Search

The huge amount of data produced by the Internet of Things need to be validated and curated to be prepared for the selection of relevant data in order to prototype models, train them, and serve the model. On the other side, blockchains and open data are also important data sources that need to be integrated into the proposed integrative models. It is difficult to find a sufficiently versatile and agnostic architecture based on the main machine learning frameworks that facilitate model development and allow continuous training to continuously improve them from the data streams. The paper describes the conceptualization, implementation, and testing of a new architecture that proposes a use case agnostic processing chain. The proposed architecture is mainly built around the Apache Submarine, an unified Machine Learning platform that facilitates the training and deployment of algorithms. Here, Internet of Things data are collected and formatted at the edge level. They are then processed and validated at the fog level. On the other hand, open data and blockchain data via Blockchain Access Layer are directly processed at the cloud level. Finally, the data are preprocessed to feed scalable machine learning algorithms. [ABSTRACT FROM AUTHOR]

To develop a breed assignment model, three main steps are generally followed: 1) The selection of breed informative single nucleotide polymorphism (SNP); 2) The training of a model, based on a reference population, that allows to classify animals to their breed of origin; and 3) The validation of the developed model on external animals i.e., that were not used in previous steps. However, there is no consensus in the literature about which methodology to follow for the first step, nor about the number of SNP to be selected. This can raise many questions when developing the model and lead to the use of sophisticated methodologies for selecting SNP (e.g., with iterative algorithms, partitions of SNP, or combination of several methods). Therefore, it may be of interest to avoid the first step by the use of all the available SNP. For this purpose, we propose the use of a genomic relationship matrix (GRM), combined or not with a machine learning method, for breed assignment. We compared it with a previously developed model based on selected informative SNP. Four methodologies were investigated: 1) The PLS_NSC methodology: selection of SNP based on a partial least square-discriminant analysis (PLS-DA) and breed assignment by classification based on the nearest shrunken centroids (NSC) method; 2) Breed assignment based on the highest mean relatedness of an animal to the reference populations of each breed (referred to mean_GRM); 3) Breed assignment based on the highest SD of the relatedness of an animal to the reference populations of each breed (referred to SD_GRM) and 4) The GRM_SVM methodology: the use of means and SD of the relatedness defined in mean_GRM and SD_GRM methodologies combined with the linear support vector machine (SVM), a machine learning method used for classification. Regarding mean global accuracies, results showed that the use of mean_GRM or GRM_SVM was not significantly different (Bonferroni corrected P > 0.0083) than the model based on a reduced SNP panel (PLS_NSC). Moreover, the mean_GRM and GRM_SVM methodology were more efficient than PLS_NSC as it was faster to compute. Therefore, it is possible to bypass the selection of SNP and, by the use of a GRM, to develop an efficient breed assignment model. In routine, we recommend the use of GRM_SVM over mean_GRM as it gave a slightly increased global accuracy, which can help endangered breeds to be maintained. The script to execute the different methodologies can be accessed on [ABSTRACT FROM AUTHOR]

Mass spectrometry (MS)-based techniques, including liquid chromatography coupling, shotgun lipidomics, MS imaging, and ion mobility, are widely used to analyze lipids. However, with enhanced separation capacity and an optimized chemical derivatization approach, comprehensive two-dimensional gas chromatography (GC×GC) can be a powerful tool to investigate some groups of small lipids in the framework of lipidomics. This study describes the optimization of a dedicated two-stage derivatization and extraction process to analyze different saturated and unsaturated fatty acids in plasma by two-dimensional gas chromatography–time-of-flight mass spectrometry (GC×GC–TOFMS) using a full factorial design. The optimized condition has a composite desirability of 0.9159. This optimized sample preparation and chromatographic condition were implemented to differentiate between positive (BT) and negative (UT) boar-tainted pigs based on fatty acid profiling in pig serum using GC×GC–TOFMS. A chemometric screening, including unsupervised (PCA, HCA) and supervised analysis (PLS–DA), as well as univariate analysis (volcano plot), was performed. The results suggested that the concentration of PUFA ω-6 and cholesterol derivatives were significantly increased in BT pigs, whereas SFA and PUFA ω-3 concentrations were increased in UT pigs. The metabolic pathway and quantitative enrichment analysis suggest the significant involvement of linolenic acid metabolism. [ABSTRACT FROM AUTHOR]

