Development and evaluation of predictive models for pregnancy risk in UK dairy cows

One suggested approach to improve the reproductive performance of dairy herds is through the targeted management of subgroups of biologically similar animals, such as those with similar probabilities of becoming pregnant, termed pregnancy risk. We aimed to use readily available farm data to develop predictive models of pregnancy risk in dairy cows. Data from a convenience sample of 108 dairy herds in the UK were collated and each herd was randomly allocated, at a ratio of 80:20, to either training or testing data sets. Following data cleaning, there were a total of 78 herds in the training data set and 20 herds in the testing data set. Data were further split by parity into nulliparous, primiparous, and multiparous subsets. An XGBoost model was trained to predict the insemination outcome in each parity subset, with predictors from farm records of breeding, calving and milk recording. Training data comprised 74,511 inseminations in 45,909 nulliparous animals, 86,420 inseminations in 39,439 primiparous animals, and 158,294 inseminations in 32,520 multiparous animals. The final models were evaluated by predicting with the testing data, comprising 31,740 inseminations in 19,647 nulliparous animals, 38,588 inseminations in 16,215 primiparous animals, and 65,049 inseminations in 12,439 multiparous animals. Model discrimination was assessed by calculating the area under receiver operating characteristic curves (AUC); model calibration was assessed by plotting calibration curves and compared across test herds by calculating the expected calibration error (ECE) in each test herd. The models were unable to discriminate between insemination outcomes with high accuracy, with an AUC of 0.63, 0.59 and 0.62 in the nulliparous, primiparous and multiparous subsets, respectively. The models were generally well-calibrated, meaning the model-predicted pregnancy risks were similar to the observed pregnancy risks. The mean (SD) ECE in the test herds was 0.038 (0.023), 0.028 (0.012) and 0.020 (0.008) in the nulliparous, primiparous and multiparous subsets respectively. The predictive models reported here could theoretically be used to identify subgroups of animals with similar pregnancy risk to facilitate targeted reproductive management; or provide information about cows' relative pregnancy risk compared with the herd average, which may support on-farm decision-making. Further research is needed to evaluate the generalizability of these predictive models and understand the source of variation in ECE between herds; however, this study demonstrates that it is possible to accurately predict pregnancy risk in dairy cows using readily available farm data.

Accuracy of early pregnancy diagnosis and determining pregnancy loss using different biomarkers and machine learning applications in dairy cattle

This study aimed to compare the accuracy of IFN-τ stimulated gene abundance (ISGs) in peripheral blood mononuclear cells (PBMCs), CL blood perfusion by Doppler ultrasound (Doppler-US), plasma concentration of P4 on Day 21 and pregnancy-associated glycoproteins (PAGs) test on Day 25 after timed-artificial insemination (TAI) for early pregnancy diagnosis in dairy cows and heifers. Holstein cows (n = 140) and heifers (n = 32) were subjected to a hormonal synchronization protocol and TAI on Day 0. On Day 21 post-TAI, blood samples were collected for PBMC isolation and plasma concentration of P4. The CL blood perfusion was evaluated by Doppler-US. Plasma samples collected on Day 25 were assayed for PAGs. The abundance of ISGs (ISG15 and RSAD2) in PBMCs was determined by RT-qPCR. Pregnancy was confirmed on Days 32 and 60 post-TAI by B-mode ultrasonography. Statistical analyses were performed by ANOVA using the MIXED procedure and GLIMMIX in SAS software. The pregnancy biomarkers were used to categorize the females as having undergone late luteolysis (LL); early embryonic mortality (EEM); late embryonic mortality (LEM); or late pregnancy loss (LPL). The abundance of ISGs, CL blood perfusion by Doppler-US, and concentrations of P4 on Day 21, and PAGs test on Day 25 were significant (P < 0.05) predictors of early pregnancy in dairy cows and heifers. Dairy cows had a greater (P = 0.01) occurrence of LL than heifers, but there was no difference (P > 0.1) for EEM, LEM, and LPL in heifers compared to cows. Cows with postpartum reproductive issues had a greater (P = 0.008) rate of LEM and a lesser (P = 0.01) rate of LPL compared to cows without reproductive issues. In summary, the CL blood perfusion by Doppler-US had the highest accuracy and the least number of false negatives, suggesting it is the best predictor of pregnancy on Day 21 post-TAI. The PAGs test was the most reliable indicator of pregnancy status on Day 25 post-TAI in dairy heifers and cows. The application of machine learning, specifically the MARS algorithm, shows promise in enhancing the accuracy of predicting early pregnancies in cows.

Current Trends in Artificial Intelligence and Bovine Mastitis Research: A Bibliometric Review Approach

Mastitis, an important disease in dairy cows, causes significant losses in herd profitability. Accurate diagnosis is crucial for adequate control. Studies using artificial intelligence (AI) models to classify, identify, predict, and diagnose mastitis show promise in improving mastitis control. This bibliometric review aimed to evaluate AI and bovine mastitis terms in the most relevant Scopus-indexed papers from 2011 to 2021. Sixty-two documents were analyzed, revealing key terms, prominent researchers, relevant publications, main themes, and keyword clusters. "Mastitis" and "machine learning" were the most cited terms, with an increasing trend from 2018 to 2021. Other terms, such as "sensors" and "mastitis detection", also emerged. The United States was the most cited country and presented the largest collaboration network. Publications on mastitis and AI models notably increased from 2016 to 2021, indicating growing interest. However, few studies utilized AI for bovine mastitis detection, primarily employing artificial neural network models. This suggests a clear potential for further research in this area.

Behavioral Adaptations in Tropical Dairy Cows: Insights into Calving Day Predictions

This study examined changes in the activity patterns of tropical dairy cows during the transition period to assess their potential for predicting calving days. This study used the AfiTag-II biosensor to monitor activity, rest time, rest per bout, and restlessness ratio in 298 prepartum and 347 postpartum Holstein Friesian cows across three lactation groups (1, 2, and ≥3). The data were analyzed using generalized linear mixed models in SPSS, and five machine learning models, including random forest, decision tree, gradient boosting, Naïve Bayes, and neural networks, were used to predict the calving day, with their performance evaluated via ROC curves and AUC metrics. For all lactations, activity levels peak on the calving day, followed by a gradual return to prepartum levels within two weeks. First-lactation cows displayed the shortest rest duration, with a prepartum rest time of 568.8 ± 5.4 (mean ± SE), which is significantly lower than higher-lactation animals. The random forest and gradient boosting displayed an effective performance, achieving AUCs of 85% and 83%, respectively. These results indicate that temporal changes in activity behavior have the potential to be a useful indicator for calving day prediction, particularly in tropical climates where seasonal variations can obscure traditional prepartum indicators.

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.

