Development of feed composition tables using a statistical screening procedure.

Millions of feed composition records generated annually by testing laboratories are valuable assets that can be used to benefit the animal nutrition community. However, it is challenging to manage, handle, and process feed composition data that originate from multiple sources, lack standardized feed names, and contain outliers. Efficient methods that consolidate and screen such data are needed to develop feed composition databases with accurate means and standard deviations (SD). Considering the interest of the animal science community in data management and the importance of feed composition tables for the animal industry, the objective was to develop a set of procedures to construct accurate feed composition tables from large data sets. A published statistical procedure, designed to screen feed composition data, was employed, modified, and programmed to operate using Python and SAS. The 2.76 million data received from 4 commercial feed testing laboratories were used to develop procedures and to construct tables summarizing feed composition. Briefly, feed names and nutrients across laboratories were standardized, and erroneous and duplicated records were removed. Histogram, univariate, and principal component analyses were used to identify and remove outliers having key nutrients outside of the mean ± 3.5 SD. Clustering procedures identified subgroups of feeds within a large data set. Aside from the clustering step that was programmed in Python to automatically execute in SAS, all steps were programmed and automatically conducted using Python followed by a manual evaluation of the resulting mean Pearson correlation matrices of clusters. The input data set contained 42, 94, 162, and 270 feeds from 4 laboratories and comprised 25 to 30 nutrients. The final database included 174 feeds and 1.48 million records. The developed procedures effectively classified by-products (e.g., distillers grains and solubles as low or high fat), forages (e.g., legume or grass-legume mixture by maturity), and oilseeds versus meal (e.g., soybeans as whole raw seeds vs. soybean meal expellers or solvent extracted) into distinct sub-populations. Results from these analyses suggest that the procedure can provide a robust tool to construct and update large feed data sets. This approach can also be used by commercial laboratories, feed manufacturers, animal producers, and other professionals to process feed composition data sets and update feed libraries.

Lameness detection in dairy cattle: single predictor v. multivariate analysis of image-based posture processing and behaviour and performance sensing.

The objective of this study was to evaluate if a multi-sensor system (milk, activity, body posture) was a better classifier for lameness than the single-sensor-based detection models. Between September 2013 and August 2014, 3629 cow observations were collected on a commercial dairy farm in Belgium. Human locomotion scoring was used as reference for the model development and evaluation. Cow behaviour and performance was measured with existing sensors that were already present at the farm. A prototype of three-dimensional-based video recording system was used to quantify automatically the back posture of a cow. For the single predictor comparisons, a receiver operating characteristics curve was made. For the multivariate detection models, logistic regression and generalized linear mixed models (GLMM) were developed. The best lameness classification model was obtained by the multi-sensor analysis (area under the receiver operating characteristics curve (AUC)=0.757±0.029), containing a combination of milk and milking variables, activity and gait and posture variables from videos. Second, the multivariate video-based system (AUC=0.732±0.011) performed better than the multivariate milk sensors (AUC=0.604±0.026) and the multivariate behaviour sensors (AUC=0.633±0.018). The video-based system performed better than the combined behaviour and performance-based detection model (AUC=0.669±0.028), indicating that it is worthwhile to consider a video-based lameness detection system, regardless the presence of other existing sensors in the farm. The results suggest that Θ2, the feature variable for the back curvature around the hip joints, with an AUC of 0.719 is the best single predictor variable for lameness detection based on locomotion scoring. In general, this study showed that the video-based back posture monitoring system is outperforming the behaviour and performance sensing techniques for locomotion scoring-based lameness detection. A GLMM with seven specific variables (walking speed, back posture measurement, daytime activity, milk yield, lactation stage, milk peak flow rate and milk peak conductivity) is the best combination of variables for lameness classification. The accuracy on four-level lameness classification was 60.3%. The accuracy improved to 79.8% for binary lameness classification. The binary GLMM obtained a sensitivity of 68.5% and a specificity of 87.6%, which both exceed the sensitivity (52.1%±4.7%) and specificity (83.2%±2.3%) of the multi-sensor logistic regression model. This shows that the repeated measures analysis in the GLMM, taking into account the individual history of the animal, outperforms the classification when thresholds based on herd level (a statistical population) are used.

The effect of routine hoof trimming on locomotion score, ruminating time, activity, and milk yield of dairy cows.