The objectives of this study were to estimate genetic parameters and identify genomic regions associated with milk urea concentration (MU) in Dual‐Purpose Belgian Blue (DPBB) cows. The data were 29,693 test‐day records of milk yield (MY), fat yield (FY), protein yield (PY), fat percentage (FP), protein percentage (PP) and MU collected between 2014 and 2020 on 2498 first parity cows (16,935 test‐day records) and 1939 second‐parity cows (12,758 test‐day records) from 49 herds in the Walloon Region of Belgium. Data of 28,266 single nucleotide polymorphisms (SNP), located on 29 Bos taurus autosomes (BTA), on 1699 animals (639 males and 1060 females) were used. Random regression test‐day models were used to estimate genetic parameters through the Bayesian Gibbs sampling method using a single chain of 100,000 iterations after a burn‐in period of 20,000. SNP solutions were estimated using a single‐step genomic best linear unbiased prediction approach. The proportion of genetic variance explained by windows of 25 consecutive SNPs (with an average size of ~2 Mb) was calculated, and regions accounting for at least 1.0% of the total additive genetic variance were used to search for candidate genes. The mean (SD) of MU was 22.89 (10.07) and 22.35 (10.07) mg/dl for first and second parity, respectively. The mean (SD) heritability estimates for daily MU were 0.18 (0.01) and 0.22 (0.02), for first and second parity, respectively. The mean (SD) genetic correlations between daily MU and MY, FY, PY, FP and PP were −0.05 (0.09), −0.07 (0.11), −0.03 (0.13), −0.05 (0.08) and −0.03 (0.11) for first parity, respectively. The corresponding values estimated for second parity were 0.02 (0.10), −0.02 (0.09), 0.02 (0.08), −0.08 (0.06) and −0.05 (0.05). The genome‐wide association analyses identified three genomic regions (BTA2, BTA3 and BTA13) associated with MU. The identified regions showed contrasting results between parities and among different stages within each parity. This suggests that different groups of candidate genes underlie the phenotypic expression of MU between parities and among different lactation stages within a parity. The results of this study can be used for future implementation and use of genomic evaluation to reduce MU in DPBB cows. [ABSTRACT FROM AUTHOR]

Simple Summary: The dairy industry is interested in developing a detection tool for milk produced from pasture to certify the protected designation of origin of certain dairy products. As the cattle's grazing influences the milk composition, the milk mid-infrared (MIR) spectra will be modified. Therefore, this study aims to develop a predictive model allowing one to use milk MIR spectrometry to classify milk as produced from a cattle diet either based on grass or not. To achieve this objective, the collection of grazing calendars from farms is needed. Unfortunately, this is hardly possible, as this information is rarely recorded. Therefore, the innovation of this research consists of using the usual farming practices developed in the southern part of Belgium combined with a large-scale milk database containing 48 milk composition-related traits. Indeed, in this geographical area, the cows are mainly on pasture between April and September. Therefore, the month of testing can be considered indirect information about their feeding. The developed models were able to distinguish with a precision of around 90% the supposed grass-based diet. Therefore, the probability of belonging to the GRASS class could be used in a tool counting the number of grazing days to confirm the labeling of dairy products. This research aims to develop a predictive model to discriminate milk produced from a cattle diet either based on grass or not using milk mid-infrared spectrometry and the month of testing (an indirect indicator of the feeding ration). The dataset contained 3,377,715 spectra collected between 2011 and 2021 from 2449 farms and 3 grazing traits defined following the month of testing. Records from 30% of the randomly selected farms were kept in the calibration set, and the remaining records were used to validate the models. Around 90% of the records were correctly discriminated. This accuracy is very good, as some records could be erroneously assigned. The probability of belonging to the GRASS modality allowed confirmation of the model's ability to detect the transition period even if the model was not trained on this data. Indeed, the probability increased from the spring to the summer and then decreased. The discrimination was mainly explained by the changes in the milk fat, mineral, and protein compositions. A hierarchical clustering from the averaged probability per farm and year highlighted 12 groups illustrating different management practices. The probability of belonging to the GRASS class could be used in a tool counting the number of grazing days. [ABSTRACT FROM AUTHOR]

