Predicting impending calving using automatically collected measures of activity and rumination in dairy cattle

Abstract Text: The objective of this study was to monitor behavioral changes in prepartum dairy cattle and predict impending calvings through the automated observation of activity and rumination. Data collection for 29 primiparous and 46 multiparous Holstein dairy cattle occurred from September 13, 2011 through May 16, 2013 at the University of Kentucky Coldstream Dairy. The HR Tag (SCR Engineers, Ltd., Israel) was used to automatically collect neck activity and rumination data in 2-hour increments. The IceQube (IceRobotics, Ltd., Scotland) collected hourly step number, hours lying, hours standing, lying bouts, and total motion data. Data collection occurred for 7 weeks prepartum and retrospective data analysis was performed using SAS (Cary, NC). Data summed by day for each cow was included in the calculation of a 7-day backward moving average and standard deviation to establish each cow’s baseline values. Least-squares means were calculated from moving averages using the MIXED procedure of SAS. Parameters exhibiting significant differences (P < 0.05; mean ± SE) on the day of calving (DAY⁰) versus the day before (DAY^-1) included: hours lying (DAY⁰: 10.15 ± 0.25 vs. DAY^-1: 10.50 ± 0.25), hours standing (DAY⁰: 13.79 ± 0.25 vs. DAY^-1: 13.47 ± 0.25), lying bouts (DAY⁰: 11.36 ± 0.38 vs. DAY^-1: 9.86 ± 0.38), and minutes ruminating (DAY⁰: 319.29 ± 12.01 vs. DAY^-1: 336.86 ± 12.01). Neck activity, step number and total motion showed no significant differences between DAY⁰ and DAY^-1. Z-scores were calculated using data summed by day for each cow, moving averages, and moving standard deviations. Deviations ≥ 1.5 from the mean dictated alert creation. The percent of cows (n = 61) showing an alert by parameter on DAY⁰ was: lying bouts (75.4%), minutes ruminating (45.9%), step number (39.3%), total motion (39.3%), hours standing (32.8%), hours lying (31.1%), and neck activity (21.3%). In comparison, the percent of cows (n = 57) showing an alert by parameter on DAY^-1was: lying bouts (15.8%), minutes ruminating (15.8%), step number (17.5%), total motion (15.8%), hours standing (28.1%), hours lying (28.1%), and neck activity (12.3%). Changes in least-squares means and alerts relative to calving indicate that these measures may be useful in predicting impending calvings without adding new technologies or parameters, but further research is necessary. 

Keywords: calving prediction, days before calving, activity and rumination