Validation strategy can result in an overoptimistic view of the ability of milk infrared spectra to predict methane emission of dairy cattle

Assessing different cross-validation schemes for predicting novel traits using sensor data: an application to dry matter intake and residual feed intake using milk spectral data

Feed efficiency is important for economic profitability of dairy farms; however, recording daily dry matter intakes (DMI) is expensive. Our objective was to investigate the potential use of milk mid-infrared (MIR) spectral data to predict proxy phenotypes for DMI based on different cross-validation schemes. We were specifically interested in comparisons between a model that included only MIR data (Model M1), a model that incorporated different energy sink predictors, such as body weight, body weight change, and milk energy (Model M2), and an extended model that incorporated both energy sinks and MIR data (Model M3). Models M2 and M3 also included various cow level variables (stage of lactation, age at calving, parity) such that any improvement in model performance from M2 to M3, whether through a smaller root mean squared error (RMSE) or a greater squared predictive correlation (R2), could indicate a potential benefit of MIR to predict residual feed intake. The data used in our study originated from a multi-institutional project on the genetics of feed efficiency in US Holsteins. Analyses were conducted on 2 different trait definitions based on different period lengths: averaged across weeks vs. averaged across 28-d. Specifically, there were 19,942 weekly records on 1,812 cows across 46 experiments or cohorts and 3,724 28-d records on 1,700 cows across 43 different cohorts. The cross-validation analyses involved 3 different k-fold schemes. First, a 10-fold cow-independent cross-validation was conducted whereby all records from any one cow were kept together in either training or test sets. Similarly, a 10-fold experiment-independent cross-validation kept entire experiments together whereas a 4-fold herd-independent cross-validation kept entire herds together in either training or test sets. Based on cow-independent cross-validation for both weekly and 28-d DMI, adding MIR predictors to energy sinks (Models M3 vs M2) significantly (P < 10-10) reduced average RMSE to 1.59 kg and increased average R2 to 0.89. However, adding MIR to energy sinks (M3) to predict DMI either within an experiment-independent or herd-independent cross-validation scheme seemed to demonstrate no merit (P > 0.05) compared with an energy sink model (M2) for either R2 or RMSE (respectively, 0.68 and 2.55 kg for M2 in herd-independent scheme). We further noted that with broader cross-validation schemes, i.e., from cow-independent to experiment-independent to herd-independent schemes, the mean and slope bias increased. Given that proxy DMI phenotypes for cows would need to be almost entirely generated in herds having no DMI or training data of their own, herd-independent cross-validation assessments of predictive performance should be emphasized. Hence, more research on predictive algorithms suitable for broader cross-validation schemes and a more earnest effort on calibration of spectrophotometers against each other should be considered.

Standardization method, testing scenario, and accuracy of the infrared prediction model affect the standardization accuracy of milk mid-infrared spectra

The widespread use of milk mid-infrared (MIR) spectroscopy for phenotype prediction has urged the application of prediction models across regions and countries. Spectra standardization is the most effective way to reduce the variability in the spectral signal provided by different instruments and labs. This study aimed to develop different standardization models for MIR spectra collected by multiple instruments, across 2 provinces of China, and investigate whether the standardization method (piecewise direct standardization, PDS, and direct standardization, DS), testing scenario (standardization of spectra collected on the same day or after 7 mo), infrared prediction model accuracy (high or low), and instrument (6 instruments from 2 brands) affect the performance of the standardization model. The results showed that the determination coefficient (R2) between absorbance values at each wavenumber provided by the primary and the secondary instruments increased from less than 0.90 to nearly 1.00 after standardization. Both PDS and DS successfully reduced spectra variation among instruments, and performed significantly better than non-standardization (P < 0.05). However, DS was more prone to overfitting than PDS. Standardization accuracy was higher when tested using spectra collected on the same time compared with those collected 7 mo after (P < 0.05), but great improvement in model transferability was obtained for both scenarios compared with the non-standardized spectra. The less accurate infrared prediction model (for C8:0 and C10:0 content) benefited the most (P < 0.05) from spectra standardization compared with the more accurate model (for total fat and protein content). For spectra collected after 7 mo from standardization, after PDS the RMSE between predictions obtained by different machines decreased on average by 86 and 94% compared with the values before standardization, for C8:0 and C10:0 respectively. The secondary instrument had no significant effect on the R2 between predictions (P > 0.05). The variation in the spectral signal provided by different instruments was successfully reduced by standardization across 2 provinces in China. This study lays the foundations for developing a national MIR spectra database to provide consistent predictions across provinces to be used in dairy farm management and breeding programs in China. Besides, this provides opportunities for data exchange and cooperation at international levels.

Prediction of likelihood of conception in dairy cows using milk mid-infrared spectra collected before the first insemination and machine learning algorithms

Accurate and ex-ante prediction of cows' likelihood of conception (LC) based on milk composition information could improve reproduction management on dairy farms. Milk composition is already routinely measured by mid-infrared (MIR) spectra, which are known to change with advancing stages of pregnancy. For lactating cows, MIR spectra may also be used for predicting the LC. Our objectives were to classify the LC at first insemination using milk MIR spectra data collected from calving to first insemination and to identify the spectral regions that contribute the most to the prediction of LC at first insemination. After quality control, 4,866 MIR spectra, milk production, and reproduction records from 3,451 Holstein cows were used. The classification accuracy and area under the curve (AUC) of 6 models comprising different predictors and 3 machine learning methods were estimated and compared. The results showed that partial least square discriminant analysis (PLS-DA) and random forest had higher prediction accuracies than logistic regression. The classification accuracy of good and poor LC cows and AUC in herd-by-herd validation of the best model were 76.35 ± 10.60% and 0.77 ± 0.11, respectively. All wavenumbers with values of variable importance in the projection higher than 1.00 in PLS-DA belonged to 3 spectral regions, namely from 1,003 to 1,189, 1,794 to 2,260, and 2,300 to 2,660 cm-1. In conclusion, the model can predict LC in dairy cows from a high productive TMR system before insemination with a relatively good accuracy, allowing farmers to intervene in advance or adjust the insemination schedule for cows with a poor predicted LC.

Lactation modeling and the effects of rotational crossbreeding on milk production traits and milk-spectra-predicted enteric methane emissions

Rotational crossbreeding has not been widely studied in relation to the enteric methane emissions of dairy cows, nor has the variation in emissions during lactation been modeled. Milk infrared spectra could be used to predict proxies of methane emissions in dairy cows. Therefore, the objective of this work was to study the effects of crossbreeding on the predicted infrared proxies of methane emissions and the variation in the latter during lactation. Milk samples were taken once from 1,059 cows reared in 2 herds, and infrared spectra of the milk were used to predict milk fat (mean ± SD; 3.79 ± 0.81%) and protein (3.68 ± 0.36%) concentrations, yield (21.4 ± 1.5 g/kg dry matter intake), methane intensity (14.2 ± 2.0 g/kg corrected milk), and daily methane production (358 ± 108 g/d). Of these cows, 620 were obtained from a 3-breed (Holstein, Montbéliarde, and Viking Red) rotational mating system, and the rest were purebred Holsteins. Milk production data and methane traits were analyzed using a nonlinear model that included the fixed effects of herd, genetic group, and parity, and the 4 parameters (a, b, c, and k) of a lactation curve modeled using the Wilmink function. Milk infrared spectra were found to be useful for direct prediction of qualitative proxies, such as methane yield and intensity, but not quantitative traits, such as daily methane production, which appears to be better estimated (450 ± 125 g/d) by multiplying a measured daily milk yield by infrared-predicted methane intensity. Lactation modeling of methane traits showed daily methane production to have a zenith curve, similar to that of milk yield but with a delayed peak (53 vs. 37 d in milk), whereas methane intensity is characterized by an upward curve that increases rapidly during the first third of lactation and then slowly till the end of lactation (10.5 g/kg at 1 d in milk to 15.2 g/kg at 300 d in milk). However, lactation modeling was not useful in explaining methane yield, which is almost constant during lactation. Lastly, the methane yield and intensity of cows from 3-breed rotational crossbreeding are not greater, and their methane production is lower than that of purebred Holsteins (452 vs. 477 g/d). Given the greater longevity of crossbred cows, and their lower replacement rate, rotational crossbreeding could be a way of mitigating the environmental impact of milk production.

Exploring the use of NIR and Raman spectroscopy for the prediction of quality traits in PDO cheeses

The aims of this proof of principle study were to compare two different chemometric approaches using a Bayesian method, Partial Least Square (PLS) and PLS-discriminant analysis (DA), for the prediction of the chemical composition and texture properties of the Grana Padano (GP) and Parmigiano Reggiano (PR) PDO cheeses by using NIR and Raman spectra and quantify their ability to distinguish between the two PDO and among their ripening periods. For each dairy chain consortium, 9 cheese samples from 3 dairy industries were collected for a total of 18 cheese samples. Three seasoning times were chosen for each dairy industry: 12, 20, and 36 months for GP and 12, 24, and 36 months for PR. A portable NIR instrument (spectral range: 950-1,650 nm) was used on 3 selected spots on the paste of each cheese sample, for a total of 54 spectra collected. An Alpha300 R confocal Raman microscope was used to collect 10 individual spectra for each cheese sample in each spot for a total of 540 Raman spectra collected. After the detection of eventual outliers, the spectra were also concatenated together (NIR + Raman). All the cheese samples were assessed in terms of chemical composition and texture properties following the official reference methods. A Bayesian approach and PLS-DA were applied to the NIR, Raman, and fused spectra to predict the PDO type and seasoning time. The PLS-DA reached the best performances, with 100% correctly identified PDO type using Raman only. The fusion of the data improved the results in 60% of the cases with the Bayesian and of 40% with the PLS-DA approach. A Bayesian approach and a PLS procedure were applied to the NIR, Raman, and fused spectra to predict the chemical composition of the cheese samples and their texture properties. In this case, the best performance in validation was reached with the Bayesian method on Raman spectra for fat (R2VAL = 0.74). The fusion of the data was not always helpful in improving the prediction accuracy. Given the limitations associated with our sample set, future studies will expand the sample size and incorporate diverse PDO cheeses.

