Evaluation of a milk test for detection of subclinical ketosis

The Use of Multilayer Perceptron Artificial Neural Networks to Detect Dairy Cows at Risk of Ketosis

Simple Summary

Ketosis is a serious metabolic disease in high-yield dairy cows, that affects productive herds throughout the world. Subclinical ketosis is one of the most dominant metabolic disorders in dairy herds during early lactation, so early detection and prevention are important for both economic and animal welfare reasons. Neural networks, which offer a high degree of accuracy in predicting various phenomena and processes where there is no clear causal correlation or there are no rules that allow the establishment of a logical cause-and-effect relationship, can be used to address problems related to prediction, classification, or control. A Multi-Layer perceptron (MLP) is a feedforward artificial neural network model that takes input data for a set of proper output. This study investigated the performance of four algorithms used to train MLP networks. The experimental results demonstrate that the MLP network model improved the accuracy of process recognition of subclinical ketosis in dairy cows. The received artificial model’s results were saved in the predictive model markup language (PMML) and can be used to describe the learning set, the algorithm used in the data mining application and related information.

Abstract

Subclinical ketosis is one of the most dominant metabolic disorders in dairy herds during lactation. Cows suffering from ketosis experience elevated ketone body levels in blood and milk, including β-hydroxybutyric acid (BHB), acetone (ACE) and acetoacetic acid. Ketosis causes serious financial losses to dairy cattle breeders and milk producers due to the costs of diagnosis and management as well as animal welfare reasons. Recent years have seen a growing interest in the use of artificial neural networks (ANNs) in various fields of science. ANNs offer a modeling method that enables the mapping of highly complex functional relationships. The purpose of this study was to determine the relationship between milk composition and blood BHB levels associated with subclinical ketosis in dairy cows, using feedforward multilayer perceptron (MLP) artificial neural networks. The results were verified based on the estimated sensitivity and specificity of selected network models, an optimum cut-off point was identified for the receiver operating characteristic (ROC) curve and the area under the ROC curve (AUC). The study demonstrated that BHB, ACE and lactose (LAC) levels, as well as the fat-to-protein ratio in milk, were important input variables in the network training process. For the identification of cows at risk of subclinical ketosis, variables such as BHB and ACE levels in milk were of particular relevance, with a sensitivity and specificity of 0.84 and 0.61, respectively. It was found that the back propagation algorithm offers opportunities to integrate artificial intelligence and dairy cattle welfare within a computerized decision support tool.

Expert opinion as priors for random effects in Bayesian prediction models: Subclinical ketosis in dairy cows as an example

Random effects regression models are routinely used for clustered data in etiological and intervention research. However, in prediction models, the random effects are either neglected or conventionally substituted with zero for new clusters after model development. In this study, we applied a Bayesian prediction modelling method to the subclinical ketosis data previously collected by Van der Drift et al. (2012). Using a dataset of 118 randomly selected Dutch dairy farms participating in a regular milk recording system, the authors proposed a prediction model with milk measures as well as available test-day information as predictors for the diagnosis of subclinical ketosis in dairy cows. While their original model included random effects to correct for the clustering, the random effect term was removed for their final prediction model. With the Bayesian prediction modelling approach, we first used non-informative priors for the random effects for model development as well as for prediction. This approach was evaluated by comparing it to the original frequentist model. In addition, herd level expert opinion was elicited from a bovine health specialist using three different scales of precision and incorporated in the prediction as informative priors for the random effects, resulting in three more Bayesian prediction models. Results showed that the Bayesian approach could naturally take the clustering structure of clusters into account by keeping the random effects in the prediction model. Expert opinion could be explicitly combined with individual level data for prediction. However in this dataset, when elicited expert opinion was incorporated, little improvement was seen at the individual level as well as at the herd level. When the prediction models were applied to the 118 herds, at the individual cow level, with the original frequentist approach we obtained a sensitivity of 82.4% and a specificity of 83.8% at the optimal cutoff, while with the three Bayesian models with elicited expert opinion, we obtained sensitivities ranged from 78.7% to 84.6% and specificities ranged from 75.0% to 83.6%. At the herd level, 30 out of 118 within herd prevalences were correctly predicted by the original frequentist approach, and 31 to 44 herds were correctly predicted by the three Bayesian models with elicited expert opinion. Further investigation in expert opinion and distributional assumption for the random effects was carried out and discussed.

