Technical note: Accuracy of an ear tag-attached accelerometer to monitor rumination and feeding behavior in feedlot cattle.

Early identification of sick cattle increases treatment success and decreases mortality. Continuous automated records of behavior can be used to identify sick cattle early in the disease process. The objective was to evaluate accuracy of an ear-attached accelerometer (SensOor) that quantified ear movements and estimated feeding and rumination time through a proprietary algorithm. Accelerometers were attached to the ear tag of 18 steers with an initial mean BW of 326 ± 46 kg. The manufacturer's proprietary software was used to determine time spent "feeding," "ruminating," "active," and "resting." Direct visual observation was used to validate the accelerometer. Sensitivity, specificity, and predictive values were calculated for rumination and feeding separately. Repeated measures were accounted for using mixed model logistic regression. Single minutes of either feeding or rumination in a run of other behavior minutes were changed to the preceding behavior. Accuracy and precision of hourly recorded feeding and rumination times were assessed using the concordance correlation coefficient adjusted for repeated measurements. Sensitivity and specificity were 95 and 76% for feeding and 49 and 96% for rumination, respectively. Concordance correlation between observations and the sensor were 0.79 (95% CI: 0.61 to 0.85) and 0.44 (95% CI: 0.23 to 0.60) for feeding and rumination, respectively. There was large variability among steers, with concordance correlations ranging from 0.09 to 0.98 for rumination time and from 0.58 to 0.96 for feeding time. We conclude that the accelerometer is a promising monitoring system for feeding behavior.