Predicting methane emissions of individual grazing dairy cows from spectral analyses of their milk samples

Data on the enteric methane emissions of individual cows are useful not just in assisting management decisions and calculating herd inventories but also as inputs for animal genetic evaluations. Data generation for many animal characteristics, including enteric methane emissions, can be expensive and time consuming, so being able to extract as much information as possible from available samples or data sources is worthy of investigation. The objective of the present study was to attempt to predict individual cow methane emissions from the information contained within milk samples, specifically the spectrum of light transmittance across different wavelengths of the mid-infrared (MIR) region of the electromagnetic spectrum. A total of 93,888 individual spot measures of methane (i.e., individual samples of an animal's breath when using the GreenFeed technology) from 384 lactations on 277 grazing dairy cows were collapsed into weekly averages expressed as grams per day; each weekly average coincided with a MIR spectral analysis of a morning or evening individual cow milk sample. Associations between the spectra and enteric methane measures were performed separately using partial least squares regression or neural networks with different tuning parameters evaluated. Several alternative definitions of the enteric methane phenotype (i.e., average enteric methane in the 6 d preceding or 6 d following taking the milk sample or the average of the 6 d before and after the milk sample, all of which also included the enteric methane emitted on the day of milk sampling), the candidate model features (e.g., milk yield, milk composition, and milk MIR) as well as validation strategy (i.e., cross-validation or leave-one-experimental treatment-out) were evaluated. Irrespective of the validation method, the prediction accuracy was best when the average of the milk MIR from the morning and evening milk sample was used and the prediction model was developed using neural networks; concurrently including milk yield and days in milk in the prediction model generated superior predictions relative to just the spectral information alone. Furthermore, prediction accuracy was best when the enteric methane phenotype was the average of at least 20 methane spot measures across a 6-d period flanking each side of the milk sample with associated spectral data. Based on the strategy that achieved the best accuracy of prediction, the correlation between the actual and predicted daily methane emissions when based on 4-fold cross-validation varied per validation stratum from 0.68 to 0.75; the corresponding range when validated on each of the 8 different experimental treatments focusing on alternative pasture grazing systems represented in the dataset varied from 0.55 to 0.71. The root mean square error of prediction across the 4-folds of cross-validation was 37.46 g/d, whereas the root mean square error averaged across all folds of leave-one-treatment-out was 37.50 g/d. Results suggest that even with the likely measurement errors contained within the MIR spectrum and gold standard enteric methane phenotype, enteric methane can be reasonably well predicted from the infrared spectrum of milk samples. What is yet to be established, however, is whether (a) genetic variation exists in this predicted enteric methane phenotype and (b) selection on estimates of genetic merit for this phenotype translate to actual phenotypic differences in enteric methane emissions.

Genetic parameters for feed intake and body weight in dairy cattle using high-throughput 3-dimensional cameras in Danish commercial farms

Recording complex phenotypes on a large scale is becoming possible with the incorporation of recently developed new technologies. One of these new technologies is the use of 3-dimensional (3D) cameras on commercial farms to measure feed intake and body weight (BW) daily. Residual feed intake (RFI) has been proposed as a proxy for feed efficiency in several species, including cattle, pigs, and poultry. Dry matter intake (DMI) and BW records are required to calculate RFI, and the use of this new technology will help increase the number of individual records more efficiently. The aim of this study was to estimate genetic parameters (including genetic correlations) for DMI and BW obtained by 3D cameras from 6,000 cows in commercial farms from the breeds Danish Holstein, Jersey, and Nordic Red. Additionally, heritabilities per parity and genetic correlations among parities were estimated for DMI and BW in the 3 breeds. Data included 158,000 weekly records of DMI and BW obtained between 2019 and 2022 on 17 commercial farms. Estimated heritability for DMI ranged from 0.17 to 0.25, whereas for BW they ranged from 0.44 to 0.58. The genetic correlations between DMI and BW were moderately positive (0.58-0.65). Genetic correlations among parities in both traits were highly correlated in the 3 breeds, except for DMI between first parity and late parities in Holstein where they were down to 0.62. Based on these results, we conclude that DMI and BW phenotypes measured by 3D cameras are heritable for the 3 dairy breeds and their heritabilities are comparable to those obtained by traditional methods (scales and feed bins). The high heritabilities and correlations of 3D measurements with the true trait in previous studies demonstrate the potential of this new technology for measuring feed intake and BW in real time. In conclusion, 3D camera technology has the potential to become a valuable tool for automatic and continuous recording of feed intake and BW on commercial farms.

Cheese yield and nutrients recovery in the curd predicted by Fourier-transform spectra from individual sheep milk samples

The objectives of this study were to explore the use of Fourier-transform infrared (FTIR) spectroscopy on individual sheep milk samples for predicting cheese-making traits, and to test the effect of the farm variability on their prediction accuracy. For each of 121 ewes from 4 farms, a laboratory model cheese was produced, and 3 actual cheese yield traits (fresh cheese, cheese solids, and cheese water) and 4 milk nutrient recovery traits (fat, protein, total solids, and energy) in the curd were measured. Calibration equations were developed using a Bayesian approach with 2 different scenarios: (1) a random cross-validation (80% calibration; 20% validation set), and (2) a leave-one-out validation (3 farms used as calibration, and the remaining one as validation set) to assess the accuracy of prediction of samples from external farms, not included in calibration set. The best performance was obtained for predicting the yield and recovery of total solids, justifying for the practical application of the method at sheep population and dairy industry levels. Performances for the remaining traits were lower, but still useful for the monitoring of the milk processing in the case of fresh curd and recovery of energy. Insufficient accuracies were found for the recovery of protein and fat, highlighting the complex nature of the relationships among the milk nutrients and their recovery in the curd. The leave-one-out validation procedure, as expected, showed lower prediction accuracies, as a result of the characteristics of the farming systems, which were different between calibration and validation sets. In this regard, the inclusion of information related to the farm could help to improve the prediction accuracy of these traits. Overall, a large contribution to the prediction of the cheese-making traits came from the areas known as "water" and "fingerprint" regions. These findings suggest that, according to the traits studied, the inclusion of water regions for the development of the prediction equation models is fundamental to maintain a high prediction accuracy. However, further studies are necessary to better understand the role of specific absorbance peaks and their contribution to the prediction of cheese-making traits, to offer reliable tools applicable along the dairy ovine chain.

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.

Prediction of detailed blood metabolic profile using milk infrared spectra and machine learning methods in dairy cattle

The adoption of preventive management decisions is crucial to dealing with metabolic impairments in dairy cattle. Various serum metabolites are known to be useful indicators of the health status of cows. In this study, we used milk Fourier-transform mid-infrared (FTIR) spectra and various machine learning (ML) algorithms to develop prediction equations for a panel of 29 blood metabolites, including those related to energy metabolism, liver function/hepatic damage, oxidative stress, inflammation/innate immunity, and minerals. For most traits, the data set comprised observations from 1,204 Holstein-Friesian dairy cows belonging to 5 herds. An exception was represented by β-hydroxybutyrate prediction, which contained observations from 2,701 multibreed cows pertaining to 33 herds. The best predictive model was developed using an automatic ML algorithm that tested various methods, including elastic net, distributed random forest, gradient boosting machine, artificial neural network, and stacking ensemble. These ML predictions were compared with partial least squares regression, the most commonly used method for FTIR prediction of blood traits. Performance of each model was evaluated using 2 cross-validation (CV) scenarios: 5-fold random (CV_r) and herd-out (CV_h). We also tested the best model's ability to classify values precisely in the 2 extreme tails, namely, the 25th (Q25) and 75th (Q75) percentiles (true-positive prediction scenario). Compared with partial least squares regression, ML algorithms achieved more accurate performance. Specifically, elastic net increased the R² value from 5% to 75% for CV_r and 2% to 139% for CV_h, whereas the stacking ensemble increased the R² value from 4% to 70% for CV_r and 4% to 150% for CV_h. Considering the best model, with the CV_r scenario, good prediction accuracies were obtained for glucose (R² = 0.81), urea (R² = 0.73), albumin (R² = 0.75), total reactive oxygen metabolites (R² = 0.79), total thiol groups (R² = 0.76), ceruloplasmin (R² = 0.74), total proteins (R² = 0.81), globulins (R² = 0.87), and Na (R² = 0.72). Good prediction accuracy in classifying extreme values was achieved for glucose (Q25 = 70.8%, Q75 = 69.9%), albumin (Q25 = 72.3%), total reactive oxygen metabolites (Q25 = 75.1%, Q75 = 74%), thiol groups (Q75 = 70.4%), total proteins (Q25 = 72.4%, Q75 = 77.2.%), globulins (Q25 = 74.8%, Q75 = 81.5%), and haptoglobin (Q75 = 74.4%). In conclusion, our study shows that FTIR spectra can be used to predict blood metabolites with relatively good accuracy, depending on trait, and are a promising tool for large-scale monitoring.