Apoptotic-Related MiRNAs Correlated with Functional and Flow Cytometric Parameters in Asthenozoospermic Holstein Bulls After Freeze-Thaw Process

Many cellular processes in spermatozoa, including apoptosis and motility, are regulated by miRNA. Different miRNAs and molecular pathways are involved in asthenozoospermia (AS) conditions, which are thought to be one of the causes of infertility with reduced sperm motility. Thirty-two semen samples from four Holstein bulls with normozoospermia (NS), total motility ≥ 70%, and progressive motility ≥ 60%, and 32 semen samples from four bulls with AS, total motility ≤ 40%, and progressive motility ≤ 32% were used to investigate the function of apoptosis-related miRNAs in the AS group. Samples were then aspirated into a 0.5 mL straw after dilution with a Tris-egg yolk extender and frozen at -196°C. After freezing, semen samples were thawed for 2 weeks at 37°C and sperm kinematic parameters, plasma membrane integrity, acrosome integrity, DNA fragmentation, apoptosis status, and expression of apoptosis-related miRNAs (miR-2114, miR-296-3p, miR-455-3p, and miR345-3p) were evaluated. Our results showed that the functional and flow cytometric parameters of the NS group were significantly better than those of the AS group. In the NS group, miR-455-3pp and miR-2412 were upregulated, while miR-345-3p was downregulated compared with the AS group. In the AS group, miR-296-39, miR-2412, and miR-345-3p levels were strongly correlated with membrane integrity, DNA fragmentation, and apoptosis status. The findings demonstrated that the selected miRNAs based on bioinformatic analysis in AS and NS samples had a substantial association with functional and flow cytometry indicators and may be involved in regulating apoptosis and motility in AS samples.

Prediction of Body Weight by Using PCA-Supported Gradient Boosting and Random Forest Algorithms in Water Buffaloes (Bubalus bubalis) Reared in South-Eastern Mexico

This study aims to use advanced machine learning techniques supported by Principal Component Analysis (PCA) to estimate body weight (BW) in buffalos raised in southeastern Mexico and compare their performance. The first stage of the current study consists of body measurements and the process of determining the most informative variables using PCA, a dimension reduction method. This process reduces the data size by eliminating the complex structure of the model and provides a faster and more effective learning process. As a second stage, two separate prediction models were developed with Gradient Boosting and Random Forest algorithms, using the principal components obtained from the data set reduced by PCA. The performances of both models were compared using R2, RMSE and MAE metrics, and showed that the Gradient Boosting model achieved a better prediction performance with a higher R2 value and lower error rates than the Random Forest model. In conclusion, PCA-supported modeling applications can provide more reliable results, and the Gradient Boosting algorithm is superior to Random Forest in this context. The current study demonstrates the potential use of machine learning approaches in estimating body weight in water buffalos, and will support sustainable animal husbandry by contributing to decision making processes in the field of animal science.

Classification of Daily Body Weight Gains in Beef Calves Using Decision Trees, Artificial Neural Networks, and Logistic Regression

The aim of the present study was to compare the predictive performance of decision trees, artificial neural networks, and logistic regression used for the classification of daily body weight gains in beef calves. A total of 680 pure-breed Simmental and 373 Limousin cows from the largest farm in the West Pomeranian Province, whose calves were fattened between 2014 and 2016, were included in the study. Pre-weaning daily body weight gains were divided into two categories: A-equal to or lower than the weighted mean for each breed and sex and B-higher than the mean. Models were developed separately for each breed. Sensitivity, specificity, accuracy, and area under the curve on a test set for the best model (random forest) were 0.83, 0.67, 0.76, and 0.82 and 0.68, 0.86, 0.78, and 0.81 for the Limousin and Simmental breeds, respectively. The most important predictors were daily weight gains of the dam when she was a calf, daily weight gains of the first calf, sex of the third calf, milk yield at first lactation, birth weight of the third calf, dam birth weight, dam hip height, and second calving season. The selected machine learning models can be used quite effectively for the classification of calves based on their daily weight gains.

High Precision Classification of Resting and Eating Behaviors of Cattle by Using a Collar-Fitted Triaxial Accelerometer Sensor

Cattle are less active than humans. Hence, it was hypothesized in this study that transmitting acceleration signals at a 1 min sampling interval to reduce storage load has the potential to improve the performance of motion sensors without affecting the precision of behavior classification. The behavior classification performance in terms of precision, sensitivity, and the F1-score of the 1 min serial datasets segmented in 3, 4, and 5 min window sizes based on nine algorithms were determined. The collar-fitted triaxial accelerometer sensor was attached on the right side of the neck of the two fattening Korean steers (age: 20 months) and the steers were observed for 6 h on day one, 10 h on day two, and 7 h on day three. The acceleration signals and visual observations were time synchronized and analyzed based on the objectives. The resting behavior was most correctly classified using the combination of a 4 min window size and the long short-term memory (LSTM) algorithm which resulted in 89% high precision, 81% high sensitivity, and 85% high F1-score. High classification performance (79% precision, 88% sensitivity, and 83% F1-score) was also obtained in classifying the eating behavior using the same classification method (4 min window size and an LSTM algorithm). The most poorly classified behavior was the active behavior. This study showed that the collar-fitted triaxial sensor measuring 1 min serial signals could be used as a tool for detecting the resting and eating behaviors of cattle in high precision by segmenting the acceleration signals in a 4 min window size and by using the LSTM classification algorithm.

Symposium review: Use of multiple biological, management, and performance data for the design of targeted reproductive management strategies for dairy cows

As the reproductive efficiency of dairy cattle continues to improve in response to better management and use of technology, novel reproductive management approaches will be required to improve herd performance, profitability, and sustainability. A potential approach currently being explored is targeted reproductive management. This approach consists of identifying cows with different reproductive and performance potential using multiple traditional and novel sources of biological, management, and performance data. Once subgroups of cows that share biological and performance features are identified, reproductive management strategies specifically designed to optimize cow performance, herd profitability, or alternative outcomes of interest are implemented on different subgroups of cows. Tailoring reproductive management to subgroups of cows is expected to generate greater gains in outcomes of interest than if the whole herd is under similar management. Major steps in the development and implementation of targeted reproductive management programs for dairy cattle include identification and validation of robust predictors of reproductive outcomes and cow performance, and the development and on-farm evaluation of reproductive management strategies for optimizing outcomes of interest for subgroups of cows. Predictors of cow performance currently explored for use in targeted management include genomic predictions; behavioral, physiological, and performance parameters monitored by sensor technologies; and individual cow and herd performance records. Once the most valuable predictive sources of variation are identified and their effects quantified, novel analytic methods (e.g., machine learning) for prediction will likely be required. These tools must identify groups of cows for targeted management in real time and with no human input. Despite some encouraging research evidence supporting the development of targeted reproductive management strategies, extensive work is required before widespread implementation by commercial farms.

Over 20 Years of Machine Learning Applications on Dairy Farms: A Comprehensive Mapping Study

Machine learning applications are becoming more ubiquitous in dairy farming decision support applications in areas such as feeding, animal husbandry, healthcare, animal behavior, milking and resource management. Thus, the objective of this mapping study was to collate and assess studies published in journals and conference proceedings between 1999 and 2021, which applied machine learning algorithms to dairy farming-related problems to identify trends in the geographical origins of data, as well as the algorithms, features and evaluation metrics and methods used. This mapping study was carried out in line with PRISMA guidelines, with six pre-defined research questions (RQ) and a broad and unbiased search strategy that explored five databases. In total, 129 publications passed the pre-defined selection criteria, from which relevant data required to answer each RQ were extracted and analyzed. This study found that Europe (43% of studies) produced the largest number of publications (RQ1), while the largest number of articles were published in the Computers and Electronics in Agriculture journal (21%) (RQ2). The largest number of studies addressed problems related to the physiology and health of dairy cows (32%) (RQ3), while the most frequently employed feature data were derived from sensors (48%) (RQ4). The largest number of studies employed tree-based algorithms (54%) (RQ5), while RMSE (56%) (regression) and accuracy (77%) (classification) were the most frequently employed metrics used, and hold-out cross-validation (39%) was the most frequently employed evaluation method (RQ6). Since 2018, there has been more than a sevenfold increase in the number of studies that focused on the physiology and health of dairy cows, compared to almost a threefold increase in the overall number of publications, suggesting an increased focus on this subdomain. In addition, a fivefold increase in the number of publications that employed neural network algorithms was identified since 2018, in comparison to a threefold increase in the use of both tree-based algorithms and statistical regression algorithms, suggesting an increasing utilization of neural network-based algorithms.