Predictive ability of host genetics and rumen microbiome for subclinical ketosis

Subclinical metabolic disorders such as ketosis cause substantial economic losses for dairy farmers in addition to the serious welfare issues they pose for dairy cows. Major hurdles in genetic improvement against metabolic disorders such as ketosis include difficulties in large-scale phenotype recording and low heritability of traits. Milk concentrations of ketone bodies, such as acetone and β-hydroxybutyric acid (BHB), might be useful indicators to select cows for low susceptibility to ketosis. However, heritability estimates reported for milk BHB and acetone in several dairy cattle breeds were low. The rumen microbial community has been reported to play a significant role in host energy homeostasis and metabolic and physiologic adaptations. The current study aims at investigating the effects of cows' genome and rumen microbial composition on concentrations of acetone and BHB in milk, and identifying specific rumen microbial taxa associated with variation in milk acetone and BHB concentrations. We determined the concentrations of acetone and BHB in milk using nuclear magnetic resonance spectroscopy on morning milk samples collected from 277 Danish Holstein cows. Imputed high-density genotype data were available for these cows. Using genomic and microbial prediction models with a 10-fold resampling strategy, we found that rumen microbial composition explains a larger proportion of the variation in milk concentrations of acetone and BHB than do host genetics. Moreover, we identified associations between milk acetone and BHB with some specific bacterial and archaeal operational taxonomic units previously reported to have low to moderate heritability, presenting an opportunity for genetic improvement. However, higher covariation between specific microbial taxa and milk acetone and BHB concentrations might not necessarily indicate a causal relationship; therefore further validation is needed before considering implementation in selection programs.

Association between blood β-hydroxybutyrate at 7 days postpartum and milk yield, disease occurrence and fertility in grazing dairy cattle with seasonal calving: a case study

Context Ketosis in grazing cattle has been sparsely studied. A large commercial grazing dairy in southern Chile, representative of a significant proportion of the systems in the country, was used in this case study. Aims The study had three objectives: (i) to establish a cut-off for β-hydroxybutyrate (BHB) concentration for subclinical ketosis (SCK), and use this to measure the proportion of cows with SCK at 7 days postpartum in spring- and autumn-calving cows; (ii) to describe the relationship of SCK and other periparturient diseases and fertility; and (iii) to compare milk yield of healthy cows and those affected by SCK in a dairy herd with autumn and spring parturitions under grazing conditions in southern Chile. Methods During 2016, 234 cows with autumn parturitions and 632 cows with spring parturitions (n = 866) were assessed for blood BHB at 7 days postpartum. A receiver operating characteristic analysis for a BHB cut-off value was completed. Models were developed for disease occurrence, culling risk, conception risk and pregnancy rate, considering SCK as the main explanatory variable. Key results In total, 810 cows were used for the final analysis. The frequency of cows with SCK, based on the cut-off value obtained (BHB ≥1.1 mmol/L), was 22.2% at 7 days postpartum. The risk of SCK was higher (P < 0.0001) in cows calving in spring (27.0%) than in autumn (10.3%), and in multiparous (24.6%) than primiparous cows (15.1%). The seasonal difference in proportion of cows with SCK was parity-dependent, because the frequency of SCK in multiparous cows was higher (P < 0.0005) in spring (32.0%) than autumn (10.1%), whereas SCK in primiparous cows showed no significant (P = 0.41) difference between spring (15.4%) and autumn (12.5%). Milk production up to 100 days-in-milk was greater (P = 0.002) in cows with SCK (3394 kg) than without SCK (3015 kg). Disease occurrence was higher (P < 0.0001) in cows with SCK and in multiparous cows (P < 0.0001). There was no difference in conception risk at first service (P = 0.62) or in overall pregnancy rate (P = 0.90) between cows with and without SCK. Conclusions Multiparous cows calving in spring had the highest risk of SCK (BHB ≥1.1 mmol/L). SCK was associated with higher milk yield and greater occurrence of other diseases, but not with reproductive performance. Implications Grazing herds have challenges with SCK that may require different management strategies depending on the calving season and the parity of the animals.