Integrating on-farm and genomic information improves the predictive ability of milk infrared prediction of blood indicators of metabolic disorders in dairy cows

BACKGROUND

Blood metabolic profiles can be used to assess metabolic disorders and to evaluate the health status of dairy cows. Given that these analyses are time-consuming, expensive, and stressful for the cows, there has been increased interest in Fourier transform infrared (FTIR) spectroscopy of milk samples as a rapid, cost-effective alternative for predicting metabolic disturbances. The integration of FTIR data with other layers of information such as genomic and on-farm data (days in milk (DIM) and parity) has been proposed to further enhance the predictive ability of statistical methods. Here, we developed a phenotype prediction approach for a panel of blood metabolites based on a combination of milk FTIR data, on-farm data, and genomic information recorded on 1150 Holstein cows, using BayesB and gradient boosting machine (GBM) models, with tenfold, batch-out and herd-out cross-validation (CV) scenarios.

RESULTS

The predictive ability of these approaches was measured by the coefficient of determination (R²). The results show that, compared to the model that includes only FTIR data, integration of both on-farm (DIM and parity) and genomic information with FTIR data improves the R² for blood metabolites across the three CV scenarios, especially with the herd-out CV: R² values ranged from 5.9 to 17.8% for BayesB, from 8.2 to 16.9% for GBM with the tenfold random CV, from 3.8 to 13.5% for BayesB and from 8.6 to 17.5% for GBM with the batch-out CV, and from 8.4 to 23.0% for BayesB and from 8.1 to 23.8% for GBM with the herd-out CV. Overall, with the model that includes the three sources of data, GBM was more accurate than BayesB with accuracies across the CV scenarios increasing by 7.1% for energy-related metabolites, 10.7% for liver function/hepatic damage, 9.6% for oxidative stress, 6.1% for inflammation/innate immunity, and 11.4% for mineral indicators.

CONCLUSIONS

Our results show that, compared to using only milk FTIR data, a model integrating milk FTIR spectra with on-farm and genomic information improves the prediction of blood metabolic traits in Holstein cattle and that GBM is more accurate in predicting blood metabolites than BayesB, especially for the batch-out CV and herd-out CV scenarios.

The Combined Use of Automated Milking System and Sensor Data to Improve Detection of Mild Lameness in Dairy Cattle

This study aimed to develop a tool to detect mildly lame cows by combining already existing data from sensors, AMSs, and routinely recorded animal and farm data. For this purpose, ten dairy farms were visited every 30–42 days from January 2020 to May 2021. Locomotion scores (LCS, from one for nonlame to five for severely lame) and body condition scores (BCS) were assessed at each visit, resulting in a total of 594 recorded animals. A questionnaire about farm management and husbandry was completed for the inclusion of potential risk factors. A lameness incidence risk (LCS ≥ 2) was calculated and varied widely between farms with a range from 27.07 to 65.52%. Moreover, the impact of lameness on the derived sensor parameters was inspected and showed no significant impact of lameness on total rumination time. Behavioral patterns for eating, low activity, and medium activity differed significantly in lame cows compared to nonlame cows. Finally, random forest models for lameness detection were fit by including different combinations of influencing variables. The results of these models were compared according to accuracy, sensitivity, and specificity. The best performing model achieved an accuracy of 0.75 with a sensitivity of 0.72 and specificity of 0.78. These approaches with routinely available data and sensor data can deliver promising results for early lameness detection in dairy cattle. While experimental automated lameness detection systems have achieved improved predictive results, the benefit of this presented approach is that it uses results from existing, routinely recorded, and therefore widely available data.

Establishment and Validation of Fourier Transform Infrared Spectroscopy (FT–MIR) Methodology for the Detection of Linoleic Acid in Buffalo Milk

Buffalo milk is a dairy product that is considered to have a higher nutritional value compared to cow’s milk. Linoleic acid (LA) is an essential fatty acid that is important for human health. This study aimed to investigate and validate the use of Fourier transform mid-infrared spectroscopy (FT-MIR) for the quantification of the linoleic acid in buffalo milk. Three machine learning models were used to predict linoleic acid content, and random forest was employed to select the most important subset of spectra for improved model performance. The validity of the FT-MIR methods was evaluated in accordance with ICH Q2 (R1) guidelines using the accuracy profile method, and the precision, the accuracy, and the limit of quantification were determined. The results showed that Fourier transform infrared spectroscopy is a suitable technique for the analysis of linoleic acid, with a lower limit of quantification of 0.15 mg/mL milk. Our results showed that FT-MIR spectroscopy is a viable method for LA concentration analysis.

Phenotypic Analysis of Fourier-Transform Infrared Milk Spectra in Dairy Goats

The infrared spectrum of bovine milk is used to predict many interesting traits, whereas there have been few studies on goat milk in this regard. The objective of this study was to characterize the major sources of variation in the absorbance of the infrared spectrum in caprine milk samples. A total of 657 goats belonging to 6 breeds and reared on 20 farms under traditional and modern dairy systems were milk-sampled once. Fourier-transform infrared (FTIR) spectra were taken (2 replicates per sample, 1314 spectra), and each spectrum contained absorbance values at 1060 different wavenumbers (5000 to 930 × cm^-1), which were treated as a response variable and analyzed one at a time (i.e., 1060 runs). A mixed model, including the random effects of sample/goat, breed, flock, parity, stage of lactation, and the residual, was used. The pattern and variability of the FTIR spectrum of caprine milk was similar to those of bovine milk. The major sources of variation in the entire spectrum were as follows: sample/goat (33% of the total variance); flock (21%); breed (15%); lactation stage (11%); parity (9%); and the residual unexplained variation (10%). The entire spectrum was segmented into five relatively homogeneous regions. Two of them exhibited very large variations, especially the residual variation. These regions are known to be affected by the absorbance of water, although they also exhibited wide variations in the other sources of variation. The average repeatability of these two regions were 45% and 75%, whereas for the other three regions it was about 99%. The FTIR spectrum of caprine milk could probably be used to predict several traits and to authenticate the origin of goat milk.

Prediction of Liver Triglyceride Content in Early Lactation Multiparous Holstein Cows Using Blood Metabolite, Mineral, and Protein Biomarker Concentrations

Bovine fatty liver syndrome (bFLS) is difficult to diagnose because a liver tissue biopsy is required to assess liver triglyceride (TG) content. We hypothesized that a blood biomarker panel could be a convenient alternative method of liver TG content assessment and bFLS diagnosis. Our objectives were to predict liver TG using blood biomarker concentrations across days in milk (DIM; longitudinal, LT) or at a single timepoint (ST; 3, 7, or 14 DIM), as well as different biomarker combination based on their perceived accessibility. Data from two separate experiments (n = 65 cows) was used for model training and validation. Response variables were based on the maximum liver TG observed in 1 and 14 DIM liver biopsies: Max TG (continuous), Low TG (TG > 13.3% dry matter; DM), Median TG (TG > 17.1% DM), and High TG (TG > 22.0% DM). Model performance varied but High TG was well predicted by sparse partial least squares—discriminate analysis models using LT and ST data, achieving balanced error rates ≤ 15.4% for several model variations during cross-validation. In conclusion, blood biomarker panels using 7 DIM, 14 DIM, or LT data may be a useful diagnostic tool for bFLS in research and field settings.

In-line near-infrared analysis of milk coupled with machine learning methods for the daily prediction of blood metabolic profile in dairy cattle

Precision livestock farming technologies are used to monitor animal health and welfare parameters continuously and in real time in order to optimize nutrition and productivity and to detect health issues at an early stage. The possibility of predicting blood metabolites from milk samples obtained during routine milking by means of infrared spectroscopy has become increasingly attractive. We developed, for the first time, prediction equations for a set of blood metabolites using diverse machine learning methods and milk near-infrared spectra collected by the AfiLab instrument. Our dataset was obtained from 385 Holstein Friesian dairy cows. Stacking ensemble and multi-layer feedforward artificial neural network outperformed the other machine learning methods tested, with a reduction in the root mean square error of between 3 and 6% in most blood parameters. We obtained moderate correlations (r) between the observed and predicted phenotypes for γ-glutamyl transferase (r = 0.58), alkaline phosphatase (0.54), haptoglobin (0.66), globulins (0.61), total reactive oxygen metabolites (0.60) and thiol groups (0.57). The AfiLab instrument has strong potential but may not yet be ready to predict the metabolic stress of dairy cows in practice. Further research is needed to find out methods that allow an improvement in accuracy of prediction equations.