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.

Estrus Prediction Models for Dairy Gyr Heifers

Simple Summary

Intraruminal devices are already being used to predict reproductive events in cattle. For this prediction, several models and approaches can be used. The aim of this study was to evaluate changes in rumen reticulum temperature (RRT) and activity (ACT) during estrus in Dairy Gyr heifers and to evaluate different models for estrus prediction. There was an increase in both RRT and ACT in the estrus period compared to the same period on the day before and the day after estrus. Among the mathematical models, Random Forest had the best performance. The present results suggest that RRT and ACT can contribute to the identification of estrus and be of value for improving the reproductive efficiency of Zebu herds in tropical regions.

Abstract

Technological devices are increasingly present in livestock activities, such as identifying the reproductive status of cows. For this, predictive models must be accurate and usable in the productive context. The aims of this study were to evaluate estrus-associated changes in reticulo-rumen temperature (RRT) and activity (ACT) in Dairy Gyr heifers provided by reticulo-rumen boluses and to test the ability of different models for estrus prediction. The RRT and ACT of 45 heifers submitted to estrus synchronization were recorded using reticulo-rumen boluses. The means of RRT and ACT at different time intervals were compared between the day before and the day of estrus manifestation. An analysis of variance of RRT and ACT was performed using mixed models. A second approach employed logistic regression, random forest, and linear discriminant analysis models using RRT, ACT, time of day, and the temperature-humidity index (THI) as predictors. There was an increase in RRT and ACT at estrus (p < 0.05) compared to the same period on the day before and on the day after estrus. The random forest model provided the best performance values with a sensitivity of 51.69% and specificity of 93.1%. The present results suggest that RRT and ACT contribute to the identification of estrus in Dairy Gyr heifers.

The Relationship between Satellite-Derived Vegetation Indices and Live Weight Changes of Beef Cattle in Extensive Grazing Conditions

The live weight (LW) and live weight change (LWC) of cattle in extensive beef production is associated with pasture availability and quality. The remote monitoring of pastures and cattle LWC can be achieved with a combination of satellite imagery and walk-over-weighing (WoW) stations. The objective of the present study is to determine the association, if any, between vegetation indices (VIs) (pasture availability) and the LWC of beef cattle in an extensive breeding operation in Northern Australia. The study also tests a suite of VIs along with variables such as rainfall and Julian day to predict the LWC of breeding cows. The VIs were calculated from Sentinel-2 satellite imagery over a 2-year period from a paddock with 378 cattle. Animal LW was measured remotely using a weighing scale at the water point. The relationship between VIs, the LWC, and LW was assessed using linear mixed-effects regression models and random forest modelling. Findings demonstrate that all VIs calculated had a significant positive relationship with the LWC and LW (p < 0.001). Machine learning predictive modelling showed that the LWC of breeding cows could be predicted from VIs, Julian day, and rainfall information, with a Lin’s Concordance Correlation Coefficient of 0.62 when using the leave-one-month-out cross-validation. The LW and LWC were greater during the wet season when VIs were higher compared to the dry season (p < 0.001). Results suggest that the remote monitoring of pasture availability, the LWC and LW is possible under extensive grazing conditions. Further, the use of VIs and other readily available data such as rainfall can be used to predict the LWC of a breeding herd in extensive conditions. Such information could be used to increase the productivity and land management in extensive beef production. The integration of these data streams offers great potential to improve the monitoring, management, and productivity of grazing or cropping enterprises.

Loop-mediated isothermal amplification (LAMP) and machine learning application for early pregnancy detection using bovine vaginal mucosal membrane

An early and accurate pregnancy diagnosis method is required to improve the reproductive performance of cows. Here we developed an easy pregnancy detection method using vaginal mucosal membrane (VMM) with application of Reverse Transcription-Loop-mediated Isothermal Amplification (RT-LAMP) and machine learning. Cows underwent artificial insemination (AI) on day 0, followed by VMM-collection on day 17-18, and pregnancy diagnosis by ultrasonography on day 30. By RNA sequencing of VMM samples, three candidate genes for pregnancy markers (ISG15 and IFIT1: up-regulated, MUC16: down-regulated) were selected. Using these genes, we performed RT-LAMP and calculated the rise-up time (RUT), the first-time absorbance exceeded 0.05 in the reaction. We next determined the cutoff value and calculated accuracy, sensitivity, specificity, positive prediction value (PPV), and negative prediction value (NPV) for each marker evaluation. The IFIT1 scored the best performance at 92.5% sensitivity, but specificity was 77.5%, suggesting that it is difficult to eliminate false positives. We then developed a machine learning model trained with RUT of each marker combination to predict pregnancy. The model created with the RUT of IFIT1 and MUC16 combination showed high specificity (86.7%) and sensitivity (93.3%), which were higher compared to IFIT1 alone. In conclusion, using VMM with RT-LAMP and machine learning algorithm can be used for early pregnancy detection before the return of first estrus.

Predicting sperm production of young dairy bulls using collection history and management factors

Selection of elite young dairy bulls by using genomic data shortened the generation interval and increased pressure to collect and market germplasm at an early age. The objectives of this study were (1) develop prediction models for daily, weekly, and monthly total sperm (TSp) production from collection history, health status, and management factors, and (2) assess the ability of these models to forecast future TSp production, as well as differences in prediction accuracy by seasonality or age of bull. Data consisted of 43,918 daily processing records from 1,037 Holstein and Jersey bulls between 10 and 28 mo of age at collection. Potential explanatory variables included year and season of collection, barn location, collection frequency, breed, scrotal circumference, TSp in previous months, health events, and age at arrival, first collection, and current collection. Linear regression, random forest (RF), Bayesian regularized neural network, model tree, multilayer perceptron neural network with multiple layers, and extreme learning machine were used to predict daily, weekly, and monthly TSp (R v3.5.1, https://www.r-project.org/). In the additive approach, all prior data were used for training; however, in the fixed-window approach, records from 3 previous months were used for age-based prediction, records from 4 previous months or 1 yr were used for the monthly date-based analyses, and records from 1 previous month or year were used for the weekly date-based analyses. Model performance was measured by root mean squared error (RMSE) and the correlation (r) between actual and predicted TSp in testing sets. In monthly analyses, RF with additive training performed best in age-based (RMSE = 13.6 billion cells, r = 0.93) and date-based (RMSE = 11.9, r = 0.94) prediction, compared with linear regression (age-based RMSE = 16.6, r = 0.89; date-based RMSE = 15.5, r = 0.90) and Bayesian regularized neural network (age-based RMSE = 14.1, r = 0.92). On average, RMSE was 0.93 or 0.14 billion cells greater with fixed 4-mo or 1-yr training windows, respectively, than in the additive analyses. The most important management variables affecting TSp were collection frequency, TSp in previous months, and age at collection. Results indicate RF models with additive training can predict TSp output of individual bulls with ≥85% accuracy up to 4 mo into the future. Spikes in accuracy were associated with sire summary times and company processing changes, and accuracy tended to stabilize when bulls reached 19 to 20 mo of age.