The aims of this study were to: (1) estimate genetic correlation for milk production traits (milk, fat and protein yields and fat and protein contents) and fatty acids (FA: C16:0, C18:1 cis-9, LCFA, SFA, and UFA) over days in milk, (2) investigate the performance of genomic predictions using single-step GBLUP (ssGBLUP) based on random regression models (RRM), and (3) identify the optimal scaling and weighting factors to be used in the construction of the H matrix. A total of 302 684 test-day records of 63.875 first lactation Walloon Holstein cows were used. Positive genetic correlations were found between milk yield and fat and protein yield (rg from 0.46 to 0.85) and between fat yield and milk FA (rg from 0.17 to 0.47). On the other hand, negative correlations were estimated between fat and protein contents (rg from −0.22 to −0.59), between milk yield and milk FA (rg from −0.22 to −0.62), and between protein yield and milk FA (rg from −0.11 to −0.19). The selection for high fat content increases milk FA throughout lactation (rg from 0.61 to 0.98). The test-day ssGBLUP approach showed considerably higher prediction reliability than the parent average for all milk production and FA traits, even when no scaling and weighting factors were used in the H matrix. The highest validation reliabilities (r2 from 0.09 to 0.38) and less biased predictions (b1 from 0.76 to 0.92) were obtained using the optimal parameters (i.e., ω = 0.7 and α = 0.6) for the genomic evaluation of milk production traits. For milk FA, the optimal parameters were ω = 0.6 and α = 0.6. However, biased predictions were still observed (b1 from 0.32 to 0.81). The findings suggest that using ssGBLUP based on RRM is feasible for the genomic prediction of daily milk production and FA traits in Walloon Holstein dairy cattle. [ABSTRACT FROM AUTHOR]

Simple Summary: Mid-infrared (MIR) spectroscopy is the method of choice to determine milk components like fat, protein and urea. We examined the potential of MIR spectra analyses for the prediction of clinical mastitis events of dairy cows additionally, or alternatively, to somatic cell count, which is routinely used as an indicator for mastitis monitoring. Prediction models based on MIR spectra and a somatic cell count-derived score (SCS) were developed and compared. A model based on MIR spectra and SCS proved more accurate at predicting mastitis than models based on either indicator alone. Consequently, MIR spectra analyses add extra value in the prediction of clinical mastitis, making them potentially useful for dairy farm management and as an auxiliary trait for the genetic evaluation of udder health. Monitoring for mastitis on dairy farms is of particular importance, as it is one of the most prevalent bovine diseases. A commonly used indicator for mastitis monitoring is somatic cell count. A supplementary tool to predict mastitis risk may be mid-infrared (MIR) spectroscopy of milk. Because bovine health status can affect milk composition, this technique is already routinely used to determine standard milk components. The aim of the present study was to compare the performance of models to predict clinical mastitis based on MIR spectral data and/or somatic cell count score (SCS), and to explore differences of prediction accuracies for acute and chronic clinical mastitis diagnoses. Test-day data of the routine Austrian milk recording system and diagnosis data of its health monitoring, from 59,002 cows of the breeds Fleckvieh (dual purpose Simmental), Holstein Friesian and Brown Swiss, were used. Test-day records within 21 days before and 21 days after a mastitis diagnosis were defined as mastitis cases. Three different models (MIR, SCS, MIR + SCS) were compared, applying Partial Least Squares Discriminant Analysis. Results of external validation in the overall time window (−/+21 days) showed area under receiver operating characteristic curves (AUC) of 0.70 when based only on MIR, 0.72 when based only on SCS, and 0.76 when based on both. Considering as mastitis cases only the test-day records within 7 days after mastitis diagnosis, the corresponding areas under the curve were 0.77, 0.83 and 0.85. Hence, the model combining MIR spectral data and SCS was performing best. Mastitis probabilities derived from the prediction models are potentially valuable for routine mastitis monitoring for farmers, as well as for the genetic evaluation of the trait udder health. [ABSTRACT FROM AUTHOR]