Pregnancy status predicted using milk mid-infrared spectra from dairy cattle

Accurate and timely pregnancy diagnosis is an important component of effective herd management in dairy cattle. Predicting pregnancy from Fourier-transform mid-infrared (FT-MIR) spectroscopy data is of particular interest because the data are often already available from routine milk testing. The purpose of this study was to evaluate how well pregnancy status could be predicted in a large data set of 1,161,436 FT-MIR milk spectra records from 863,982 mixed-breed pasture-based New Zealand dairy cattle managed within seasonal calving systems. Three strategies were assessed for defining the nonpregnant cows when partitioning the records according to pregnancy status in the training population. Two of these used records for cows with a subsequent calving only, whereas the third also included records for cows without a subsequent calving. For each partitioning strategy, partial least squares discriminant analysis models were developed, whereby spectra from all the cows in 80% of herds were used to train the models, and predictions on cows in the remaining herds were used for validation. A separate data set was also used as a secondary validation, whereby pregnancy diagnosis had been assigned according to the presence of pregnancy-associated glycoproteins (PAG) in the milk samples. We examined different ways of accounting for stage of lactation in the prediction models, either by including it as an effect in the prediction model, or by pre-adjusting spectra before fitting the model. For a subset of strategies, we also assessed prediction accuracies from deep learning approaches, utilizing either the raw spectra or images of spectra. Across all strategies, prediction accuracies were highest for models using the unadjusted spectra as model predictors. Strategies for cows with a subsequent calving performed well in herd-independent validation with sensitivities above 0.79, specificities above 0.91 and area under the receiver operating characteristic curve (AUC) values over 0.91. However, for these strategies, the specificity to predict nonpregnant cows in the external PAG data set was poor (0.002-0.04). The best performing models were those that included records for cows without a subsequent calving, and used unadjusted spectra and days in milk as predictors, with consistent results observed across the training, herd-independent validation and PAG data sets. For the partial least squares discriminant analysis model, sensitivity was 0.71, specificity was 0.54 and AUC values were 0.68 in the PAG data set; and for an image-based deep learning model, the sensitivity was 0.74, specificity was 0.52 and the AUC value was 0.69. Our results demonstrate that in pasture-based seasonal calving herds, confounding between pregnancy status and spectral changes associated with stage of lactation can inflate prediction accuracies. When the effect of this confounding was reduced, prediction accuracies were not sufficiently high enough to use as a sole indicator of pregnancy status.

Integrating diverse data sources to predict disease risk in dairy cattle-a machine learning approach

Livestock farming is currently undergoing a digital revolution and becoming increasingly data-driven. Yet, such data often reside in disconnected silos making them impossible to leverage their full potential to improve animal well-being. Here, we introduce a precision livestock farming approach, bringing together information streams from a variety of life domains of dairy cattle to study whether including more and diverse data sources improves the quality of predictions for eight diseases and whether using more complex prediction algorithms can, to some extent, compensate for less diverse data. Using three machine learning approaches of varying complexity (from logistic regression to gradient boosted trees) trained on data from 5,828 animals in 165 herds in Austria, we show that the prediction of lameness, acute and chronic mastitis, anestrus, ovarian cysts, metritis, ketosis (hyperketonemia), and periparturient hypocalcemia (milk fever) from routinely available data gives encouraging results. For example, we can predict lameness with high sensitivity and specificity (F1 = 0.74). An analysis of the importance of individual variables to prediction performance shows that disease in dairy cattle is a product of the complex interplay between a multitude of life domains, such as housing, nutrition, or climate, that including more and diverse data sources increases prediction performance, and that the reuse of existing data can create actionable information for preventive interventions. Our findings pave the way toward data-driven point-of-care interventions and demonstrate the added value of integrating all available data in the dairy industry to improve animal well-being and reduce disease risk.

Genetic parameters of differential somatic cell count, milk composition, and cheese-making traits measured and predicted using spectral data in Holstein cows

Mastitis is one of the most prevalent diseases in dairy cattle and is the cause of considerable economic losses. Alongside somatic cell count (SCC), differential somatic cell count (DSCC) has been recently introduced as a new indicator of intramammary infection. The DSCC is expressed as a count or a proportion (%) of polymorphonuclear neutrophils plus lymphocytes (PMN-LYM) in milk somatic cells. These numbers are complemented to total somatic cell count or to 100 by macrophages (MAC). The aim of this study was to investigate the genetic variation and heritability of DSCC, and its correlation with milk composition, udder health indicators, milk composition, and technological traits in Holstein cattle. Data used in the analysis consisted in single test-day records from 2,488 Holstein cows reared in 36 herds located in northern Italy. Fourier-transform infrared (FTIR) spectroscopy was used to predict missing information for some milk coagulation and cheese-making traits, to increase sample size and improve estimation of the genetic parameters. Bayesian animal models were implemented via Gibbs sampling. Marginal posterior means of the heritability estimates were 0.13 for somatic cell score (SCS); 0.11 for DSCC, MAC proportion, and MAC count; and 0.10 for PMN-LYM count. Posterior means of additive genetic correlations between SCS and milk composition and udder health were low to moderate and unfavorable. All the relevant genetic correlations between the SCC traits considered and the milk traits (composition, coagulation, cheese yield and nutrients recovery) were unfavorable. The SCS showed genetic correlations of -0.30 with the milk protein proportion, -0.56 with the lactose proportion and -0.52 with the casein index. In the case of milk technological traits, SCS showed genetic correlations of 0.38 with curd firming rate (k₂₀), 0.45 with rennet coagulation time estimated using the curd firming over time equation (RCT_eq), -0.39 with asymptotic potential curd firmness, -0.26 with maximum curd firmness (CF_max), and of -0.31 with protein recovery in the curd. Differential somatic cell count expressed as proportion was correlated with SCS (0.60) but had only 2 moderate genetic correlations with milk traits: with lactose (-0.32) and CF_max (-0.33). The SCS was highly correlated with the log PMN-LYM count (0.79) and with the log MAC count (0.69). The 2 latter traits were correlated with several milk traits: fat (-0.38 and -0.43 with PMN-LYM and MAC counts, respectively), lactose percentage (-0.40 and -0.46), RCT_eq (0.53 and 0.41), t_max (0.38 and 0.48). Log MAC count was correlated with k₂₀ (+0.34), and log PMN-LYM count was correlated with CF_max (-0.26) and weight of water curd as percentage of weight of milk processed (-0.26). The results obtained offer new insights into the relationships between the indicators of udder health and the milk technological traits in Holstein cows.

Disentangling data dependency using cross-validation strategies to evaluate prediction quality of cattle grazing activities using machine learning algorithms and wearable sensor data

Wearable sensors have been explored as an alternative for real-time monitoring of cattle feeding behavior in grazing systems. To evaluate the performance of predictive models such as machine learning (ML) techniques, data cross-validation (CV) approaches are often employed. However, due to data dependencies and confounding effects, poorly performed validation strategies may significantly inflate the prediction quality. In this context, our objective was to evaluate the effect of different CV strategies on the prediction of grazing activities in cattle using wearable sensor (accelerometer) data and ML algorithms. Six Nellore bulls (average live weight of 345 ± 21 kg) had their behavior visually classified as grazing or not-grazing for a period of 15 days. Elastic Net Generalized Linear Model (GLM), Random Forest (RF), and Artificial Neural Network (ANN) were employed to predict grazing activity (grazing or not-grazing) using 3-axis accelerometer data. For each analytical method, three CV strategies were evaluated: holdout, leave-one-animal-out (LOAO), and leave-one-day-out (LODO). Algorithms were trained using similar dataset sizes (holdout: n = 57,862; LOAO: n = 56,786; LODO: n = 56,672). Overall, GLM delivered the worst prediction accuracy (53%) compared to the ML techniques (65% for both RF and ANN), and ANN performed slightly better than RF for LOAO (73%) and LODO (64%) across CV strategies. The holdout yielded the highest nominal accuracy values for all three ML approaches (GLM: 59%, RF: 76%, and ANN: 74%), followed by LODO (GLM: 49%, RF: 61%, and ANN: 63%) and LOAO (GLM: 52%, RF: 57%, and ANN: 57%). With a larger dataset (i.e., more animals and grazing management scenarios), it is expected that accuracy could be increased. Most importantly, the greater prediction accuracy observed for holdout CV may simply indicate a lack of data independence and the presence of carry-over effects from animals and grazing management. Our results suggest that generalizing predictive models to unknown (not used for training) animals or grazing management may incur poor prediction quality. The results highlight the need for using management knowledge to define the validation strategy that is closer to the real-life situation, i.e., the intended application of the predictive model.