Predicting dairy cattle heat stress using machine learning techniques

The objectives of the study were to use a heat stress scoring system to evaluate the severity of heat stress on dairy cows using different heat abatement techniques. The scoring system ranged from 1 to 4, where 1 = no heat stress; 2 = mild heat stress; 3 = severe heat stress; and 4 = moribund. The accuracy of the scoring system was then predicted using 3 machine learning techniques: logistic regression, Gaussian naïve Bayes, and random forest. To predict the accuracy of the scoring system, these techniques used factors including temperature-humidity index, respiration rate, lying time, lying bouts, total steps, drooling, open-mouth breathing, panting, location in shade or sprinklers, somatic cell score, reticulorumen temperature, hygiene body condition score, milk yield, and milk fat and protein percent. Three different treatments, namely, portable shade structure, portable polyvinyl chloride pipe sprinkler system, or control with no heat abatement, were considered, where each treatment was replicated 3 times with 3 second-trimester lactating cows. Results indicate that random forest outperformed the other 2 methods, with respect to both accuracy and precision, in predicting the sprinkler group's score. Both logistic regression and random forest were consistent in predicting scores for control, shade, and combined groups. The mean probability of predicting non-heat-stressed cows was highest for cows in the sprinkler group. Finally, the logistic regression method worked best for predicting heat-stressed cows in control, shade, and combined. The insights gained from these results could aid dairy producers to detect heat stress before it becomes severe, which could decrease the negative effects of heat stress, such as milk loss.

Predicting the likelihood of conception to first insemination of dairy cows using milk mid-infrared spectroscopy

The objective of this study was to examine the ability of milk mid-infrared (MIR) spectroscopy and other on-farm data, such as milk yield, milk composition, stage of lactation, calving age, days in milk at insemination, and somatic cell count, to identify cows that were most or least likely to conceive to first insemination. A total of 16,628 spectral and milk production records of 7,040 cows from 29 commercial dairy herds across 3 Australian states were used. Three models, comprising different explanatory variables, were tested. Model 1 included features that are readily available on farms participating in milk recording, such as milk yield, milk composition, somatic cell count, days from calving to insemination, and calving season. Days in milk and age at calving were incorporated into model 1 to form model 2. In model 3, MIR was added to model 2, but to avoid double counting, milk composition traits of model 2 were removed. The models were first trained on extreme data [i.e., including cows that (1) conceived to first insemination and (2) cows with no conception event recorded and with only 1 insemination]. Then, the models were validated in a fresh data set with all cows regardless of conception outcomes present to test for their ability to identify cows that conceived or did not conceive to first insemination. To do this, we ranked the predicted probability of all cows in the validation set and then selected the top and bottom records in varying proportions from 5 to 40% (i.e., where the model predicted the highest versus lowest likelihood of conception to first insemination, respectively) and compared with the actual values. The model's performance was evaluated through herd-year by herd-year external validation and measured as the proportion of selected records being correct. The results show that when more cows are selected (i.e., descending confidence), the accuracy of the models was reduced, and selecting the 10% of cows with the highest confidence of predictions produces optimal accuracy. Irrespective of the proportions, none of the models could predict cows that conceived to first insemination, with an accuracy around 0.48. When attempting to predict the bottom 10% of cows, which had the least likelihood of conception to first insemination, model 1 had prediction accuracy around 0.64. Compared with model 1, the addition of days in milk and calving age (model 2) resulted in a negligible improvement in prediction accuracy (0.01 to 0.03). Model 3 had the highest prediction accuracy (0.76), which implies that in the models tested, MIR is of primary importance in the prediction of fertility of dairy cows. In conclusion, this study indicates that MIR and other milk recording data could be used to identify cows with potential difficulty in getting pregnant to first insemination with promising accuracy.

Analysis of Cattle Social Transitional Behaviour: Attraction and Repulsion

Understanding social interactions in livestock groups could improve management practices, but this can be difficult and time-consuming using traditional methods of live observations and video recordings. Sensor technologies and machine learning techniques could provide insight not previously possible. In this study, based on the animals' location information acquired by a new cooperative wireless localisation system, unsupervised machine learning approaches were performed to identify the social structure of a small group of cattle yearlings (n=10) and the social behaviour of an individual. The paper first defined the affinity between an animal pair based on the ranks of their distance. Unsupervised clustering algorithms were then performed, including K-means clustering and agglomerative hierarchical clustering. In particular, K-means clustering was applied based on logical and physical distance. By comparing the clustering result based on logical distance and physical distance, the leader animals and the influence of an individual in a herd of cattle were identified, which provides valuable information for studying the behaviour of animal herds. Improvements in device robustness and replication of this work would confirm the practical application of this technology and analysis methodologies.

Using Sensor Data to Detect Lameness and Mastitis Treatment Events in Dairy Cows: A Comparison of Classification Models

The aim of this study was to develop classification models for mastitis and lameness treatments in Holstein dairy cows as the target variables based on continuous data from herd management software with modern machine learning methods. Data was collected over a period of 40 months from a total of 167 different cows with daily individual sensor information containing milking parameters, pedometer activity, feed and water intake, and body weight (in the form of differently aggregated data) as well as the entered treatment data. To identify the most important predictors for mastitis and lameness treatments, respectively, Random Forest feature importance, Pearson’s correlation and sequential forward feature selection were applied. With the selected predictors, various machine learning models such as Logistic Regression (LR), Support Vector Machine (SVM), K-nearest neighbors (KNN), Gaussian Naïve Bayes (GNB), Extra Trees Classifier (ET) and different ensemble methods such as Random Forest (RF) were trained. Their performance was compared using the receiver operator characteristic (ROC) area-under-curve (AUC), as well as sensitivity, block sensitivity and specificity. In addition, sampling methods were compared: Over- and undersampling as compensation for the expected unbalanced training data had a high impact on the ratio of sensitivity and specificity in the classification of the test data, but with regard to AUC, random oversampling and SMOTE (Synthetic Minority Over-sampling) even showed significantly lower values than with non-sampled data. The best model, ET, obtained a mean AUC of 0.79 for mastitis and 0.71 for lameness, respectively, based on testing data from practical conditions and is recommended by us for this type of data, but GNB, LR and RF were only marginally worse, and random oversampling and SMOTE even showed significantly lower values than without sampling. We recommend the use of these models as a benchmark for similar self-learning classification tasks. The classification models presented here retain their interpretability with the ability to present feature importances to the farmer in contrast to the “black box” models of Deep Learning methods.

A review of traditional and machine learning methods applied to animal breeding

The current livestock management landscape is transitioning to a high-throughput digital era where large amounts of information captured by systems of electro-optical, acoustical, mechanical, and biosensors is stored and analyzed on a daily and hourly basis, and actionable decisions are made based on quantitative and qualitative analytic results. While traditional animal breeding prediction methods have been used with great success until recently, the deluge of information starts to create a computational and storage bottleneck that could lead to negative long-term impacts on herd management strategies if not handled properly. A plethora of machine learning approaches, successfully used in various industrial and scientific applications, made their way in the mainstream approaches for livestock breeding techniques, and current results show that such methods have the potential to match or surpass the traditional approaches, while most of the time they are more scalable from a computational and storage perspective. This article provides a succinct view on what traditional and novel prediction methods are currently used in the livestock breeding field, how successful they are, and how the future of the field looks in the new digital agriculture era.