Genome-wide association analysis for β-hydroxybutyrate concentration in Milk in Holstein dairy cattle

BACKGROUND

Ketosis in dairy cattle has been shown to cause a high morbidity in the farm and substantial financial losses to dairy farmers. Ketosis symptoms, however, are difficult to identify, therefore, the amount of ketone bodies (mainly β-hydroxybutyric acid, BHB) is used as an indicator of subclinical ketosis in cows. It has also been shown that milk BHB concentrations have a strong correlation with ketosis in dairy cattle. Mid-infrared spectroscopy (MIR) has recently became a fast, cheap and high-throughput method for analyzing milk components. The aim of this study was to perform a genome-wide association study (GWAS) on the MIR-predicted milk BHB to identify genomic regions, genes and pathways potentially affecting subclinical ketosis in North American Holstein dairy cattle.

RESULTS

Several significant regions were identified associated with MIR-predicted milk BHB concentrations (indicator of subclinical ketosis) in the first lactation (SCK1) and second and later lactations (SCK2) in Holstein dairy cows. The strongest association was located on BTA6 for SCK1 and BTA14 on SCK2. Several SNPs on BTA6 were identified in regions and variants reported previously to be associated with susceptibility to ketosis and clinical mastitis in Jersey and Holstein dairy cattle, respectively. One highly significant SNP on BTA14 was found within the DGAT1 gene with known functions on fat metabolism and inflammatory response in dairy cattle. A region on BTA6 and three SNPs on BTA20 were found to overlap between SCK1 and SCK2. However, a novel region on BTA20 (55-63 Mb) for SCK2 was also identified, which was not reported in previous association studies. Enrichment analysis of the list of candidate genes within the identified regions for MIR-predicted milk BHB concentrations yielded molecular functions and biological processes that may be involved in the inflammatory response and lipid metabolism in dairy cattle.

CONCLUSIONS

The results of this study confirmed several SNPs and genes identified in previous studies as associated with ketosis susceptibility and immune response, and also found a novel region that can be used for further analysis to identify causal variations and key regulatory genes that affect clinical/ subclinical ketosis.

Symposium review: Building a better cow-The Australian experience and future perspectives

Genomic selection has led to opportunities for developing new breeding values that rely on phenotypes in dedicated reference populations of genotyped cows. In Australia, it has been applied to 2 novel traits: feed efficiency, which was released in 2015 as feed saved breeding values, and heat tolerance genomic breeding values, released for the first time in 2017. Feed saved is already included in the national breeding objective, which is focused on profitability and designed to be in line with farmer preferences. Our future focus is on traits associated with animal health, either directly or in combination with predictor traits, such as mid-infrared spectral data and, into the future, automated data capture. Although it is common for many evaluated traits to have genomic reliabilities ranging between 60 and 75%, many new, genomic information-only traits are likely to have reliabilities of less than 50%. Pooling of phenotype data internationally and investing in maintenance of reference populations is one option to increase the reliability of these traits; the other is to apply improved genomic prediction methods. For example, advances in the use of sequence data, in addition to gene expression studies, can lead to improved persistence of genomic breeding values across breeds and generations and potentially lead to greater reliabilities. Lower genomic reliabilities of novel traits could reduce the overall index reliability. However, provided these traits contribute to the overall breeding objective (e.g., profit), they are worth including. Bull selection tools and personalized genetic trends are already available, but increased access to economic and automatic capture farm data may see even better use of data to improve farm management and selection decisions.