Comparison of Single-Breed and Multi-Breed Training Populations for Infrared Predictions of Novel Phenotypes in Holstein Cows

In general, Fourier-transform infrared (FTIR) predictions are developed using a single-breed population split into a training and a validation set. However, using populations formed of different breeds is an attractive way to design cross-validation scenarios aimed at increasing prediction for difficult-to-measure traits in the dairy industry. This study aimed to evaluate the potential of FTIR prediction using training set combining specialized and dual-purpose dairy breeds to predict different phenotypes divergent in terms of biological meaning, variability, and heritability, such as body condition score (BCS), serum β-hydroxybutyrate (BHB), and kappa casein (k-CN) in the major cattle breed, i.e., Holstein-Friesian. Data were obtained from specialized dairy breeds: Holstein (468 cows) and Brown Swiss (657 cows), and dual-purpose breeds: Simmental (157 cows), Alpine Grey (75 cows), and Rendena (104 cows), giving a total of 1461 cows from 41 multi-breed dairy herds. The FTIR prediction model was developed using a gradient boosting machine (GBM), and predictive ability for the target phenotype in Holstein cows was assessed using different cross-validation (CV) strategies: a within-breed scenario using 10-fold cross-validation, for which the Holstein population was randomly split into 10 folds, one for validation and the remaining nine for training (10-fold_HO); an across-breed scenario (BS_HO) where the Brown Swiss cows were used as the training set and the Holstein cows as the validation set; a specialized multi-breed scenario (BS+HO_10-fold), where the entire Brown Swiss and Holstein populations were combined then split into 10 folds, and a multi-breed scenario (Multi-breed), where the training set comprised specialized (Holstein and Brown Swiss) and dual-purpose (Simmental, Alpine Grey, and Rendena) dairy cows, combined with nine folds of the Holstein cows. Lastly a Multi-breed CV2 scenario was implemented, assuming the same number of records as the reference scenario and using the same proportions as the multi-breed. Within-Holstein, FTIR predictions had a predictive ability of 0.63 for BCS, 0.81 for BHB, and 0.80 for k-CN. Using a specific breed (Brown Swiss) as the training set for prediction in the Holstein population reduced the prediction accuracy by 10% for BCS, 7% for BHB, and 11% for k-CN. Notably, the combination of Holstein and Brown Swiss cows in the training set increased the predictive ability of the model by 6%, which was 0.66 for BCS, 0.85 for BHB, and 0.87 for k-CN. Using multiple specialized and dual-purpose animals in the training set outperforms the 10-fold_HO (standard) approach, with an increase in predictive ability of 8% for BCS, 7% for BHB, and 10% for k-CN. When the Multi-breed CV2 was implemented, no improvement was observed. Our findings suggest that FTIR prediction of different phenotypes in the Holstein breed can be improved by including different specialized and dual-purpose breeds in the training population. Our study also shows that predictive ability is enhanced when the size of the training population and the phenotypic variability are increased.

Evaluating the performance of machine learning methods and variable selection methods for predicting difficult-to-measure traits in Holstein dairy cattle using milk infrared spectral data

Fourier-transform infrared (FTIR) spectroscopy is a powerful high-throughput phenotyping tool for predicting traits that are expensive and difficult to measure in dairy cattle. Calibration equations are often developed using standard methods, such as partial least squares (PLS) regression. Methods that employ penalization, rank-reduction, and variable selection, as well as being able to model the nonlinear relations between phenotype and FTIR, might offer improvements in predictive ability and model robustness. This study aimed to compare the predictive ability of 2 machine learning methods, namely random forest (RF) and gradient boosting machine (GBM), and penalized regression against PLS regression for predicting 3 phenotypes differing in terms of biological meaning and relationships with milk composition (i.e., phenotypes measurable directly and not directly in milk, reflecting different biological processes which can be captured using milk spectra) in Holstein-Friesian cattle under 2 cross-validation scenarios. The data set comprised phenotypic information from 471 Holstein-Friesian cows, and 3 target phenotypes were evaluated: (1) body condition score (BCS), (2) blood β-hydroxybutyrate (BHB, mmol/L), and (3) κ-casein expressed as a percentage of nitrogen (κ-CN, % N). The data set was split considering 2 cross-validation scenarios: samples-out random in which the population was randomly split into 10-folds (8-folds for training and 1-fold for validation and testing); and herd/date-out in which the population was randomly assigned to training (70% herd), validation (10%), and testing (20% herd) based on the herd and date in which the samples were collected. The random grid search was performed using the training subset for the hyperparameter optimization and the validation set was used for the generalization of prediction error. The trained model was then used to assess the final prediction in the testing subset. The grid search for penalized regression evidenced that the elastic net (EN) was the best regularization with increase in predictive ability of 5%. The performance of PLS (standard model) was compared against 2 machine learning techniques and penalized regression using 2 cross-validation scenarios. Machine learning methods showed a greater predictive ability for BCS (0.63 for GBM and 0.61 for RF), BHB (0.80 for GBM and 0.79 for RF), and κ-CN (0.81 for GBM and 0.80 for RF) in samples-out cross-validation. Considering a herd/date-out cross-validation these values were 0.58 (GBM and RF) for BCS, 0.73 (GBM and RF) for BHB, and 0.77 (GBM and RF) for κ-CN. The GBM model tended to outperform other methods in predictive ability around 4%, 1%, and 7% for EN, RF, and PLS, respectively. The prediction accuracies of the GBM and RF models were similar, and differed statistically from the PLS model in samples-out random cross-validation. Although, machine learning techniques outperformed PLS in herd/date-out cross-validation, no significant differences were observed in terms of predictive ability due to the large standard deviation observed for predictions. Overall, GBM achieved the highest accuracy of FTIR-based prediction of the different phenotypic traits across the cross-validation scenarios. These results indicate that GBM is a promising method for obtaining more accurate FTIR-based predictions for different phenotypes in dairy cattle.

Performance of different portable and hand-held near-infrared spectrometers for predicting beef composition and quality characteristics in the abattoir without meat sampling

The availability of portable and handheld NIR instruments on the market opens up new possibilities in meat analysis. However, there is lack of research comparing different NIR instruments for evaluating beef characteristics from spectra obtained directly on the meat surface. Our aim, therefore, was to build and test calibration and prediction models for predicting beef characteristics, and to compare the performances of three NIR instruments differing in size and characteristics: a transportable visible-NIR spectrometer (Vis-NIRS), a portable (NIRS), and a hand-held Micro-NIRS. Spectra were collected from 178 beef samples (Longissimus thoracis muscle) from the meat surface in the abattoir. The spectra were subjected to different mathematical pretreatments then partial least square regressions. The results showed that all instruments predicted dry matter, protein and lipids with R²_VAL 0.23 to 0.70; pH and cooking loss R²_VAL 0.19 to 0.25; and color R²_VAL 0.35 to 0.77. Overall, the prediction performances of the three instruments were similar, although Micro-NIRS performed better in some respects.

Integrating genomic and infrared spectral data improves the prediction of milk protein composition in dairy cattle

Background

Over the past decade, Fourier transform infrared (FTIR) spectroscopy has been used to predict novel milk protein phenotypes. Genomic data might help predict these phenotypes when integrated with milk FTIR spectra. The objective of this study was to investigate prediction accuracy for milk protein phenotypes when heterogeneous on-farm, genomic, and pedigree data were integrated with the spectra. To this end, we used the records of 966 Italian Brown Swiss cows with milk FTIR spectra, on-farm information, medium-density genetic markers, and pedigree data. True and total whey protein, and five casein, and two whey protein traits were analyzed. Multiple kernel learning constructed from spectral and genomic (pedigree) relationship matrices and multilayer BayesB assigning separate priors for FTIR and markers were benchmarked against a baseline partial least squares (PLS) regression. Seven combinations of covariates were considered, and their predictive abilities were evaluated by repeated random sub-sampling and herd cross-validations (CV).

Results

Addition of the on-farm effects such as herd, days in milk, and parity to spectral data improved predictions as compared to those obtained using the spectra alone. Integrating genomics and/or the top three markers with a large effect further enhanced the predictions. Pedigree data also improved prediction, but to a lesser extent than genomic data. Multiple kernel learning and multilayer BayesB increased predictive performance, whereas PLS did not. Overall, multilayer BayesB provided better predictions than multiple kernel learning, and lower prediction performance was observed in herd CV compared to repeated random sub-sampling CV.

Conclusions

Integration of genomic information with milk FTIR spectral can enhance milk protein trait predictions by 25% and 7% on average for repeated random sub-sampling and herd CV, respectively. Multiple kernel learning and multilayer BayesB outperformed PLS when used to integrate heterogeneous data for phenotypic predictions.

Breed of goat affects the prediction accuracy of milk coagulation properties using Fourier-transform infrared spectroscopy

The prediction of traditional goat milk coagulation properties (MCP) and curd firmness over time (CF_t) parameters via Fourier-transform infrared (FTIR) spectroscopy can be of significant economic interest to the dairy industry and can contribute to the breeding objectives for the genetic improvement of dairy goat breeds. Therefore, the aims of this study were to (1) explore the variability of milk FTIR spectra from 4 goat breeds (Camosciata delle Alpi, Murciano-Granadina, Maltese, and Sarda), and to assess the possible discriminant power of milk FTIR spectra among breeds, (2) assess the viability to predict coagulation traits by using milk FTIR spectra, and (3) quantify the effect of the breed on the prediction accuracy of MCP and CF_t parameters. In total, 611 individual goat milk samples were used. Analysis of variance of measured MCP and CF_t parameters was carried out using a mixed model including the farm and pendulum as random factors, and breed, parity, and days in milk as fixed factors. Milk spectra for each goat were collected over the spectral range from wavenumber 5,011 to 925 × cm^-1. Discriminant analysis of principal components was used to assess the ability of FTIR spectra to identify breed of origin. A Bayesian model was used to calibrate equations for each coagulation trait. The accuracy of the model and the prediction equation was assessed by cross-validation (CRV; 80% training and 20% testing set) and stratified CRV (SCV; 3 breeds in the training set, one breed in the testing set) procedures. Prediction accuracy was assessed by using coefficient of determination of validation (R²_VAL), the root mean square error of validation (RMSE_VAL), and the ratio performance deviation. Moreover, measured and FTIR predicted traits were compared in the SCV procedure by assessing their least squares means for the breed effect, Pearson correlations, and variance heteroscedasticity. Results showed the feasibility of using FTIR spectra and multivariate analyses to correctly assign milk samples to their breeds of origin. The R²_VAL values obtained with the CRV procedure were moderate to high for the majority of coagulation traits, with RMSE_VAL and ratio performance deviation values increasing as the coagulation process progresses from rennet addition. Prediction accuracy obtained with the SCV were strongly influenced by the breed, presenting general low values restricting a practical application. In addition, the low Pearson correlation coefficients of Sarda breed for all the traits analyzed, and the heteroscedastic variances of Camosciata delle Alpi, Murciano-Granadina, and Maltese breeds, further indicated that it is fundamental to consider the differences existing among breeds for the prediction of milk coagulation traits.