We investigated the use of different Legendre polynomial orders to estimate genetic parameters for milk production and fatty acid (FA) traits in the first lactation Walloon Holstein cows. The data set comprised 302,684 test‐day records of milk yield, fat and protein contents, and FAs generated by mid‐infrared (MIR) spectroscopy, C16:0 (palmitic acid), C18:1 cis‐9 (oleic acid), LCFAs (long‐chain FAs), SFAs (saturated FAs) and UFAs (unsaturated FAs) were studied. The models included random regression coefficients for herd‐year of calving (h), additive genetic (a) and permanent environment (p) effects. The selection of the best random regression model (RRM) was based on the deviance information criterion (DIC), and genetic parameters were estimated via a Bayesian approach. For all analysed random effects, DIC values decreased as the order of the Legendre polynomials increased. Best‐fit models had fifth‐order (degree 4) for the p effect and ranged from second‐ to fifth‐order (degree 1–4) for the a and h effects (LEGhap: LEG555 for milk yield and protein content; LEG335 for fat content and SFA; LEG545 for C16:0 and UFA; and LEG535 for C18:1 cis‐9 and LCFA). Based on the best‐fit models, an effect of overcorrection was observed in early lactation (5–35 days in milk [DIM]). On the contrary, third‐order (LEG333; degree 2) models showed flat residual trajectories throughout lactation. In general, the estimates of genetic variance tended to increase over DIM, for all traits. Heritabilities for milk production traits ranged from 0.11 to 0.58. Milk FA heritabilities ranged from low‐to‐high magnitude (0.03–0.56). High Spearman correlations (>0.90 for all bulls and >0.97 for top 100) were found among breeding values for 155 and 305 DIM between the best RRM and LEG333 model. Therefore, third‐order Legendre polynomials seem to be most parsimonious and sufficient to describe milk production and FA traits in Walloon Holstein cows. [ABSTRACT FROM AUTHOR]

In recent years, cannabis and hemp-based products have become increasingly popular for recreational use, edibles, beverages, health care products, and medicines. The rapid detection and differentiation of phytocannabinoids is, therefore, essential to assess the potency and the therapeutic and nutritional values of cannabis cultivars. Here, we implemented SpiderMass technology for in vivo detection of cannabidiolic acid (CBDA) and ∆9-tetrahydrocannabinolicacid (∆9-THCA), and other endogenous organic plant compounds, to access distribution gradients within the plants and differentiate between cultivars. The SpiderMass system is composed of an IR-laser handheld microsampling probe connected to a mass spectrometer through a transfer tube. The analysis was performed on different plant organs from freshly cultivated cannabis plants in only a few seconds. SpiderMass analysis easily discriminated the two acid phytocannabinoid isomers via MS/MS, and the built statistical models differentiated between four cannabis cultivars. Different abundancies of the two acid phytocannabinoids were found along the plant as well as between different cultivars. Overall, these results introduce direct analysis by SpiderMass as a compelling analytical alternative for rapid hemp analysis. [ABSTRACT FROM AUTHOR]