Assessment of associations between transition diseases and reproductive performance of dairy cows using survival analysis and decision tree algorithms

This study aimed to evaluate the associations between transition cow conditions and diseases TD with fertility in Holstein cows, and to compare analytic methods for doing so. Kaplan-Meier, Cox proportional hazard, and decision tree models were used to analyze the associations of TD with the pregnancy risk at 120 and 210 DIM from a 1-year cohort with 1946 calvings from one farm. The association between TD and fertility was evaluated as follows: 1 cows with TD whether complicated with another TD or not TD-all, versus healthy cows, and 2 cows with uncomplicated TD TD-single, versus cows with multiple TD TD+; complicated cases, versus healthy cows. The occurrence of twins, milk fever, retained placenta, metritis, ketosis, displaced abomasum, and clinical mastitis were recorded. Using Kaplan-Meier models, in primiparous cows the 120 DIM pregnancy risk was 62% (95% CI: 57-67 %) for healthy animals. This was not significantly different for TD-single (58%; 95% CI: 51-66 %) but was reduced for TD+ (45%; 95% CI: 33-60 %). Among healthy primiparous cows, 80% (95% CI: 75-84 %) were pregnant by 210 DIM, but pregnancy risk at that time was reduced for primiparous cows with TD-single (72%; 95% CI: 65-79 %) and TD+ (62%; 95% CI: 49-75 %). In healthy multiparous cows, the 120 DIM pregnancy risk was 53% (95% CI: 49-56 %), which was reduced for TD-single (36%; 95% CI: 31-42 %) and TD+ (30%; 95% CI: 24-38 %). The 210 DIM pregnancy risk for healthy multiparous cows was 70% (95% CI: 67-72 %), being higher than the 210 DIM pregnancy risk for multiparous cows with TD-single (47%; 95% CI: 42-53 %) or TD+ (46%; 95% CI: 38-54 %). Cows with TD-all presented similar pregnancy risk estimates as for TD + . Cox proportional hazards regressions provided similar magnitudes of effects as the Kaplan-Meier estimates. Survival analysis and decision tree models identified parity as the most influential variable affecting fertility. Both modeling techniques concurred that TD + had a greater effect than TD-single on the probability of pregnancy at 120 and 210 DIM. Decision trees for individual TD identified that displaced abomasum affected fertility at 120 DIM in primiparous while metritis was the most influential TD at 120 and 210 DIM for multiparous cows. The data were too sparse to assess multiple interactions in multivariable Cox proportional hazard models for individual TD. Machine learning helped to explore interactions between individual TD to study their hierarchical effect on fertility, identifying conditional relationships that merit further investigation.

Boosted trees to predict pneumonia, growth, and meat percentage of growing-finishing pigs1

In pig production, efficiency is benefiting from uniform growth in pens resulting in single deliveries from a pen of possibly all animals in the targeted weight range. Abnormalities, like pneumonia or aberrant growth, reduce production efficiency as it reduces the uniformity and might cause multiple deliveries per batch and pigs delivered with a low meat yield or outside the targeted weight range. Early identification of pigs prone to develop these abnormalities, for example, at the onset of the growing-finishing phase, would help to prevent heterogeneous pens through management interventions. Data about previous production cycles at the farm combined with data from the piglet’s own history may help in identifying these abnormalities. The aim of this study, therefore, was to predict at the onset of the growing-finishing phase, that is, at 3 mo in advance, deviant pigs at slaughter with a machine-learning technique called boosted trees. The dataset used was extracted from the farm management system of a research center. It contained over 70,000 records of individual pigs born between 2004 and 2016, including information on, for example, offspring, litter size, transfer dates between production stages, their respective locations within the barns, and individual live-weights at several production stages. Results obtained on an independent test set showed that at a 90% specificity rate, the sensitivity was 16% for low meat percentage, 20% for pneumonia and 36% for low lifetime growth rate. For low lifetime growth rate, this meant an almost three times increase in positive predictive value compared to the current situation. From these results, it was concluded that routine performance information available at the onset of the growing-finishing phase combined with data about previous production cycles formed a moderate base to identify pigs prone to develop pneumonia (AUC > 0.60) and a good base to identify pigs prone to develop growth aberrations (AUC > 0.70) during the growing-finishing phase. The mentioned information, however, was not a sufficient base to identify pigs prone to develop low meat percentage (AUC < 0.60). The shown ability to identify growth aberrations and pneumonia can be considered a good first step towards the development of an early warning system for pigs in the growing-finishing phase.

Machine learning algorithms, bull genetic information, and imbalanced datasets used in abortion incidence prediction models for Iranian Holstein dairy cattle

The ability to predict abortion incidence, especially in regions with high abortion rates (e.g., Iran), helps improve reproductive performance and, thereby, dairy farm profitability. The objective of this study was to predict pregnancy loss in Iranian dairy herds. For this purpose, the cow history records and bull genetic information available at 6 large commercial dairy farms with cows calved between 2005 and 2014 were extracted from an on-farm record-keeping software. Using WEKA, 12 commonly used machine learning (ML) algorithms were applied to the dataset. The algorithms belonged to 5 classifier groups which were Bayes, meta, functions, rules, and trees. The original dataset including herd-cow factors was randomly divided into 2 subsets: a training dataset and a test one (at a ratio of 60:40). The original dataset was combined with the bull genetic information to create a full dataset. The average abortion rate was 15.4 %, which represented an imbalanced dataset. Therefore, 2 down- and up-sampling techniques were additionally implemented on the original dataset (more specifically on the training one) to create 2 balanced datasets. This ultimately yielded 4 datasets; original, full, down-sampling, and up-sampling. Different algorithms and models were evaluated based on F-measure and area under the curve (AUC). Based on the results obtained, ML algorithms exhibited a high performance in predicting abortion when applied to the balanced dataset. However, their performance varied from 32.3 % (poor) to 69.2 % (medium upward) when applied to the imbalanced original dataset. In addition to the imbalance in the original dataset, the reason for these poor results were attributed to the high proportion of unknown risk factors underlying abortion incidence. Even when including the bull genetic information, it did not lead to any significant improvements in the prediction model. From among the datasets used, the Bayes algorithms outperformed the others in predicting pregnancy losses while rules had the worst performance. Furthermore, while the Bayes algorithms were not affected by the type of dataset (balanced or imbalanced), substantial increases in F-measure and AUC were observed for rules, trees, and functions with balanced datasets. Overall, the balanced models outperformed the others, with the down-sampling method exhibiting the highest performance. Despite the fact that the prediction models used in this study did not perform as expected, it was shown that they can be beneficially used to predict and reduce pregnancy losses, despite their moderate accuracy, especially when used for herds with high abortion rates and low reproductive performances.

