment and the separation steps while altering the gradient profile, different chromatographs were obtained that proved to be equally efficient to determine lactoferrin. For this novel 1-step reversed-phase HPLC method, repeatability was very high over a wide range of concentrations (1.88% intraday to 5.89% interday). The limit of detection was 35.46 μg/mL [signal:noise ratio (S/N) = 3], whereas the limit of quantification was 50.86 μg/mL (S/N = 10). Omitting the pretreatment step caused a degradation of the column’s lifetime (to approximately 2,000 samples). As a result, the lactoferrin elution time changed, but neither the accuracy nor the separation ability of the method was significantly influenced. We observed that this degradation could be easily avoided or detained by centrifuging the samples to remove fat or by extensive cleaning of the column after every 5 samples.     

基于近红外光谱结合机器学习的鳕鱼品种二分类方法研究

目的 鳕鱼假冒伪劣事件层出不穷,为保障鳕鱼肉制品购买者能够买到放心的鳕鱼产品,探索适合分析鳕鱼近红外光谱数据的机器学习模型,实现鳕鱼品种的快速二分类。方法 选取挪威大西洋真鳕、冰岛黑线鳕等8种鳕鱼,对其研磨物进行傅里叶变换近红外光谱测试,并采用Min-Max归一化和独立成分分析法对近红外光谱数据进行预处理和降维,进一步分别使用9种机器学习模型进行二分类,通过6项指标对比各个模型的预测效果,从中选出最适合鳕鱼二分类的模型。结果 本文提出的独立成分析法结合支持向量机的鳕鱼品种二分类模型的预测准确率可达到97.2%,F1分数可达到97.3%,召回率达到99.4%。结论 本研究可实现较为准确的大西洋鳕鱼和非大西洋鳕鱼品种的分类,为鳕鱼品种鉴别提供了方法依据。     

Near-infrared spectroscopy: A possible routine analysis to detect the use of frozen curds in the production of Buffalo Mozzarella cheese

Mozzarella di Bufala Campana is a pasta filata cheese with a Protected Designation of Origin whose specifications require the use of fresh milk, forbidding the use of frozen curds. The goal is to develop a routine analysis to identify frozen curd presence. The Buffalo Mozzarella samples were analysed by the near-infrared analytical technique, and the spectral data were processed through an artificial neural network. The results make it possible to identify the use of frozen curd in samples of Buffalo Mozzarella for up to 9 days of storage. The model reported very high accuracy either in training (0.5% of bad prediction) or in tests (6.8% of bad prediction).     

Cheddar cheese classification based on flavor quality using a novel extraction method and Fourier transform infrared spectroscopy

Analysis of Cheddar cheese flavor using trained sensory and grading panels is expensive and time consuming. A rapid and simple solvent extraction procedure in combination with Fourier transform infrared spectroscopy was developed for classifying Cheddar cheese based on flavor quality. Fifteen Cheddar cheese samples from 2 commercial production plants were ground into powders using liquid nitrogen. The water-soluble compounds from the cheese powder, without interfering compounds such as fat and protein, were extracted using water, chloroform, and ethanol. Aliquots (10 μL) of the extract were placed on a zinc selenide crystal, vacuum dried, and scanned in the mid-infrared region (4,000 to 700 cm⁻¹). The infrared spectra were analyzed by soft independent modeling of class analogy (SIMCA) for pattern recognition. Sensory flavor quality of these cheeses was determined by trained quality assurance personnel in the production facilities. The SIMCA models provided 3-dimensional classification plots in which all the 15 cheese samples formed well-separated clusters. The orientation of the clusters in 3-dimensional space correlated well with their cheese flavor characteristics (fermented, unclean, low flavor, sour, good Cheddar, and so on). The discrimination of the samples in the SIMCA plot was mainly due to organic acids, fatty acids and their esters, and amino acids (1,450 to 1,350 and 1,200 to 990 cm⁻¹), which are known to contribute significantly to cheese flavor. The total analysis time, including the sample preparation time, was less than 20 min per sample. This technique can be a rapid, inexpensive, and simple tool to the cheese industry for predicting the flavor quality of cheese.     