Quantifying the level of linkage disequilibrium (LD), non‐random association of alleles at two or more loci, is important to determine the number of markers needed for genomic selection. The aims of this study were to evaluate the extent of LD in Dual‐Purpose Belgian Blue (DPBB) and to compare the level of LD in DPBB with that of Walloon Holstein. Data of 28,427 single nucleotide polymorphisms (SNP), located on 29 Bos taurus autosomes (BTA), of 639 DPBB and 398 Holstein bulls were used. The level of LD between pairwise SNPs separated by up to 10 Mb was evaluated, separately for each breed, using the squared correlation of the alleles at two loci. The analysis of molecular variance showed that the percentage of variation within populations (85.48%) was higher than between populations (14.52%). However, permutation tests showed a significant genetic differentiation between the two studied populations (p <.01). The average LD found between adjacent SNP pairs in DPBB (0.16 (SD = 0.22)) was generally lower than in Holstein (0.23 (SD = 0.27)). The proportion of SNPs in useful LD (r2 > 0.30) within a genomic distance of ≤0.10 Mb between SNPs was 18.58% and 28.23% in DPBB and Holstein bulls, respectively. In both breeds, the effective population size decreased over generations; however, the decline was greater in DPBB than that in Holstein. Based on results, it can be concluded that at least 68,000 SNPs are needed for implementing genomic selection in DPBB cattle with enough accuracy. [ABSTRACT FROM AUTHOR]

Assignment of individual cattle to a specific breed can often not rely on pedigree information. This is especially the case for local breeds for which the development of genomic assignment tools is required to allow individuals of unknown origin to be included to their herd books. A breed assignment model can be based on two specific stages: (a) the selection of breed‐informative markers and (b) the assignment of individuals to a breed with a classification method. However, the performance of combination of methods used in these two stages has been rarely studied until now. In this study, the combination of 16 different SNP panels with four classification methods was developed on 562 reference genotypes from 12 cattle breeds. Based on their performances, best models were validated on three local breeds of interest. In cross‐validation, 14 models had a global cross‐validation accuracy higher than 90%, with a maximum of 98.22%. In validation, best models used 7,153 or 2,005 SNPs, based on a partial least squares‐discriminant analysis (PLS‐DA) and assigned individuals to breeds based on nearest shrunken centroids. The average validation sensitivity of the first two best models for the three local breeds of interest were 98.33% and 97.5%. Moreover, results reported in this study suggest that further studies should consider the PLS‐DA method when selecting breed‐informative SNPs. [ABSTRACT FROM AUTHOR]

BACKGROUND: A robust proxy for estimating methane (CH4) emissions of individual dairy cows would be valuable especially for selective breeding. This study aimed to improve the robustness and accuracy of prediction models that estimate daily CH4 emissions from milk Fourier transform mid‐infrared (FT‐MIR) spectra by (i) increasing the reference dataset and (ii) adjusting for routinely recorded phenotypic information. Prediction equations for CH4 were developed using a combined dataset including daily CH4 measurements (n = 1089; g d−1) collected using the SF6 tracer technique (n = 513) and measurements using respiration chambers (RC, n = 576). Furthermore, in addition to the milk FT‐MIR spectra, the variables of milk yield (MY) on the test day, parity (P) and breed (B) of cows were included in the regression analysis as explanatory variables. RESULTS: Models developed based on a combined RC and SF6 dataset predicted the expected pattern in CH4 values (in g d−1) during a lactation cycle, namely an increase during the first weeks after calving followed by a gradual decrease until the end of lactation. The model including MY, P and B information provided the best prediction results (cross‐validation statistics: R2 = 0.68 and standard error = 57 g CH4 d−1). CONCLUSIONS: The models developed accounted for more of the observed variability in CH4 emissions than previously developed models and thus were considered more robust. This approach is suitable for large‐scale studies (e.g. animal genetic evaluation) where robustness is paramount for accurate predictions across a range of animal conditions. © 2020 Society of Chemical Industry [ABSTRACT FROM AUTHOR]

Copyright of Biotechnologie, Agronomie, Societe et Environnement is the property of Les Presses Agronomiques de Gembloux and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or email articles for individual use. This abstract may be abridged. No warranty is given about the accuracy of the copy. Users should refer to the original published version of the material for the full abstract. (Copyright applies to all Abstracts.)