Goat farm variability affects milk Fourier-transform infrared spectra used for predicting coagulation properties

Driven by the large amount of goat milk destined for cheese production, and to pioneer the goat cheese industry, the objective of this study was to assess the effect of farm in predicting goat milk-coagulation and curd-firmness traits via Fourier-transform infrared spectroscopy. Spectra from 452 Sarda goats belonging to 14 farms in central and southeast Sardinia (Italy) were collected. A Bayesian linear regression model was used, estimating all spectral wavelengths' effects simultaneously. Three traditional milk-coagulation properties [rennet coagulation time (min), time to curd firmness of 20 mm (min), and curd firmness 30 min after rennet addition (mm)] and 3 curd-firmness measures modeled over time [rennet coagulation time estimated according to curd firmness change over time (RCT_eq), instant curd-firming rate constant, and asymptotical curd firmness] were considered. A stratified cross validation (SCV) was assigned, evaluating each farm separately (validation set; VAL) and keeping the remaining farms to train (calibration set) the statistical model. Moreover, a SCV, where 20% of the goats randomly taken (10 replicates per farm) from the VAL farm entered the calibration set, was also considered (SCV₈₀). To assess model performance, coefficient of determination (R²_VAL) and the root mean squared error of validation were recorded. The R²_VAL varied between 0.14 and 0.45 (instant curd-firming rate constant and RCT_eq, respectively), albeit the standard deviation was approximating half of the mean for all the traits. Although average results of the 2 SCV procedures were similar, in SCV₈₀, the maximum R²_VAL increased at about 15% across traits, with the highest observed for time to curd firmness of 20 mm (20%) and the lowest for RCT_eq (6%). Further investigation evidenced important variability among farms, with R²_VAL for some of them being close to 0. Our work outlined the importance of considering the effect of farm when developing Fourier-transform infrared spectroscopy prediction equations for coagulation and curd-firmness traits in goats.

Validation of milk mid-infrared spectroscopy for predicting the metabolic status of lactating dairy cows in Australia

Increased concentrations of some serum biomarkers are known to be associated with impaired health of dairy cows. Therefore, being able to predict these biomarkers, especially in the early stage of lactation, would enable preventive management decision. Some health biomarkers may also be used as phenotypes for genetic improvement for improved animal health. In this study, we validated the accuracy and robustness of models for predicting serum concentrations of β-hydroxybutyrate (BHB), fatty acids, and urea nitrogen, using milk mid-infrared (MIR) spectroscopy. The data included 3,262 blood samples of 3,027 lactating Holstein-Friesian cows from 19 dairy herds in Southeastern Australia, collected in the period from July 2017 to April 2020. The models were developed using partial least squares regression and were validated using 10-fold random cross-validation, herd-year by herd-year external validation, and year by year validation. The coefficients of determination (R²) for prediction of serum BHB, fatty acids, and urea obtained through random cross-validation were 0.60, 0.42, and 0.87, respectively. For the herd-year by herd-year external validation, the prediction accuracies held up comparatively well, with R² values of 0.49, 0.33, and 0.67 for of serum BHB, fatty acids, and urea, respectively. When the models were developed using data from a single year to predict data collected in future years, the R² remained comparable, however, the root mean squared errors increased substantially (4-10 times larger than compared with that of herd-year by herd-year external validation) which could be due to machine differences in spectral response, the change in spectral response of individual machines over time, or other differences associated with farm management between seasons. In conclusion, the mid-infrared equations for predicting serum BHB, fatty acids, and urea have been validated. The prediction equations could be used to help farmers detect cows with metabolic disorders in early lactation in addition to generating novel phenotypes for genetic improvement purposes.

Short communication: Iodine content in bovine milk is lowly heritable and shows limited genetic variation

Milk and dairy products are considered important sources of iodine in several countries. Despite this, there is a paucity of studies that have investigated sources of variation of milk iodine, especially on a large scale. So far, it is not clear if milk iodine content could be increased through breeding in dairy cattle. Recently, a mid-infrared spectroscopy prediction model has been developed for an indirect quantification of iodine content in cow milk, as it is a faster and less expensive method that allows the prediction at population level. The model has coefficient of determination and ratio of performance to deviation in external validation of 0.57 and 1.44, respectively, and it was used in the present study to predict the iodine content from historical milk spectral data to investigate phenotypic and genetic aspects in the Italian Holstein cattle. Based on the accuracy of the model, the prediction was interpreted as proxy for the real milk iodine concentration (IOD_P). The data set comprised 33,776 test-day records with IOD_P from 4,072 cows. Data of IOD_P were transformed through natural logarithm to achieve a normal distribution. The effect of parity, lactation stage, and month of sampling were investigated, and genetic parameters were estimated using a test-day repeatability animal model. Milk IOD_P decreased with parities and was the lowest in early lactation. Heritability of IOD_P was low (0.025) and it was positively genetically correlated with milk yield and negatively with fat content. Results suggested that it would be challenging to directly improve this trait through breeding strategies in dairy cattle, because IOD_P is mainly affected by temporary environmental factors and thus, cannot be easily improved through genetics. Although preliminary, findings of this study suggest that it would be more convenient to develop feeding and management strategies to drive milk iodine level than to put efforts and resources into breeding strategies. Further studies should validate IOD_P as an indicator trait of milk iodine content by improving reference data and estimating genetic correlation between predicted and measured values.

Genetic and Non-Genetic Variation of Milk Total Antioxidant Activity Predicted from Mid-Infrared Spectra in Holstein Cows

Simple Summary

The total antioxidant activity (TAA) of food is important for human health and results from the contribution of different nutraceutical compounds. Direct determination of TAA in food is time-consuming and expensive. Infrared technologies allow the prediction of difficult-to-measure traits with certain accuracy in several organic matrices, including TAA of bovine milk. In order to understand the background of TAA and identify potential strategies to improve this feature in bovine milk, we explored its non-genetic sources of variation and estimated heritability and correlations with traits of economic interest in a large database of Holstein cows.

Abstract

Food antioxidants enhance products shelf life and stability during technological treatments through the maintenance of their physical and chemical properties. Moreover, they are endowed with several positive effects on human health, including cell membranes preservation, enzyme functionality, and DNA integrity. Milk has been described in relation to a wide array of fat soluble and water-soluble antioxidant compounds, in particular vitamin A, C, and E, lactoferrin and peptides derived from casein and whey proteins. The total antioxidant activity (TAA) of milk is a novel and scarcely explored trait, defined as the sum of antioxidant contributions of the aforementioned compounds. On this background, the aims of the present study were to investigate the variability of milk TAA on a large scale exploiting predictions obtained through mid-infrared (MIR) spectroscopy and to estimate genetic parameters of this trait in Holstein cows. Individual milk samples were collected between January 2011 and December 2018 during the routine milk recording procedure. Samples were analysed for gross composition through MIR spectroscopy and MIR spectra were stored. Milk TAA was then predicted (pTAA) from the stored milk MIR spectra (111,653 test-day records of 9519 cows in 344 herds) using the previously developed prediction model; considering the prediction accuracy, pTAA might be considered a proxy of the TAA determined through the reference method. Overall, pTAA averaged 7.16 mmoL/L of Trolox equivalents, showed a nadir around 40 days after calving and increased thereafter, following a linear trend up to the end of lactation. The lowest pTAA was observed in milk sampled from June to September. Milk pTAA was heritable (0.401 ± 0.015) and genetically associated to fat yield (0.366 ± 0.049), crude protein (CP) yield (0.238 ± 0.052), fat percentage (0.616 ± 0.022) and CP percentage (0.754 ± 0.015). The official selection index of Italian Holstein put the 49% of the emphasis on fat and protein yield and percentage; therefore, it derives that an indirect favourable selection for milk pTAA should be already in progress in Italian Holstein population.