A comparison of machine learning and logistic regression in modelling the association of body condition score and submission rate

The effect of body condition score (BCS) on reproductive outcomes is complex, dynamic and non-linear with interaction and confounding. The flexibility inherent in machine learning algorithms makes them attractive for analysing complex data. This study was designed to compare the ability of a range of machine learning techniques in estimating the probability of service within 21 days of the planned start of mating. We hypothesised that if there were complex and unknown interactions or non-linearity in the data, some machine learning algorithms would result in superior model performance compared to regression models. For a period of six months from the planned start of calving, BCS was visually assessed once a month for 6127 cows on 8 commercial New Zealand dairy farms by a trained veterinarian using the DairyNZ 10-point range for every cow in the herd. Cow, lactation and reproductive data was extracted from the national herd database. This data was used to predict probability of service within 21 days of planned start of mating (PSM) using mixed multivariable logistic regression and decision tree, k-nearest neighbour, random forest and neural network analysis. Models were adjusted for herd, cow age, breed, days in milk, BCS at calving, BCS change between calving and mating, BCS change after mating, volume adjusted milk protein and fat concentration pre-mating. Models were constructed on a training data set using 10-fold cross validation repeated 10 times and evaluated on a test data set using discrimination and calibration techniques. In all models, days calved at PSM was the most important variable for predicting submission rate, followed by BCS at PSM. Factors associated with an increased probability of insemination were calving at a BCS of 5.0, losing less BCS after calving, having a higher BCS at nadir, losing BCS rapidly after calving, nadir occurring before PSM and calving early. All the models except for the decision tree had an area under the receiver operating characteristic curve (AUC) in the range 0.68-0.73 indicating good overall discriminatory power, but calibration analysis suggested all models were better at predicting cows that got inseminated than correctly identifying animals that did not get inseminated. Overall, the machine learning techniques were no better than a generalised logistic regression model. These results highlight the importance of BCS targets at calving and indicate BCS loss, milk characteristics and days calves may be useful indicators identifying cows at risk of poor reproductive outcomes.

Multicolor flow cytometric analysis of cryopreserved bovine sperm: A tool for the evaluation of bull fertility

The study aimed at the analysis of the functional status of cryopreserved bovine sperm using multicolor flow cytometry. The value of sperm functional traits as predictors of bull fertility was further evaluated through a retrospective fertility study. For this purpose, 20 Holstein-Friesian bulls serving as mature sperm donors in an artificial insemination (AI) center were selected based on their annual 56-d non-return rate (%) after at least 1,000 AI, and were accordingly classified as high (HF; n_HF = 10 bulls) or low fertility bulls (LF; n_LF = 10 bulls). Four to 5 cryopreserved ejaculates per bull (91 ejaculates in total) were examined immediately after thawing (0 h) and after a 3-h incubation at 38°C (3 h). A panel of 5 fluorochromes including calcein violet, propidium iodide, pycoerythrin-conjugated lectin of Arachis hypogea, Fluo-4, and cyanine dye DiIC₁(5) was configured by means of a 3-laser flow cytometer, to simultaneously assess sperm esterase activity, plasma membrane integrity, acrosomal status, intracellular Ca²⁺ levels, and mitochondrial membrane potential, respectively. The % relative size of 18 sperm sub-populations showing 2 or more of a combination of the following features was determined: high esterase activity (C_pos), intact plasma membrane (PI_neg), unstained acrosome (PNA_neg), low intracellular Ca²⁺ levels (F_neg), and high mitochondrial membrane potential (M_pos). In both fertility groups, M_pos cells comprised more than 90 and 84% of PI_negPNA_neg sperm at 0 and 3 h, respectively. The percentage of C_posPI_negPNA_negF_negM_pos sperm did not differ between HF and LF ejaculates; however, the percentage of F_neg cells within the PI_negPNA_neg and PI_negM_pos sperm populations at 0 h was higher in the HF than in the LF bulls. Applying the random forest ensemble learning method, approximately two-thirds of ejaculates could be correctly assigned to their fertility group. The fraction of F_neg sperm within the PI_negM_pos population at 0 h was the most important fertility predictor among the 18 defined sperm populations. In conclusion, multicolor flow cytometry offered an insight into the functional heterogeneity of cryopreserved bovine sperm. Indeed, the ability of viable sperm to retain low Ca²⁺ levels differed between bulls of diverse fertility. A classifier based on selected sperm populations assessed through multicolor flow cytometry could contribute to the prognosis of bull fertility after AI.

Classifying the fertility of dairy cows using milk mid-infrared spectroscopy

The objective of this study was to investigate the potential of milk mid-infrared (MIR) spectroscopy, MIR-derived traits including milk composition, milk fatty acids, and blood metabolic profiles (fatty acids, β-hydroxybutyrate, and urea), and other on-farm data for discriminating cows of good versus poor likelihood of conception to first insemination (i.e., pregnant vs. open). A total of 6,488 spectral and milk production records of 2,987 cows from 19 commercial dairy herds across 3 Australian states were used. Seven models, comprising different explanatory variables, were examined. Model 1 included milk production; concentrations of fat, protein, and lactose; somatic cell count; age at calving; days in milk at herd test; and days from calving to insemination. Model 2 included, in addition to the variables in model 1, milk fatty acids and blood metabolic profiles. The MIR spectrum collected before first insemination was added to model 2 to form model 3. Fat, protein, and lactose percentages, milk fatty acids, and blood metabolic profiles were removed from model 3 to create model 4. Model 5 and model 6 comprised model 4 and either fertility genomic estimated breeding value or principal components obtained from a genomic relationship matrix derived using animal genotypes, respectively. In model 7, all previously described sources of information, but not MIR-derived traits, were used. The models were developed using partial least squares discriminant analysis. The performance of each model was evaluated in 2 ways: 10-fold random cross-validation and herd-by-herd external validation. The accuracy measures were sensitivity (i.e., the proportion of pregnant cows that were correctly classified), specificity (i.e., the proportion of open cows that were correctly classified), and area under the curve (AUC) for the receiver operating curve. The results showed that in all models, prediction accuracy obtained through 10-fold random cross-validation was higher than that of herd-by-herd external validation, with the difference in AUC ranging between 0.01 and 0.09. In the herd-by-herd external validation, using basic on-farm information (model 1) was not sufficient to classify good- and poor-fertility cows; the sensitivity, specificity, and AUC were around 0.66. Compared with model 1, adding milk fatty acids and blood metabolic profiles (model 2) increased the sensitivity, specificity, and AUC by 0.01, 0.02, and 0.02 unit, respectively (i.e., 0.65, 0.63, and 0.678). Incorporating MIR spectra into model 2 resulted in sensitivity, specificity, and AUC values of 0.73, 0.63, and 0.72, respectively (model 3). The comparable prediction accuracies observed for models 3 and 4 mean that useful information from MIR-derived traits is already included in the spectra. Adding the fertility genomic estimated breeding value and animal genotypes (model 7) produced the highest prediction accuracy, with sensitivity, specificity, and AUC values of 0.75, 0.66, and 0.75, respectively. However, removing either the fertility estimated breeding value or animal genotype from model 7 resulted in a reduction of the prediction accuracy of only 0.01 and 0.02, respectively. In conclusion, this study indicates that MIR and other on-farm data could be used to classify cows of good and poor likelihood of conception with promising accuracy.

