Technical note: Evaluation of an ear-attached movement sensor to record cow feeding behavior and activity
Bikker, J.P. ; Laar, H. van; Rump, P. ; Doorenbos, J. ; Meurs, K. van; Griffioen, G.M. ; Dijkstra, J. - \ 2014
Journal of Dairy Science 97 (2014)5. - ISSN 0022-0302 - p. 2974 - 2979.
dairy-cattle - monitoring rumination - automatic system - coefficient - agreement - ovulation - time
The ability to monitor dairy cow feeding behavior and activity could improve dairy herd management. A 3-dimensional accelerometer (SensOor; Agis Automatisering BV, Harmelen, the Netherlands) has been developed that can be attached to ear identification tags. Based on the principle that behavior can be identified by ear movements, a proprietary model classifies sensor data as “ruminating,” “eating,” “resting,” or “active.” The objective of the study was to evaluate this sensor on accuracy and precision. First, a pilot evaluation of agreement between 2 independent observers, recording behavior from 3 cows for a period of approximately 9 h each, was performed. Second, to evaluate the sensor, the behavior of 15 cows was monitored both visually (VIS) and with the sensor (SENS), with approximately 20 h of registration per cow, evenly distributed over a 24-h period, excluding milking. Cows were chosen from groups of animals in different lactation stages and parities. Each minute of SENS and VIS data was classified into 1 of 9 categories (8 behaviors and 1 transition behavior) and summarized into 4 behavioral groups, namely ruminating, eating, resting, or active, which were analyzed by calculating kappa (¿) values. For the pilot evaluation, a high level of agreement between observers was obtained, with ¿ values of =0.96 for all behavioral categories, indicating that visual observation provides a good standard. For the second trial, relationships between SENS and VIS were studied by ¿ values on a minute basis and Pearson correlation and concordance correlation coefficient analysis on behavior expressed as percentage of total registration time. Times spent ruminating, eating, resting, and active were 42.6, 15.9, 31.6, and 9.9% (SENS) respectively, and 42.1, 13.0, 30.0, and 14.9% (VIS), respectively. Overall ¿ for the comparison of SENS and VIS was substantial (0.78), with ¿ values of 0.85, 0.77, 0.86, and 0.47 for “ruminating,” “eating,” “resting,” and “active,” respectively. Pearson correlation and concordance correlation coefficients between SENS and VIS for “ruminating,” “eating,” “resting,” and “active” were 0.93, 0.88, 0.98, and 0.73, and 0.93, 0.75, 0.97, and 0.35, respectively. In conclusion, the results provide strong evidence that the present ear sensor technology can be used to monitor ruminating and resting behavior of freestall-housed dairy cattle. Our results also suggest that this technology shows promise for monitoring eating behavior, whereas more work is needed to determine its suitability to monitor activity of dairy cattle.
Lameness detection based on multivariate continuous sensing of milk yield, rumination, and neck activity
Hertem, T. van; Maltz, E. ; Antler, A. ; Romanini, C.E.B. ; Viazzi, S. ; Bahr, C. ; Schlageter-Tello, A. ; Lokhorst, C. ; Berckmans, D. ; Halachmi, I. - \ 2013
Journal of Dairy Science 96 (2013). - ISSN 0022-0302 - p. 4286 - 4298.
limb movement variables - dairy-cattle - risk-factors - monitoring rumination - clinical lameness - locomotion score - gait assessment - foot disorders - lying behavior - cows
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 4 d before diagnosis; the slope coefficient of the daily milk yield 4 d before diagnosis; the nighttime to daytime neck activity ratio 6 d before diagnosis; the milk yield week difference ratio 4 d before diagnosis; the milk yield week difference 4 d before diagnosis; the neck activity level during the daytime 7 d before diagnosis; the ruminating time during nighttime 6 d 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.