A comparison of 4 different machine learning algorithms to predict lactoferrin content in bovine milk from mid-infrared spectra

Lactoferrin (LF) is a glycoprotein naturally present in milk. Its content varies throughout lactation, but also with mastitis; therefore it is a potential additional indicator of udder health beyond somatic cell count. Condequently, there is an interest in quantifying this biomolecule routinely. First prediction equations proposed in the literature to predict the content in milk using milk mid-infrared spectrometry were built using partial least square regression (PLSR) due to the limited size of the data set. Thanks to a large data set, the current study aimed to test 4 different machine learning algorithms using a large data set comprising 6,619 records collected across different herds, breeds, and countries. The first algorithm was a PLSR, as used in past investigations. The second and third algorithms used partial least square (PLS) factors combined with a linear and polynomial support vector regression (PLS + SVR). The fourth algorithm also used PLS factors, but included in an artificial neural network with 1 hidden layer (PLS + ANN). The training and validation sets comprised 5,541 and 836 records, respectively. Even if the calibration prediction performances were the best for PLS + polynomial SVR, their validation prediction performances were the worst. The 3 other algorithms had similar validation performances. Indeed, the validation root mean squared error (RMSE) ranged between 162.17 and 166.75 mg/L of milk. However, the lower standard deviation of cross-validation RMSE and the better normality of the residual distribution observed for PLS + ANN suggest that this modeling was more suitable to predict the LF content in milk from milk mid-infrared spectra (R²v = 0.60 and validation RMSE = 162.17 mg/L of milk). This PLS +ANN model was then applied to almost 6 million spectral records. The predicted LF showed the expected relationships with milk yield, somatic cell score, somatic cell count, and stage of lactation. The model tended to underestimate high LF values (higher than 600 mg/L of milk). However, if the prediction threshold was set to 500 mg/L, 82% of samples from the validation having a content of LF higher than 600 mg/L were detected. Future research should aim to increase the number of those extremely high LF records in the calibration set.

Predicting the Content of 20 Minerals in Beef by Different Portable Near-Infrared (NIR) Spectrometers

The aim of this study was to test the predictability of a detailed mineral profile of beef using different portable near-infrared spectrometers (NIRS). These devices are rapid, chemical waste-free, cheap, nondestructive tools that can be used directly on the meat surface in the work environment without the need to take samples. We compared a transportable Visible-NIRS (weight 5.6 kg; wavelength 350-1830 nm), a portable NIRS (2.0 kg; 950-1650 nm), and a hand-held Micro-NIRS (0.06 kg; 905-1649 nm) to predict the contents of 20 minerals (measured by ICP-OES) in 178 beef samples (Longissimus thoracis muscle) using different mathematical pretreatments of the spectra and partial least square regressions. The externally validated results show that Fe, P, Mg, S, Na, and Pb have some potential for prediction with all instruments (R2VAL: 0.40-0.83). Overall, the prediction performances of the three instruments were similar, although the smallest (Micro-NIRS) exhibited certain advantages.

Infrared Spectrometry as a High-Throughput Phenotyping Technology to Predict Complex Traits in Livestock Systems

High-throughput phenotyping technologies are growing in importance in livestock systems due to their ability to generate real-time, non-invasive, and accurate animal-level information. Collecting such individual-level information can generate novel traits and potentially improve animal selection and management decisions in livestock operations. One of the most relevant tools used in the dairy and beef industry to predict complex traits is infrared spectrometry, which is based on the analysis of the interaction between electromagnetic radiation and matter. The infrared electromagnetic radiation spans an enormous range of wavelengths and frequencies known as the electromagnetic spectrum. The spectrum is divided into different regions, with near- and mid-infrared regions being the main spectral regions used in livestock applications. The advantage of using infrared spectrometry includes speed, non-destructive measurement, and great potential for on-line analysis. This paper aims to review the use of mid- and near-infrared spectrometry techniques as tools to predict complex dairy and beef phenotypes, such as milk composition, feed efficiency, methane emission, fertility, energy balance, health status, and meat quality traits. Although several research studies have used these technologies to predict a wide range of phenotypes, most of them are based on Partial Least Squares (PLS) and did not considered other machine learning (ML) techniques to improve prediction quality. Therefore, we will discuss the role of analytical methods employed on spectral data to improve the predictive ability for complex traits in livestock operations. Furthermore, we will discuss different approaches to reduce data dimensionality and the impact of validation strategies on predictive quality.

Large-scale phenotyping in dairy sector using milk MIR spectra: Key factors affecting the quality of predictions

Methods and technologies enabling the estimation at large scale of important traits for the dairy sector are of great interest. Those phenotypes are necessary to improve herd management, animal genetic evaluation, and milk quality control. In the recent years, the research was very active to predict new phenotypes from the mid-infrared (MIR) analysis of milk. Models were developed to predict phenotypes such as fine milk composition, milk technological properties or traits related to cow health, fertility and environmental impact. Most of models were developed within research contexts and often not designed for routine use. The implementation of models at a large scale to predict new traits of interest brings new challenges as the factors influencing the robustness of models are poorly documented. The first objective of this work is to highlight the impact on prediction accuracy of factors such as the variability of the spectral and reference data, the spectral regions used and the complexity of models. The second objective is to emphasize methods and indicators to evaluate the quality of models and the quality of predictions generated under routine conditions. The last objective is to outline the issues and the solutions linked with the use and transfer of models on large number of instruments. Based on partial least square regression and 10 datasets including milk MIR spectra and reference quantitative values for 57 traits of interest, the impact of the different factors is illustrated by evaluating the influence on the validation root mean square error of prediction (RMSEP). In the displayed examples, all factors, when well set up, increase the quality of predictions, with an improvement of the RMSEP ranging from 12% to 43%. This work also aims to underline the need for and the complementarity between different validation procedures, statistical parameters and quality assurance methods. Finally, when using and transferring models, the impact of the spectral standardization on the prediction reproducibility is highlighted with an improvement up to 86% with the tested models, and the monitoring of individual spectrometer stability over time appears essential. This list inspired from our experience is of course not exhaustive. The displayed results are only examples and not general rules and other aspects play a role in the quality of final predictions. However, this work highlights good practices, methods and indicators to increase and evaluate quality of phenotypes predicted at a large scale. The results obtained argue for the development of guidelines at international levels, as well as international collaborations in order to constitute large and robust datasets and enable the use of models in routine conditions.

Short communication: Prediction of hyperketonemia in dairy cows in early lactation using on-farm cow data and net energy intake by partial least square discriminant analysis

The objectives of this study were (1) to evaluate if hyperketonemia in dairy cows (defined as plasma β-hydroxybutyrate ≥1.0 mmol/L) can be predicted using on-farm cow data either in current or previous lactation week, and (2) to study if adding individual net energy intake (NEI) can improve the predictive ability of the model. Plasma β-hydroxybutyrate concentration, on-farm cow data (milk yield, percentage of fat, protein and lactose, fat- and protein-corrected milk yield, body weight, body weight change, dry period length, parity, and somatic cell count), and NEI of 424 individual cows were available weekly through lactation wk 1 to 5 postpartum. To predict hyperketonemia in dairy cows, models were first trained by partial least square discriminant analysis, using on-farm cow data in the same or previous lactation week. Second, NEI was included in models to evaluate the improvement of the predictability of the models. Through leave-one trial-out cross-validation, models were evaluated by accuracy (the ratio of the sum of true positive and true negative), sensitivity (68.2% to 84.9%), specificity (61.5% to 98.7%), positive predictive value (57.7% to 98.7%), and negative predictive value (66.2% to 86.1%) to predict hyperketonemia of dairy cows. Through lactation wk 1 to 5, the accuracy to predict hyperketonemia using data in the same week was 64.4% to 85.5% (on-farm cow data only), 66.1% to 87.0% (model including NEI), and using data in the previous week was 58.5% to 82.0% (on-farm cow data only), 59.7% to 85.1% (model including NEI). An improvement of the accuracy of the model due to including NEI ranged among lactation weeks from 1.0% to 4.4% when using data in the same lactation week and 0.2% to 6.6% when using data in the previous lactation week. In conclusion, trained models via partial least square discriminant analysis have potential to predict hyperketonemia in dairy cows not only using data in the current lactation week, but also using data in the previous lactation week. Net energy intake can improve the accuracy of the model, but only to a limited extent. Besides NEI, body weight, body weight change, milk fat, and protein content were important variables to predict hyperketonemia, but their rank of importance differed across lactation weeks.

Heritability estimates of predicted blood β-hydroxybutyrate and nonesterified fatty acids and relationships with milk traits in early-lactation Holstein cows

At the beginning of lactation, high-producing cows commonly experience an unbalanced energy status that is often responsible for the onset of metabolic disorders and impaired health and performance. Blood β-hydroxybutyrate (BHB) and nonesterified fatty acids (NEFA) are indicators of excessive fat mobilization and circulating ketone bodies. Recently, prediction models based on mid-infrared (MIR) spectroscopy have been developed to assess blood BHB and NEFA from routinely collected individual milk samples. This study aimed to estimate genetic parameters of blood BHB and NEFA predicted from milk MIR spectra and to assess their phenotypic and genetic correlations with milk production and composition traits in early-lactation Holstein cows. The data set comprised the first test-day record within lactation and spectra of individual milk samples (n = 22,718) of 13,106 Holstein cows collected from 5 to 35 d in milk (DIM). Blood BHB and NEFA were predicted from milk MIR spectra using previously developed prediction models. Genetic parameters of blood metabolites and milk traits were estimated for the whole observational period (5-35 DIM) and within 6 classes of DIM. Blood BHB and NEFA showed similar genetic variation across DIM, with the highest heritability in the first 10 d after calving (0.31 ± 0.06 and 0.19 ± 0.05 for BHB and NEFA, respectively). The genetic correlation between BHB and NEFA was moderate (0.51 ± 0.05). Genetic correlations of BHB with milk yield, SCS, protein percentage, lactose percentage, and urea nitrogen content were similar to, or at least in the same direction as, the correlations of NEFA with the same traits, whereas opposite correlations were observed with fat percentage and fat-to-protein ratio. Results of the current study suggest that blood BHB and NEFA predicted from milk MIR spectra have genetic variation that is potentially exploitable for breeding purposes. Therefore, they could be used as indicator traits of hyperketonemia in a selection index aimed to reduce the susceptibility of dairy cows to metabolic disorders in early lactation.