Prediction of metabolic status of dairy cows in early lactation with on-farm cow data and machine learning algorithms

Metabolic status of dairy cows in early lactation can be evaluated using the concentrations of plasma β-hydroxybutyrate (BHB), free fatty acids (FFA), glucose, insulin, and insulin-like growth factor 1 (IGF-1). These plasma metabolites and metabolic hormones, however, are difficult to measure on farm. Instead, easily obtained on-farm cow data, such as milk production traits, have the potential to predict metabolic status. Here we aimed (1) to investigate whether metabolic status of individual cows in early lactation could be clustered based on their plasma values and (2) to evaluate machine learning algorithms to predict metabolic status using on-farm cow data. Through lactation wk 1 to 7, plasma metabolites and metabolic hormones of 334 cows were measured weekly and used to cluster each cow into 1 of 3 clusters per week. The cluster with the greatest plasma BHB and FFA and the lowest plasma glucose, insulin, and IGF-1 was defined as poor metabolic status; the cluster with the lowest plasma BHB and FFA and the greatest plasma glucose, insulin, and IGF-1 was defined as good metabolic status; and the intermediate cluster was defined as average metabolic status. Most dairy cows were classified as having average or good metabolic status, and a limited number of cows had poor metabolic status (10-50 cows per lactation week). On-farm cow data, including dry period length, parity, milk production traits, and body weight, were used to predict good or average metabolic status with 8 machine learning algorithms. Random Forest (error rate ranging from 12.4 to 22.6%) and Support Vector Machine (SVM; error rate ranging from 12.4 to 20.9%) were the top 2 best-performing algorithms to predict metabolic status using on-farm cow data. Random Forest had a higher sensitivity (range: 67.8-82.9% during wk 1 to 7) and negative predictive value (range: 89.5-93.8%) but lower specificity (range: 76.7-88.5%) and positive predictive value (range: 58.1-78.4%) than SVM. In Random Forest, milk yield, fat yield, protein percentage, and lactose yield had important roles in prediction, but their rank of importance differed across lactation weeks. In conclusion, dairy cows could be clustered for metabolic status based on plasma metabolites and metabolic hormones. Moreover, on-farm cow data can predict cows in good or average metabolic status, with Random Forest and SVM performing best of all algorithms.

Comparing regression, naive Bayes, and random forest methods in the prediction of individual survival to second lactation in Holstein cattle

In this study, we compared multiple logistic regression, a linear method, to naive Bayes and random forest, 2 nonlinear machine-learning methods. We used all 3 methods to predict individual survival to second lactation in dairy heifers. The data set used for prediction contained 6,847 heifers born between January 2012 and June 2013, and had known survival outcomes. Each animal had 50 genomic estimated breeding values available at birth and up to 65 phenotypic variables that accumulated over time. Survival was predicted at 5 moments in life: at birth, at 18 mo, at first calving, at 6 wk after first calving, and at 200 d after first calving. The data sets were randomly split into 70% training and 30% testing sets to evaluate model performance for 20-fold validation. The methods were compared for accuracy, sensitivity, specificity, area under the curve (AUC) value, contrasts between groups for the prediction outcomes, and increase in surviving animals in a practical scenario. At birth and 18 mo, all methods had overlapping performance; no method significantly outperformed the other. At first calving, 6 wk after first calving, and 200 d after first calving, random forest and naive Bayes had overlapping performance, and both machine-learning methods outperformed multiple logistic regression. Overall, naive Bayes has the highest average AUC at all decision points up to 200 d after first calving. Random forest had the highest AUC at 200 d after first calving. All methods obtained similar increases in survival in the practical scenario. Despite this, the methods appeared to predict the survival of individual heifers differently. All methods improved over time, but the changes in mean model outcomes for surviving and non-surviving animals differed by method. Furthermore, the correlations of individual predictions between methods ranged from r = 0.417 to r = 0.700; the lowest correlations were at first calving for all methods. In short, all 3 methods were able to predict survival at a population level, because all methods improved survival in a practical scenario. However, depending on the method used, predictions for individual animals were quite different between methods.

Physiological characteristics and effects on fertility of the first follicular wave dominant follicle in cattle

The first follicular wave emerges soon after ovulation, and its dominant follicle (DF) develops during the first 8–10 days of the estrous cycle in cattle. And, the first-wave DF is a non-ovulatory follicle, because it develops during the first half of the estrous cycle simultaneously with the corpus luteum (CL), which produces and secretes progesterone. Regarding the characteristics of development and the mechanisms of deviation in the DF during the follicular wave, the first-wave DF has been well studied. However, the characteristics of the first-wave DF, such as growth, blood flow in the follicular wall, concentration of sex steroid hormones in the peripheral blood and follicular fluid, amounts of mRNA in granulosa cells, as well as the characteristics of the CL formed after the first-wave DF and the influence of the first-wave DF on fertility (conception rate), have not been well studied. Additionally, the first-wave DF synthesizes and secretes 17β-estradiol (E₂), and plasma E₂ concentration increases during the early stage of the estrous cycle. Consequently, there is a possibility that the first-wave DF might affect the fertility in cattle. In this review, to provide the new perspective on reproductive physiology in cattle, characteristics of the first-wave DF were examined in detail and its characteristics were compared with that of the second-wave DF. In addition, the locational effects of the first-wave DF and CL on conception rate are discussed.

Prediction of slaughter age in pigs and assessment of the predictive value of phenotypic and genetic information using random forest

The weight of a pig and the rate of its growth are key elements in pig production. In particular, predicting future growth is extremely useful, since it can help in determining feed costs, pen space requirements, and the age at which a pig reaches a desired slaughter weight. However, making these predictions is challenging, due to the natural variation in how individual pigs grow, and the different causes of this variation. In this paper, we used machine learning, namely random forest (RF) regression, for predicting the age at which the slaughter weight of 120 kg is reached. Additionally, we used the variable importance score from RF to quantify the importance of different types of input data for that prediction. Data of 32,979 purebred Large White pigs were provided by Topigs Norsvin, consisting of phenotypic data, estimated breeding values (EBVs), along with pedigree and pedigree-genetic relationships. Moreover, we presented a 2-step data reduction procedure, based on random projections (RPs) and principal component analysis (PCA), to extract features from the pedigree and genetic similarity matrices for use as inputs in the prediction models. Our results showed that relevant phenotypic features were the most effective in predicting the output (age at 120 kg), explaining approximately 62% of its variance (i.e., R2 = 0.62). Estimated breeding value, pedigree, or pedigree-genetic features interchangeably explain 2% of additional variance when added to the phenotypic features, while explaining, respectively, 38%, 39%, and 34% of the variance when used separately.

Effects of human chorionic gonadotropin treatment after artificial inseminations on conception rate with the first follicular wave dominant follicle in the ovary ipsilateral to the corpus luteum in lactating dairy cows

We examined the effect of human chorionic gonadotropin (hCG) treatment 5 days after artificial insemination (AI) on conception rate when the first-wave dominant follicle (DF) in the ovaries was either ipsilateral or contralateral to the corpus luteum (CL) in lactating dairy cows. 577 cows from 4 dairy farms were divided into the following two groups 5 days after AI using transrectal ultrasonography: (1) the ipsilateral group (IG; n = 348), in which the DF was ipsilateral to the CL, and (2) the contralateral group (CG; n = 229), in which the DF was contralateral to the CL. IG and CG were further subdivided into two groups: non-treatment groups, which received no treatment (IG, n = 220; CG, n = 128), and hCG treatment group, that was administrated 1500 IU hCG 5 days after AI (IG, n = 143; CG, n = 86). Pregnancy was diagnosed by rectal palpation or transrectal ultrasonography from 53 to 67 days after AI. Conception rate was significantly (P < 0.01) higher in the hCG treatment group of IG (40.6%) than in the non-treatment group of IG (21.4%); however, there was no difference in the non-treatment (51.7%) and hCG treatment (43.0%) groups of CG. Parity, farm, days in milk at AI, interaction between the farm and hCG treatment and interaction between the farm and location of the first-wave DF and CL did not affect conception rate. Our results suggest that conception rate can be improved by administrating hCG only to cows with the first wave DF ipsilateral to the CL.