Recent publications indicate that single‐step models are suitable to estimate breeding values, dominance deviations and total genetic values with acceptable quality. Additive single‐step methods implicitly extend known number of allele information from genotyped to non‐genotyped animals. This theory is well derived in an additive setting. It was recently shown, at least empirically, that this basic strategy can be extended to dominance with reasonable prediction quality. Our study addressed two additional issues. It illustrated the theoretical basis for extension and validated genomic predictions to dominance based on single‐step genomic best linear unbiased prediction theory. This development was then extended to include inbreeding into dominance relationships, which is a currently not yet solved issue. Different parametrizations of dominance relationship matrices were proposed. Five dominance single‐step inverse matrices were tested and described as C1, C2, C3, C4 and C5. Genotypes were simulated for a real pig population (n = 11,943 animals). In order to avoid any confounding issues with additive effects, pseudo‐records including only dominance deviations and residuals were simulated. SNP effects of heterozygous genotypes were summed up to generate true dominance deviations. We added random noise to those values and used them as phenotypes. Accuracy was defined as correlation between true and predicted dominance deviations. We conducted five replicates and estimated accuracies in three sets: between all (S1), non‐genotyped (S2) and inbred non‐genotyped (S3) animals. Potential bias was assessed by regressing true dominance deviations on predicted values. Matrices accounting for inbreeding (C3, C4 and C5) best fit. Accuracies were on average 0.77, 0.40 and 0.46 in S1, S2 and S3, respectively. In addition, C3, C4 and C5 scenarios have shown better accuracies than C1 and C2, and dominance deviations were less biased. Better matrix compatibility (accuracy and bias) was observed by re‐scaling diagonal elements to 1 minus the inbreeding coefficient (C5). [ABSTRACT FROM AUTHOR]

In the context of genetic research, but even more for routine genetic evaluations, a recurrent problem is the difficulty to obtain measurements for certain required phenotypes. This can be caused by limited facilities, difficulties in sampling, expensive analyses, limited staff relative to number of animals, and many other reasons. It may therefore be necessary to select which animals to phenotype while maintaining the phenotyping quality and avoiding bias or omitting some animals of interest. This implies developing strategies to do the phenotyping in a non-biased way using a priori available minimum knowledge. An example could be predicting post weaning growth of cattle based on earlier information as birth weight and weaning weight to optimize the use of feed intake or methane measurement facilities. In pigs, similar situations might occur for growth or emissions. However, another important example is boar taint (BT) which shows that not only sequential weighing, but also partial measurements of BT may require attention. BT is defined as urinary and fecal and associated with 2 major compounds, androstenone (AND) and skatole (SKA). The classical method of odor discrimination, "Human Nose Score" (HNS), is notoriously unreliable. Even if odor is known to be complex, adding AND and SKA to HNS would help reveal better the boar taint condition (i.e., phenotype). However, the measurements of AND and SKA or other compounds in fat are expensive. Therefore, it would be important to optimize these measurements by selecting relevant fat samples taken at slaughterhouses. We tested the following optimization strategy. The objective was to obtain a good distribution of a posteriori measured AND and SKA in the selected samples given initial asymmetric populations with positive HNS (3 to 10% declared positives). In a first step, the strategy pooled 2 available HNS for 997 across 22 contemporary groups (CG) and added limited chemical measurements (218/997) available for older samples. The data were obtained from a boar progeny testing scheme; therefore, CG were linked through connecting sires from artificial insemination having sons in different CG. A multi-trait BLUP model was used adjusting for metabolic weight using a 4-generation pedigree. A total of 15% of the animals in the last 6 contemporary groups were selected using first positive HNS (17/300), then keeping the highest and lowest predictions and maintaining diversity by not repeating boars from the same sire-family (i.e., when a family was already present, the next animal was taken). A posteriori chemical analyses were performed on selected samples to compare predicted and observed classifications in tainted and untainted. Research showed an accuracy of 0.73 (Table 1), leading to 58% tainted and 42% untainted compared with the expected ratio. Improvements of this strategy are currently in development, including using genomics. [ABSTRACT FROM AUTHOR]