Transfer of melamine from feed to milk and from milk to cheese and whey in lactating dairy cows fed single oral doses

A study was conducted to evaluate the excretion pattern, after a single oral dose, of melamine from feed into milk, and the subsequent transfer to cheese and whey. The transfer of cyanuric acid was also investigated. Twenty-four lactating Holstein cows were randomly allocated to 4 treatments and received single doses of melamine as follows: 0.05, 0.50, 5.00, and 50.00 g/cow for groups D1, D2, D3, and D4, respectively. Individual milk samples were collected for melamine and cyanuric acid analyses on d 1, 2, 3, 4, 5, and 7. Milk collected individually from the second milking after melamine ingestion was used to make cheese on a laboratory scale. Melamine and cyanuric acid were extracted using a solid-phase extraction cartridge, and analyses were carried out by liquid chromatography-mass spectrometry/mass spectrometry. Maximal melamine concentrations occurred between 6 and 18 h after treatment and increased with log dose (linear and quadratic), ranging from 0.019 to 35.105 mg/kg. More than 60% of the melamine that was transferred to the milk was observed within 30 h after melamine ingestion. Melamine was not detected (limit of detection was 0.002 mg/kg) in milk 5 d after treatment in group D1, and 7 d after treatment in groups D2, D3, and D4. Blood urea nitrogen was not influenced by melamine ingestion. During cheese making, melamine was transferred mainly to the whey fraction. Cyanuric acid was not detected in any of the samples (milk, cheese, or whey). The excretion pattern of melamine in milk and whey may represent a health concern when cows ingest more than 0.50 g of melamine/d. However, only at intake levels of 5 and 50 g/d did cheese exceed the limits as set forth by the European Union. The results confirmed that melamine contamination of milk and milk products may be related not only to direct contamination, but also to adulteration of animal feeds.     

Relationships between cheese-processing conditions and curd and cheese properties to improve the yield of Idiazabal cheese made in small artisan dairies: A multivariate approach

Very diverse cutting and cooking intensity processes are currently used in small artisan dairies to manufacture Idiazabal cheese. The combination of the technical settings used during cheese manufacturing is known to affect cheese composition and yield, as well as whey losses. However, the information regarding the effect on microstructure and texture of cheese is scarce, especially in commercial productions. Therefore, the effect of moderate- and high-intensity cutting and cooking processes on whey losses, curd-grain characteristics, microstructure and cheese properties, and yield were analyzed. Three trials were monitored in each of 2 different small dairies during the cheesemaking of Idiazabal cheese, which is a semihard cheese made from raw sheep milk. The role and know-how of the cheesemakers are crucial in these productions because they determine the cutting point and handle semi-automatic vats. The 2 dairies studied used the following settings: dairy A used moderate-intensity cutting and cooking conditions, and dairy B used high-intensity cutting and cooking settings. Multiple relationships between cheese-processing conditions and curd, whey, and cheese properties as well as yield were obtained from a partial least square regression analysis. An increased amount of fat and casein losses were generated due to a combination of an excessively firm gel at cutting point together with high-intensity cutting and cooking processes. The microstructural analysis revealed that the porosity of the protein matrix of curd grains after cooking and cheese after pressing was the main feature affected, developing a less porous structure with a more intense process. Moderate-intensity cutting and cooking processes were associated with a higher cheese yield, regardless of the longer pressing process applied. No significant differences were observed in cheese composition. After 1 mo of ripening, however, the cheese was more brittle and adhesive when the high-intensity cutting and cooking process was applied. This could be associated with the composition, characteristics, and size distribution of curd grains due to differences in the compaction degree during pressing. These results could help to modify specific conditions used in cheesemaking, especially improving the process in those small dairies where the role of the cheesemaker is crucial.     