Machine-learning-based calving prediction from activity, lying, and ruminating behaviors in dairy cattle

The objective of this study was to use automated activity, lying, and rumination monitors to characterize prepartum behavior and predict calving in dairy cattle. Data were collected from 20 primiparous and 33 multiparous Holstein dairy cattle from September 2011 to May 2013 at the University of Kentucky Coldstream Dairy. The HR Tag (SCR Engineers Ltd., Netanya, Israel) automatically collected neck activity and rumination data in 2-h increments. The IceQube (IceRobotics Ltd., South Queensferry, United Kingdom) automatically collected number of steps, lying time, standing time, number of transitions from standing to lying (lying bouts), and total motion, summed in 15-min increments. IceQube data were summed in 2-h increments to match HR Tag data. All behavioral data were collected for 14 d before the predicted calving date. Retrospective data analysis was performed using mixed linear models to examine behavioral changes by day in the 14 d before calving. Bihourly behavioral differences from baseline values over the 14 d before calving were also evaluated using mixed linear models. Changes in daily rumination time, total motion, lying time, and lying bouts occurred in the 14 d before calving. In the bihourly analysis, extreme values for all behaviors occurred in the final 24 h, indicating that the monitored behaviors may be useful in calving prediction. To determine whether technologies were useful at predicting calving, random forest, linear discriminant analysis, and neural network machine-learning techniques were constructed and implemented using R version 3.1.0 (R Foundation for Statistical Computing, Vienna, Austria). These methods were used on variables from each technology and all combined variables from both technologies. A neural network analysis that combined variables from both technologies at the daily level yielded 100.0% sensitivity and 86.8% specificity. A neural network analysis that combined variables from both technologies in bihourly increments was used to identify 2-h periods in the 8 h before calving with 82.8% sensitivity and 80.4% specificity. Changes in behavior and machine-learning alerts indicate that commercially marketed behavioral monitors may have calving prediction potential.

Estrous detection by monitoring ventral tail base surface temperature using a wearable wireless sensor in cattle

In the present study, the ventral tail base surface temperature (ST) was monitored using a wearable wireless sensor for estrus detection in cattle. Relationships among ST, behavioral estrus expression, ovulation, and changes in hormone profiles during the estrous cycle were examined. Holstein Friesian or Japanese Black female cattle were used in summer (August-September), autumn (October-November) and winter (January-February; three animals per season). On Day 11 of the estrous cycle (Day 0=the day of ovulation), the sensor was attached to the surface of the ventral tail base and ST was measured every 2min until Day 11 of the next estrous cycle. Hourly maximum ST values were used for analysis. To exclude circadian rhythm and seasonal effects, ST changes were expressed as residual temperatures (RT=actual ST - mean ST for the same hour on the previous 3days). Obvious circadian rhythms of the ST were observed and daily changes in the ST significantly differed among seasons. There was no significant seasonal difference, however, in the RT. The mean RT increased significantly ∼24 compared with ∼48h before ovulation. The mean maximum RT was 1.27±0.30°C, which was observed 5.6±2.4h after the onset of estrus, 2.4±1.3h before LH peak, and 26.9±1.2h before ovulation. The ST of the ventral tail base could be monitored throughout the estrous cycle and could detect a substantial change around the time of expression of behavioral estrus. Calculation and analysis of the RT could be useful for automatic estrous detection.

The Importance of the Periconception Period: Immediate Effects in Cattle Breeding and in Assisted Reproduction Such as Artificial Insemination and Embryo Transfer

In livestock breeding, the successful outcome is largely depending on the "periconception environment" which, in a narrow sense, refers to the genital tract, where gametogenesis and embryogenesis occur. During these early stages of development, gametes and embryos are known to be particularly sensitive to alterations in their microenvironment. However, as the microenvironment somehow reflects what is going on in the external world, we must widen our definition of "periconception environment" and refer to all events taking place around the time of conception, including metabolic state and health and nutrition of the dam. In modern dairy cows that have to manage an optimal reproductive performance with continued growth and high milk yield, the periconception period is particularly challenging. The metabolic priority for growth and lactation is known to generate adverse conditions hampering optimal ovarian function, oocyte maturation, and development of embryo/fetus. In addition, by using artificial reproductive technologies (ARTs), gametes and/or embryos of livestock are exposed to unnatural conditions outside the male and female genital tract. Artificial insemination, the most widely used technique, is currently yielding pregnancy rates similar to natural mating, and calves produced by AI are equally viable after natural mating. In contrast, other ART, such as multiple ovulation and embryo transfer, have been reported to induce changes in gene expression and DNA methylation patterns with potential consequences for development.Finally, the "periconceptional" environment has been shown to not only influence the successful establishment of pregnancy but also the long-term health and productivity of the offspring. Hence, the optimization of management around the time of conception might open doors to improve animal production and product quality.

Applying data mining techniques for increasing implantation rate by selecting best sperms for intra-cytoplasmic sperm injection treatment

BACKGROUND AND OBJECTIVE

Aspiration of a good-quality sperm during intracytoplasmic sperm injection (ICSI) is one of the main concerns. Understanding the influence of individual sperm morphology on fertilization, embryo quality, and pregnancy probability is one of the most important subjects in male factor infertility. Embryologists need to decide the best sperm for injection in real time during ICSI cycle. Our objective is to predict the quality of zygote, embryo, and implantation outcome before injection of each sperm in an ICSI cycle for male factor infertility with the aim of providing a decision support system on the sperm selection.

METHODS

The information was collected from 219 patients with male factor infertility at the infertility therapy center of Alzahra hospital in Rasht from 2012 through 2014. The prepared dataset included the quality of zygote, embryo, and implantation outcome of 1544 injected sperms into the related oocytes. In our study, embryo transfer was performed at day 3. Each sperm was represented with thirteen clinical features. Data preprocessing was the first step in the proposed data mining algorithm. After applying more than 30 classifiers, 9 successful classifiers were selected and evaluated by 10-fold cross validation technique using precision, recall, F1, and AUC measures. Another important experiment was measuring the effect of each feature in prediction process.

RESULTS

In zygote and embryo quality prediction, IBK and RandomCommittee models provided 79.2% and 83.8% F1, respectively. In implantation outcome prediction, KStar model achieved 95.9% F1, which is even better than prediction of human experts. All these predictions can be done in real time.

CONCLUSIONS

A machine learning-based decision support system would be helpful in sperm selection phase of ICSI cycle to improve the success rate of ICSI treatment.

A novel behavioral model of the pasture-based dairy cow from GPS data using data mining and machine learning techniques

A better understanding of the behavior of individual grazing dairy cattle will assist in improving productivity and welfare. Global positioning systems (GPS) applied to cows could provide a means of monitoring grazing herds while overcoming the substantial efforts required for manual observation. Any model of behavioral prediction using GPS needs to be accurate and robust by accounting for inter-cow variation as well as atmospheric effects. We evaluated the performance using a series of machine learning algorithms on GPS data collected from 40 pasture-based dairy cows over 4 mo. A feature extraction step was performed on the collected raw GPS data, which resulted in 43 different attributes. The evaluated behaviors were grazing, resting, and walking. Classifier learners were built using 10 times 10-fold cross validation and tested on an independent test set. Results were evaluated using a variety of statistical significance tests across all parameters. We found that final model selection depended upon level of performance and model complexity. The classifier learner deemed most suitable for this particular problem was JRip, a rule-based learner (classification accuracy=0.85; false positive rate=0.10; F-measure=0.76; area under the receiver operating curve=0.87). This model will be used in further studies to assess the behavior and welfare of pasture-based dairy cows.