Evaluating genetically heat stress (HS) of cattle is still a major challenge, also because of the need to have access to HS sensitive phenotypes relating their reaction to temperature and humidity as described by the temperature-humidity index (THI). A solution could be to add the information of data from sensors. However, activity (ACT) and rumination (RUM) sensors reflect no "production" [i.e., milk yields (MY), milk composition, but also intake or growth] related aspects which is suboptimal. Therefore, the objectives of this study were thus to assess the possible usefulness for adding sensors data for HS genetic evaluation based on a dairy situation. But the context described can be easily enlarged and include beef cattle and pigs. Strategy was illustrated using daily SenseHub collar records that were obtained from October 2019 to July 2022 in six dairy herds in Wallonia in Southern Belgium. A total of 453 Walloon Holstein cows were followed during this period for activity time. Meteorological data and a HS sensitive "production" trait (here MY) were also obtained from 2015 to 2022 for 1,740 cows from the same herds. In total 32,154 MY, 130,867 ACT and 130,848 RUM records were available. The thresholds at which the different traits start to be affected by HS were estimated at a THI of 63, 64 and 66, respectively, for MY, ACT and RUM. A three-trait random regression reaction norm model was fitted with these different thresholds. Estimated heritability values were 0.19 ± 0.03, 0.14 ± 0.06 and 0.21 ± 0.05 respectively for MY, ACT and RUM using AIREML. Ratios of variance associated to THI and constant genetic effect were 0.002, 0.02 and 0.04 indicating a stronger reaction by RUM but also ACT than MY to THI. However, it is uncertain if the reactions of these sensor traits are reflected in the production traits. Estimated genetic correlations of regressions on THI for ACT and RUM showed values of 0.51 ± 0.70 and 0.08 ± 0.54 with MY and 0.77 ± 0.69 between ACT and RUM. The first eigenvector of the correlation matrix explained 65% of the total variability and standardized coefficients of this eigenvector were 0.42, 0.69 and 0.59 for MY, ACT and RUM. A total of 50% of the variance associated with the regression on THI for MY could be explained by the regressions on THI for ACT and RUM. Adding the constant genetic effects for ACT and RUM only marginally increased to 51% the ability to explain the effect of THI on MY. Based on these results, reactions of different indicator traits, here sensor data, may reflect a different reaction than production and therefore economically important traits. Additional research, also beyond dairy cattle, should continue to address this topic that is very important for the validation of sensor data detecting HS. [ABSTRACT FROM AUTHOR]

Mid-infrared (MIR) spectroscopy is the method of choice for the standard milk recording system, to determine milk components including fat, protein, lactose and urea. Since milk composition is related to health and metabolic status of a cow, MIR spectra could be potentially used for disease detection. In dairy production, mastitis is one of the most prevalent diseases. The aim of this study was to develop a calibration equation to predict mastitis events from routinely recorded MIR spectra data. A further aim was to evaluate the use of test day somatic cell score (SCS) as covariate on the accuracy of the prediction model. The data for this study is from the Austrian milk recording system and its health monitoring system (GMON). Test day data including MIR spectra data was merged with diagnosis data of Fleckvieh, Brown Swiss and Holstein Friesian cows. As prediction variables, MIR absorbance data after first derivatives and selection of wavenumbers, corrected for days in milk, were used. The data set contained roughly 600,000 records and was split into calibration and validation sets by farm. Calibration sets were made to be balanced (as many healthy as mastitis cases), while the validation set was kept large and realistic. Prediction was done with Partial Least Squares Discriminant Analysis, key indicators of model fit were sensitivity and specificity. Results were extracted for association between spectra and diagnosis with different time windows (days between diagnosis and test days) in validation. The comparison of different sets of predictor variables (MIR, SCS, MIR + SCS) showed an advantage in prediction for MIR + SCS. For this prediction model, specificity was 0.79 and sensitivity was 0.68 in time window -7 to +7 days (calibration and validation). Corresponding values for MIR were 0.71 and 0.61, for SCS they were 0.81 and 0.62. In general, prediction of mastitis performed better with a shorter distance between test day and mastitis event, yet even for time windows of -21 to +21 days, prediction accuracies were still reasonable, with sensitivities ranging from 0.50 to 0.57 and specificities remaining unchanged (0.71 to 0.85). Additional research to further improve prediction equation, and studies on genetic correlations among clinical mastitis, SCS and MIR predicted mastitis are planned. [ABSTRACT FROM AUTHOR]