Front-face fluorescence spectroscopy combined with chemometrics to detect high proteinaceous matter in milk and whey ultrafiltration permeate

Proteinaceous matter can leak into the permeate stream during ultrafiltration (UF) of milk and whey and lead to financial losses. Although manufacturers can measure protein content in the finished permeate powders, there is currently no rapid monitoring tool during UF to identify protein leak. This study applied front-face fluorescence spectroscopy (FFFS) and chemometrics to identify the fluorophore of interest associated with the protein leak, develop predictive models to quantify true protein content, and classify the types of protein leak in permeate streams. Crude protein (CP), nonprotein nitrogen (NPN), true protein (TP), tryptone-equivalent peptide (TEP), α-lactalbumin (α-LA), and β-lactoglobulin (β-LG) contents were measured for 37 lots of whey permeate and 29 lots of milk permeate from commercial manufacturers. Whey permeate contained more TEP than did milk permeate, whereas milk permeate contained more α-LA and β-LG than did whey permeate. The types of protein leak were thus identified for predictive model development. Based on excitation-emission matrix (EEM) of high- and low-TP permeates, tryptophan excitation spectra were collected for predictive model development, measuring TP content in permeate. With external validation, a useful model for quality control purposes was developed, with a root mean square error of prediction of 0.22% (dry basis) and a residual prediction deviation of 2.8. Moreover, classification models were developed using partial least square discriminant analysis. These classification methods can detect high TP level, high TEP level, and presence of α-LA or β-LG with 83.3%, 84.8%, and 98.5% cross-validated accuracy, respectively. This method showed that FFFS and chemometrics can rapidly detect protein leaks and identify the types of protein leak in UF permeate. Implementation of this method in UF processing plants can reduce financial loss from protein leaks and maintain high-quality permeate production.     

Quality and nutritional characteristics of traditional curd cheese enriched with thermo-coagulated acid whey protein and indigenous Lactococcus lactis strain

Consumer interest in a healthy lifestyle and health-promoting natural products is a major driving force for the increasing global demand for bio-functional and sustainable dairy foods. Supplementation of curd cheese with thermo-coagulated acid whey protein (TAWP) led to 8–10% higher contents of moisture, 23–31% of lactose, 12–13% of unsaturated, 5–6% of monounsaturated, 63% of polyunsaturated, and 3–4% of long-chain fatty acids. Lipid quality indices – TI, AI, h/H, and Omega 6/3 among others, were also significantly better than those of control cheese. The addition of indigenous Lactococcus lactis strain enhanced the flavour of cheese samples, decreased the counts of yeast and mould up to 1 log cfu g⁻¹ after 10 days of storage. The replacement of curd with TAWP resulted in novel cheese with overall sensory perception above 77 points and added nutritional and functional value, however, body and texture parameters of modified cheese samples require an improvement.     

A comparison and joint use of NIR and MIR spectroscopic methods for the determination of some parameters in European Emmental cheese

The determination of winter cheese chemical properties, namely, fat, sodium chloride (NaCl), pH, non protein nitrogen (NPN), total nitrogen (TN) and water soluble nitrogen (WSN) was done using spectroscopic technologies with different wavelength zones. The Emmental cheeses provided from different European countries were studied. A total of 91 cheeses produced during the winter time in Austria (n=4), Finland (n=6), Germany (n=13), France (n=30) and Switzerland (n=38) were analysed by near infrared (NIR) and mid infrared (MIR) spectroscopies. The combination of these two spectral regions (sum of their spectra) was also studied. The partial least square (PLS) regression with the leave one-out cross validation technique was used to build up calibration models using data set designated as calibration set. These models were validated with another data set designated as validation set. The obtained results suggest the use of the NIR for the determination of fat and TN contents, and the MIR for NaCl and NPN contents as well as for the pH. Similar results were obtained for WSN using the two techniques together. The combined spectra of both NIR and MIR did improve the results, while providing comparable results to those obtained from either the NIR or MIR spectroscopy.     