Invited review: Opportunities for genetic improvement of metabolic diseases

Metabolic disorders are disturbances to one or more of the metabolic processes in dairy cattle. Dysfunction of any of these processes is associated with the manifestation of metabolic diseases or disorders. In this review, data recording, incidences, genetic parameters, predictors, and status of genetic evaluations were examined for (1) ketosis, (2) displaced abomasum, (3) milk fever, and (4) tetany, as these are the most prevalent metabolic diseases where published genetic parameters are available. The reported incidences of clinical cases of metabolic disorders are generally low (less than 10% of cows are recorded as having a metabolic disease per herd per year or parity/lactation). Heritability estimates are also low and are typically less than 5%. Genetic correlations between metabolic traits are mainly positive, indicating that selection to improve one of these diseases is likely to have a positive effect on the others. Furthermore, there may also be opportunities to select for general disease resistance in terms of metabolic stability. Although there is inconsistency in published genetic correlation estimates between milk yield and metabolic traits, selection for milk yield may be expected to lead to a deterioration in metabolic disorders. Under-recording and difficulty in diagnosing subclinical cases are among the reasons why interest is growing in using easily measurable predictors of metabolic diseases, either recorded on-farm by using sensors and milk tests or off-farm using data collected from routine milk recording. Some countries have already initiated genetic evaluations of metabolic disease traits and currently most of these use clinical observations of disease. However, there are opportunities to use clinical diseases in addition to predictor traits and genomic information to strengthen genetic evaluations for metabolic health in the future.

Capitalizing on fine milk composition for breeding and management of dairy cows

The challenge of managing and breeding dairy cows is permanently adapting to changing production circumstances under socio-economic constraints. If managing and breeding address different timeframes of action, both need relevant phenotypes that allow for precise monitoring of the status of the cows, and their health, behavior, and well-being as well as their environmental impact and the quality of their products (i.e., milk and subsequently dairy products). Milk composition has been identified as an important source of information because it could reflect, at least partially, all these elements. Major conventional milk components such as fat, protein, urea, and lactose contents are routinely predicted by mid-infrared (MIR) spectrometry and have been widely used for these purposes. But, milk composition is much more complex and other nonconventional milk components, potentially predicted by MIR, might be informative. Such new milk-based phenotypes should be considered given that they are cheap, rapidly obtained, usable on a large scale, robust, and reliable. In a first approach, new phenotypes can be predicted from MIR spectra using techniques based on classical prediction equations. This method was used successfully for many novel traits (e.g., fatty acids, lactoferrin, minerals, milk technological properties, citrate) that can be then useful for management and breeding purposes. An innovation was to consider the longitudinal nature of the relationship between the trait of interest and the MIR spectra (e.g., to predict methane from MIR). By avoiding intermediate steps, prediction errors can be minimized when traits of interest (e.g., methane, energy balance, ketosis) are predicted directly from MIR spectra. In a second approach, research is ongoing to detect and exploit patterns in an innovative manner, by comparing observed with expected MIR spectra directly (e.g., pregnancy). All of these traits can then be used to define best practices, adjust feeding and health management, improve animal welfare, improve milk quality, and mitigate environmental impact. Under the condition that MIR data are available on a large scale, phenotypes for these traits will allow genetic and genomic evaluations. Introduction of novel traits into the breeding objectives will need additional research to clarify socio-economic weights and genetic correlations with other traits of interest.

Routine detection of hyperketonemia in dairy cows using Fourier transform infrared spectroscopy analysis of β-hydroxybutyrate and acetone in milk in combination with test-day information

The objective of this study was to assess the quality of a diagnostic model for the detection of hyperketonemia in early lactation dairy cows at test days. This diagnostic model comprised acetone and β-hydroxybutyrate (BHBA) concentrations in milk, as determined by Fourier transform infrared (FTIR) spectroscopy, in addition to other available test-day information. Plasma BHBA concentration was determined at a regular test day in 1,678 cows between 5 and 60 d in milk, originating from 118 randomly selected farms in the Netherlands. The observed prevalence of hyperketonemia (defined as plasma BHBA ≥1,200 µmol/L) was 11.2%. The value of FTIR predictions of milk acetone and milk BHBA concentrations as single tests for hyperketonemia were found limited, given the relatively large number of false positive test-day results. Therefore, a multivariate logistic regression model with a random herd effect was constructed, using parity, season, milk fat-to-protein ratio, and FTIR predictions of milk acetone and milk BHBA as predictive variables. This diagnostic model had 82.4% sensitivity and 83.8% specificity at the optimal cutoff value (defined as maximum sum of sensitivity and specificity) for the detection of hyperketonemia at test days. Increasing the cutoff value of the model to obtain a specificity of 95% increased the predicted value of a positive test result to 56.5%. Confirmation of test-positive samples with wet chemistry analysis of milk acetone or milk BHBA concentrations (serial testing) improved the diagnostic performance of the test procedure. The presented model was considered not suitable for individual detection of cows with ketosis due to the length of the test-day interval and the low positive predictive values of the investigated test procedures. The diagnostic model is, in our opinion, valuable for herd-level monitoring of hyperketonemia, especially when the model is combined with wet chemistry analysis of milk acetone or milk BHBA concentrations. By using the diagnostic model in combination with wet chemistry milk BHBA analysis, 84% of herds were correctly classified at a 10% alarm-level prevalence. As misclassification of herds may particularly occur when only a limited number of fresh cows are sampled, we suggest using prevalence estimates over several consecutive test days to evaluate feeding and management practices in smaller dairy farms.