Symposium review: Genomic selection for reducing environmental impact and adapting to climate change

The world has been warming as greenhouse gases accumulate. Worldwide from 1880 to 2012, the average surface temperature has increased by about 0.85°C and by 0.12°C per decade since 1951. The world's cattle population is a contributor to atmospheric methane, a potent greenhouse gas, in addition to suffering from high temperatures combined with humidity. This makes research into reducing the global footprint of dairy cows of importance on a long-term horizon, while improving tolerance to heat could alleviate the effects of rising temperatures. In December 2017, genomic estimated breeding values for heat tolerance in dairy cattle were released for the first time in Australia. Currently, heat tolerance is not included in the Balanced Performance Index (Australia's national selection index), and the correlation between heat tolerance breeding values and Balanced Performance Index is -0.20, so over time, heat tolerance has worsened due to lack of selection pressure. However, in contrast, sizable reductions in greenhouse gas emissions have been achieved as a favorable response to selecting for increased productivity, longevity, and efficiency, with opportunities for even greater gains through selecting for cow emissions directly. Internationally considerable research effort has been made to develop breeding values focused on reducing methane emissions using individual cow phenotypes. This requires (1) definition of breeding objectives and selection criteria and (2) assembling a sufficiently large data set for genomic prediction. Selecting for heat tolerance and reduced emissions directly may improve resilience to changing environments while reducing environmental impact.

Diagnosing the pregnancy status of dairy cows: How useful is milk mid-infrared spectroscopy?

Pregnancy diagnosis is an essential part of successful breeding programs on dairy farms. Milk composition alters with pregnancy, and this is well documented. Fourier-transform mid-infrared (MIR) spectroscopy is a rapid and cost-effective method for providing milk spectra that reflect the detailed composition of milk samples. Therefore, the aim of this study was to assess the ability of MIR spectroscopy to predict the pregnancy status of dairy cows. The MIR spectra and insemination records were available from 8,064 Holstein cows of 19 commercial dairy farms in Australia. Three strategies were studied to classify cows as open or pregnant using partial least squares discriminant analysis models with random cow-independent 10-fold cross-validation and external validation on a cow-independent test set. The first strategy considered 6,754 MIR spectra after insemination used as independent variables in the model. The results showed little ability to detect the pregnancy status as the area under the receiver operating characteristic curve was 0.63 and 0.65 for cross-validation and testing, respectively. The second strategy, involving 1,664 records, aimed to reduce noise in the MIR spectra used as predictors by subtracting a spectrum before insemination (i.e., open spectrum) from the spectrum after insemination. The accuracy was comparable with the first approach, showing no superiority of the method. Given the limited results for these models when using combined data from all stages after insemination, the third strategy explored separate models at 7 stages after insemination comprising 348 to 1,566 records each (i.e., progressively greater gestation) with single MIR spectra after insemination as predictors. The models developed using data recorded after 150 d of pregnancy showed promising prediction accuracy with the average value of area under the receiver operating characteristic curve of 0.78 and 0.76 obtained through cross-validation and testing, respectively. If this can be confirmed on a larger data set and extended to somewhat earlier stages after insemination, the model could be used as a complementary tool to detect fetal abortion.

Symposium review: Big data, big predictions: Utilizing milk Fourier-transform infrared and genomics to improve hyperketonemia management

Negative animal health and performance outcomes are associated with disease incidences that can be labor-intensive, costly, and cumbersome for many farms. Amelioration of unfavorable outcomes through early detection and treatment of disease has emphasized the value of improving health monitoring. Although the value is recognized, detecting hyperketonemia (HYK) is still difficult for many farms to do practically and efficiently. Increasing data streams available to farms presents opportunities to use data to better monitor cow and herd health; however, challenges remain with regard to validating, integrating, and interpreting data. During the transition to lactation period, useful data are presented in the form of milk production and composition, milk Fourier-transform infrared (FTIR) wavelength absorbance, cow management records, and genomics, which have been employed to monitor postpartum onset of HYK. Attempts to predict postpartum HYK from test-day milk and performance variables incorporated into multiple linear regression models have demonstrated sufficient accuracy to monitor monthly herd prevalence; however, they lacked the sensitivity and specificity for individual cow diagnostics. Subsequent artificial neural network prediction models employing FTIR data and milk composition variables achieved 83 and 81% sensitivity and specificity for individual cow diagnostics. Although these results fail to reach the diagnostic goals of 90%, they are achieved without individual cow blood samples, which may justify acceptance of lower performance. The caveat is that these models depend on milk analysis, which is traditionally done every 4 weeks. This infrequent sampling allows for a single diagnostic sample for about half of the fresh cows. Benefits to farms are greatly improved if postpartum cows can be milk-tested weekly. Additionally, this allows for close monitoring of somatic cell count and may open the door for use of other herd health monitoring tools. Future improvements in these models may be achievable by maximizing sensitivity at the expense of specificity and may be most economical in disorders for which the cost of treatment is less than that of mistreating (e.g., HYK). Genomic predictions for HYK may be improved by incorporating genome-wide associated SNP and further utilized for precision management of HYK risk groups. Development and validation of HYK prediction models may provide producers with individual cow and herd-level management tools.

Combined use of milk infrared spectra and genotypes can improve prediction of milk fat composition

It has been shown that milk infrared (IR) spectroscopy can be used to predict detailed milk fat composition. In addition, polymorphisms with substantial effects on milk fat composition have been identified. In this study, we investigated the combined use of milk IR spectroscopy and genotypes of dairy cows on the accuracy of predicting milk fat composition. Milk fat composition data based on gas chromatography and milk IR spectra were available for 1,456 Dutch Holstein Friesian cows. In addition, genotypes for the diacylglycerol acyltransferase 1 (DGAT1) K232A and stearoyl-CoA desaturase 1 (SCD1) A293V polymorphisms and a SNP located in an intron of the fatty acid synthase (FASN) gene were available. Adding SCD1 genotypes to the milk IR spectra resulted in a considerable improvement of the prediction accuracy for the unsaturated fatty acids C10:1, C12:1, C14:1 cis-9, and C16:1 cis-9 and their corresponding unsaturation indices. Adding DGAT1 genotypes to the milk IR spectra resulted in an improvement of the prediction accuracy for C16:1 cis-9 and C16 index. Adding genotypes of the FASN SNP to the IR spectra did not improve prediction of milk fat composition. This study demonstrated the potential of combining milk IR spectra with genotypic information from 3 polymorphisms to predict milk fat composition. We hypothesize that prediction accuracy of milk fat composition can be further improved by combining milk IR spectra with genomic breeding values.

Prediction of blood metabolites from milk mid-infrared spectra in early-lactation cows

Dairy cows commonly experience an unbalanced energy status in early lactation, and this condition can lead to the onset of several metabolic disorders. Blood metabolic profile testing is a valid tool to monitor and detect the most common early lactation disorders, but blood sampling and analysis are time-consuming and expensive, and the procedure is invasive and stressful for the cows. Mid-infrared (MIR) spectroscopy is routinely used to analyze milk composition, being a cost-effective and nondestructive method. The present study aimed to assess the feasibility of using routine milk MIR spectra for the prediction of main blood metabolites in dairy cows, and to investigate associations between measured blood metabolites and milk traits. Twenty herds of Holstein Friesian, Brown Swiss, or Simmental cows located in Northeast Italy were visited 1 to 4 times between December 2017 and June 2018, and blood and milk samples were collected from all lactating cows within 35 d in milk. Concentrations of main blood metabolites and milk MIR spectra were recorded from 295 blood and milk samples and used to develop prediction models for blood metabolic traits through backward interval partial least squares analysis. Blood β-hydroxybutyrate (BHB), urea, and nonesterified fatty acids were the most predictable traits, with coefficients of determination of 0.63, 0.58, and 0.52, respectively. On the contrary, predictive performance for blood glucose, triglycerides, cholesterol, glutamic oxaloacetic transaminase, and glutamic pyruvic transaminase were not accurate. Associations of blood BHB and urea with their respective contents in milk were moderate to strong, whereas all other correlations were weak. Predicted blood BHB showed an improved performance in detecting cows with hyperketonemia (blood BHB ≥ 1.2 mmol/L), compared with commercial calibration equation for milk BHB. Results highlighted the opportunity of using milk MIR spectra to predict blood metabolites and thus to collect routine information on the metabolic status of early-lactation cows at a population level.