PSIII-7 Using automated monitoring devices to predict pregnancy by artificial insemination in grazing beef cows.

Abstract We recently validated an automated monitoring device (i.e., accelerometers) for estrous detection in grazing beef cows with 91% of accuracy, but whether this methodology predicts pregnancy by artificial insemination (AI) still unclear. We hypothesized that the variables (rumination and activity) measured by the accelerometer software are associated positively with the probability of pregnancy/AI in grazing beef cows. Beef cows (n = 100) were fitted with an ear-mounted accelerometer tag (Sense Hub Cow Calf; Merck Animal Health, USA) in the middle third of the right ear on the first day of a 14-day estrous synchronization protocol. On the last day of the synchronization, cows were fitted with an estrous detection patch (ESTROTECTTM, Rockway Inc, USA). A split-time AI was performed 66 or 90 hours later based on estrus by the ESTROTECT, and the diameter of the largest follicle was measured. During estrus the following variables from the accelerometer software were collected: peak heat value (highest heat value), heat duration (hours between estrus onset and end), heat index (index value from the software), proestrus duration (hours from CIDR removal to estrus onset), daily eating (average eating value), raw activity (highest activity value), and peak activity (value of activity at the moment peak heat). Pregnancy diagnosis was performed 30 days post-AI. The relationship between the accelerometer variables and pregnancy/AI was analyzed using mixed-effects models (PROC MIXED; SAS). 92% of cows displayed estrus. Peak heat value (94.8 ± 1.32 vs 99.4 ± 1.17; P = 0.011), heat duration (20.31 ± 0.56 vs 21.9 ± 0.62 hours; P = 0.052), heat index (90.83 ± 1.37 vs. 97.55 ± 1.18; P = 0.0004), and proestrus duration (66.70 ± 3.53 vs. 73.67 ± 3.47 hours P= 0.001) were reduced in pregnant than to non-pregnant cows. Daily eating tended to be greater in pregnant cows (280.7 ± 32.36 vs. 236.44 ± 31.53; P=0.077). No significant association between raw activity or peak activity and pregnancy/AI was observed (P >0.1). For each unit of increase in the follicular diameter, heat index increases 0.92 (P=0.049). There was a quadratic association between follicular diameter and the percentage of increase in activity (P=0.041). ROC curve analysis testing the ability of the accelerometer variables to predict pregnancy/AI harvested accuracies from 0.6055 to 0.7211. In conclusion, accelerometers show potential for pregnancy prediction in grazing beef cows, but more calibration might be necessary to increase the accuracy. The reduction in activity-related variables in pregnant cows suggests that when displaying estrus, the control of its intensity is important for the pregnancy. Perhaps cows that display estrus that are not of very high intensity save more energy which helps with the establishment of pregnancy.