Bovine subclinical ketosis in dairy herds in Iran

Subclinical ketosis (SCK) is defined as elevated concentrations of ketone bodies in the absence of clinical signs of ketosis. It is an important metabolic disease in dairy cattle during early lactation and is associated with losses in milk production and several other periparturient diseases. Limited information is available regarding the prevalence of SCK in dairy herds in Iran. The objectives of this study were (i) to determine the incidence of SCK in the dairy herds in Kerman province of Iran using serum beta-hydroxybutyrate (BHB) concentrations, and (ii) to investigate the relationship between serum concentrations of BHB and glucose of cows with SCK. In the present study, 90 multiparous Holstein cows (4-6 years old) from 11 commercial dairy herds were evaluated 3-4 weeks after calving. The distribution of blood BHB concentrations seemed to suggest a cut-off point of 1200 micromol/L between cows with and without SCK. At this cut-off point, 14.4% of tested cows (13/90) were classified as subclinically ketotic, with the prevalence rate within herd ranging from 10% to 20%. Cows with SCK were detected in all the investigated dairies except one. Blood glucose concentrations in cows with SCK were significantly lower (p < or = 0.05) than in cows without SCK, and serum BHB and glucose concentration were inversely correlated (r = -0.43, p < or = 0.05). The results suggest that, using a cut-off of 1200 micromol/L, BHB concentrations can be used during early lactation for diagnosis and to make management decisions for prevention and treatment.

Incidence of subclinical ketosis in cows supplemented with a monensin controlled-release capsule in Holstein cattle, Florida, USA

The objective of this study was to determine the effect of a monensin controlled-release capsule on the proportion of cows with subclinical ketosis (SK). During July to August 2001, 300 cows dried-off 50-70 days before expected parturition were randomly assigned to either a treatment (n = 150, oral capsule, 335 mg/d of monesin for 95 d) or control group (no capsule, n = 150). At 14 days postpartum, a milk sample was obtained and evaluated for beta-hydroxybutyrate (BHBA) using a semi-quantitative ketone test strip. In a sub-sample of 50 cows per group a blood sample was taken and analyzed for BHBA using an ELISA kit. Milk BHBA > or = 200 micromol/L was used as the cut-off value for diagnosis of SK. The incidence of SK based on the milk test was statistically different between groups (P < or = 0.05) with a value of 26.6% for control and 14.5% for cows treated with monensin, respectively. Cows treated with monensin were 0.68 times less likely to give a positive result for milk BHBA than non-treated cows (0.53-0.80; 95% CI). Serum BHBA concentrations did not differ between groups (0.81 +/- 0.09 mmol/L versus 0.70 +/- 0.07 mmol/L for controls and treated, respectively; P > 0.05). However, for each incremental increase in serum BHBA of 0.1 mmol/L occurrence of SK increased 52% (OR = 1.52; 1.21-1.91; 95% CI).

Diurnal variation and the effect of feed restriction on plasma and milk metabolites in TMR-fed dairy cows