鲸鱼算法改进极限学习机的葡萄酒品质评价研究

[目的]解决近红外光谱中冗余信息过多的问题,提升葡萄酒品质评价模型的准确性,并构建一种快速无损的葡萄酒品质评价方法。[方法]运用竞争性自适应重加权采样法进行特征波长筛选,提出了鲸鱼算法改进极限学习机的葡萄酒品质评价模型。通过自适应重加权采样法等多种特征波长筛选方法,确定了最适用于葡萄酒光谱特征波长筛选的方法;针对ELM的初值权值与隐含层偏置选取问题,利用鲸鱼优化方法对初值权值与隐含层偏置进行优化,从而构建了一种基于鲸鱼优化算法改进的极限学习机葡萄酒品质评价模型。[结果]与GA-ELM、PSO-ELM和传统的ELM模型相比,WOA-ELM的准确率最高,达到了0.944 5,GA-ELM的准确率为0.929 0,PSO-ELM的准确率为0.906 1,传统的ELM方法准确率为0.817 7。[结论]通过智能算法优化ELM模型的参数,可以有效提高葡萄酒品质评价的准确性。     

Estimation of body condition score change in dairy cows in a seasonal calving pasture-based system using routinely available milk mid-infrared spectra and machine learning techniques

Body condition score (BCS) is a subjective estimate of body reserves in cows. Body condition score and its change in early lactation have been associated with cow fertility and health. The aim of the present study was to estimate change in BCS (ΔBCS) using mid-infrared spectra of the milk, with a particular focus on estimating ΔBCS in cows losing BCS at the fastest rate (i.e., the cows most of interest to the producer). A total of 73,193 BCS records (scale 1 to 5) from 6,572 cows were recorded. Daily BCS was interpolated from cubic splines fitted through the BCS records, and subsequently used to calculate daily ΔBCS. Body condition score change records were merged with milk mid-infrared spectra recorded on the same week. Both morning (a.m.) and evening (p.m.) spectra were available. Two different statistical methods were used to estimate ΔBCS: partial least squares regression and a neural network (NN). Several combinations of variables were included as model features, such as days in milk (DIM) only, a.m. spectra only and DIM, p.m. spectra only and DIM, and a.m. and p.m. spectra as well as DIM. The data used to estimate ΔBCS were either based on the first 120 DIM or all 305 DIM. Daily ΔBCS had a standard deviation of 1.65 × 10⁻³ BCS units in the 305 DIM data set and of 1.98 × 10⁻³ BCS units in the 120 DIM data set. Each data set was divided into 4 sub-data sets, 3 of which were used for training the prediction model and the fourth to test it. This process was repeated until all the sub-data sets were considered as the test data set once. Using all 305 DIM, the lowest root mean square error of validation (RMSEV; 0.96 × 10⁻³ BCS units) and the strongest correlation between actual and estimated ΔBCS (0.82) was achieved with NN using a.m. and p.m. spectra and DIM. Using the 120 DIM data, the lowest RMSEV (0.98 × 10⁻³ BCS units) and the strongest correlation between actual and estimated ΔBCS (0.87) was achieved with NN using DIM and either a.m. spectra only or a.m. and p.m. spectra together. The RMSEV for records in the lowest 2.5% ΔBCS percentile per DIM in early lactation was reduced up to a maximum of 13% when spectra and DIM were both considered in the model compared with a model that considered just DIM. The performance of the NN using DIM and a.m. spectra only with the 120 DIM data was robust across different strata of farm, parity, year of sampling, and breed. Results from the present study demonstrate the ability of mid-infrared spectra of milk coupled with machine learning techniques to estimate ΔBCS; specifically, the inclusion of spectral data reduced the RMSEV over and above using DIM alone, particularly for cows losing BCS at the fastest rate. This approach can be used to routinely generate estimates of ΔBCS that can subsequently be used for farm decisions.