The objective of this study was to quantify the effect of hoof trimming on cow behavior (ruminating time, activity, and locomotion score) and performance (milk yield) over time. Data were gathered from a commercial dairy farm in Israel where routine hoof trimming is done by a trained hoof trimmer twice per year on the entire herd. In total, 288 cows spread over 6 groups with varying production levels were used for the analysis. Cow behavior was measured continuously with a commercial neck activity logger and a ruminating time logger (HR-Tag, SCR Engineers Ltd., Netanya, Israel). Milk yield was recorded during each milking session with a commercial milk flow sensor (Free Flow, SCR Engineers Ltd.). A trained observer assigned on the spot 5-point locomotion scores during 19 nighttime milking occasions between 22 October 2012 and 4 February 2013. Behavioral and performance data were gathered from 1wk before hoof trimming until 1wk after hoof trimming. A generalized linear mixed model was used to statistically test all main and interactive effects of hoof trimming, parity, lactation stage, and hoof lesion presence on ruminating time, neck activity, milk yield, and locomotion score. The results on locomotion scores show that the proportional distribution of cows in the different locomotion score classes changes significantly after trimming. The proportion of cows with a locomotion score ≥3 increases from 14% before to 34% directly after the hoof trimming. Two months after the trimming, the number of cows with a locomotion score ≥3 reduced to 20%, which was still higher than the baseline values 2wk before the trimming. The neck activity level was significantly reduced 1d after trimming (380±6 bits/d) compared with before trimming (389±6 bits/d). Each one-unit increase in locomotion score reduced cow activity level by 4.488 bits/d. The effect of hoof trimming on ruminating time was affected by an interaction effect with parity. The effect of hoof trimming on locomotion scores was affected by an interaction effect with lactation stage and tended to be affected by interaction effects with hoof lesion presence, indicating that cows with a lesion reacted different to the trimming than cows without a lesion did. The results show that the routine hoof trimming affected dairy cow behavior and performance in this farm.

Lameness detection based on multivariate continuous sensing of milk yield, rumination, and neck activity.

The objective of this study was to develop and validate a mathematical model to detect clinical lameness based on existing sensor data that relate to the behavior and performance of cows in a commercial dairy farm. Identification of lame (44) and not lame (74) cows in the database was done based on the farm's daily herd health reports. All cows were equipped with a behavior sensor that measured neck activity and ruminating time. The cow's performance was measured with a milk yield meter in the milking parlor. In total, 38 model input variables were constructed from the sensor data comprising absolute values, relative values, daily standard deviations, slope coefficients, daytime and nighttime periods, variables related to individual temperament, and milk session-related variables. A lame group, cows recognized and treated for lameness, to not lame group comparison of daily data was done. Correlations between the dichotomous output variable (lame or not lame) and the model input variables were made. The highest correlation coefficient was obtained for the milk yield variable (rMY=0.45). In addition, a logistic regression model was developed based on the 7 highest correlated model input variables (the daily milk yield 4d before diagnosis; the slope coefficient of the daily milk yield 4d before diagnosis; the nighttime to daytime neck activity ratio 6d before diagnosis; the milk yield week difference ratio 4d before diagnosis; the milk yield week difference 4d before diagnosis; the neck activity level during the daytime 7d before diagnosis; the ruminating time during nighttime 6d before diagnosis). After a 10-fold cross-validation, the model obtained a sensitivity of 0.89 and a specificity of 0.85, with a correct classification rate of 0.86 when based on the averaged 10-fold model coefficients. This study demonstrates that existing farm data initially used for other purposes, such as heat detection, can be exploited for the automated detection of clinically lame animals on a daily basis as well.

Analysis of individual classification of lameness using automatic measurement of back posture in dairy cattle.

Currently, diagnosis of lameness at an early stage in dairy cows relies on visual observation by the farmer, which is time consuming and often omitted. Many studies have tried to develop automatic cow lameness detection systems. However, those studies apply thresholds to the whole population to detect whether or not an individual cow is lame. Therefore, the objective of this study was to develop and test an individualized version of the body movement pattern score, which uses back posture to classify lameness into 3 classes, and to compare both the population and the individual approach under farm conditions. In a data set of 223 videos from 90 cows, 76% of cows were correctly classified, with an 83% true positive rate and 22% false positive rate when using the population approach. A new data set, containing 105 videos of 8 cows that had moved through all 3 lameness classes, was used for an ANOVA on the 3 different classes, showing that body movement pattern scores differed significantly among cows. Moreover, the classification accuracy and the true positive rate increased by 10 percentage units up to 91%, and the false positive rate decreased by 4 percentage units down to 6% when based on an individual threshold compared with a population threshold.