The objective was to study the diurnal variation in metabolites in plasma and milk of dairy cows fed total mixed rations (TMR) with a low-energy (LE) or high-energy content (HE) expected to give a minor and a major diurnal variation, respectively. Further, the purpose was to quantify and compare the responses in plasma and milk parameters when cows changed from ad libitum to restrictive feeding. Eight multiparous, early-lactating Danish Holstein cows were used in a cross-over design with two consecutive 14-day periods. Blood and milk samples were collected hourly on day 11 of each period and on days 12-14 of each period, the cows were fed restrictively (65% of ad libitum dry-matter intake). The concentration of beta-hydroxybutyrate (BHB) in plasma was significantly higher in the evening for cows fed the HE TMR, than for cows fed the LE TMR. There was a significant diurnal variation in BHB in milk, with the highest concentrations between milkings and the lowest concentrations at milking. Non-esterified fatty acids (NEFA) in plasma showed significant diurnal variation that was caused by high concentrations in the morning. Plasma glucose did not show any diurnal variation. It has been argued that feeding a TMR removes diurnal changes related to feeding, which is contrary to earlier diurnal studies where concentrates have been fed twice daily. Feed restriction increased (P < 0.001) NEFA and BHB in plasma by 121 and 90%, respectively, while the glucose concentration decreased (P < 0.001) by 19%. Milk concentrations of BHB, citrate and fat increased (P < 0.001) by 163, 11 and 26%, respectively, because of feed restriction, while there were no changes in milk protein and lactose. The relatively high increase in BHB during feed restriction suggests that BHB is more advantageous as a milk indicator of metabolic status in dairy cows than citrate and fat.

Concentrations of ketones in milk in early lactation, and reproductive performance of dairy cows

Samples of blood and milk were taken from clinically ketotic cows at 12-hour intervals before and after treatment The concentration of beta-hydroxybutyrate was determined in plasma and milk, and the concentration of acetone was determined in milk. Measurement of milk acetone had the optimal combination of sensitivity, specificity and positive predictive value to identify ketotic cows and was subsequently used in a comparison of the fertility of ketotic and normal cows. Two samples of milk were taken approximately two weeks apart, between 12 and 60 days after calving, from 410 cows on three farms and the concentration of acetone in the milk was measured. Thirty cows with milk acetone concentrations of at least 0.4 mmol/litre were compared with paired control cows with milk acetone concentrations less than 0.3 mmol/litre. The ketotic cows had a significantly longer calving-to-conception interval and a significantly higher culling rate due to a failure to conceive (P<0.05).

Ketone bodies in milk and blood of dairy cows: relationship between concentrations and utilization for detection of subclinical ketosis

The objective of this study was to investigate the relationship between the concentrations of the different ketone bodies in milk and blood and to evaluate these concentrations for the detection of subclinical ketosis. A total of 60 multiparous cows were used. Concentrations of acetone, acetoacetate, and beta-hydroxybutyrate were analyzed quantitatively in blood and milk, and the Ketolac strip test was used for semiquantitative determination of beta-hydroxybutyrate in milk. Cows were defined subclinically ketotic when their concentration of blood beta-hydroxybutyrate was over 1200 micromol/L. High correlation coefficients were observed between blood acetone and blood acetoacetate, and between blood and milk acetone. On the contrary, concentrations of milk and blood beta-hydroxybutyrate were poorly correlated with the other concentrations of ketone bodies. The Ketolac strip test overestimated the concentrations of beta-hydroxybutyrate in milk. For the detection of subclinical ketosis, the best sensitivity-specificity combination was obtained with the determination of acetoacetate in blood or milk, with threshold concentrations of 125 and 50 micromol/L, respectively. Determination of beta-hydroxybutyrate in the milk via an enzymatic analysis or via the Ketolac strip test provided valuable results, with threshold concentrations of 70 and 100 micromol/L, respectively. The simplicity of use of the Ketolac strip test makes it a valuable way to investigate subclinical ketosis.

Evaluation of eight cow-side ketone tests in milk for detection of subclinical ketosis in dairy cows

The objective of this study was to evaluate eight cowside ketone tests when used with milk for detection of subclinical ketosis. A total of 469 dairy cows in the first week of lactation were studied. Twelve percent of these cows had subclinical ketosis, defined as >1400 micromol of beta-hydroxybutyrate/L of blood serum. The Pink test liquid and the Ketolac test strip were highly sensitive for subclinical ketosis when used with milk. The Uriscan and Rapignost test strips were poorly sensitive; the Ketostix, Ketur-Test, and Medi-Test-Keton test strips and the Acetonreagenz test tablet were insensitive for subclinical ketosis when used with milk. Pink and Ketolac milk ketone tests are potentially useful tools for use in a routine monitoring program to detect subclinical ketosis in early postpartal dairy cows.