The objectives of this study were to evaluate the performance of the SCR eSense ear tag automated activity monitor (AAM) to detect estrus behavior in Holstein heifers and to determine the optimal time from estrus alert to artificial insemination (AI) using sex-sorted or conventional semen. In total, 281 heifers were fitted with the AAM once eligible for breeding (>13.5 m of age). For the first AI, estrus was synchronized using 500 μg of cloprostenol (PGF), given 14 d apart, and heifers were given estrus detection patches (Estrotect™) after the second PGF. Heifers were inseminated at randomly attributed times after high activity alert from the AAM system or if the estrus patch had ≥ 50% colour change. Most heifers received sex-sorted semen for the first AI and conventional semen for subsequent inseminations. Pregnancy diagnosis was performed at 30 d post AI and heifers had four opportunities to become pregnant. In a subset of heifers (n = 149), ovaries were scanned every 12 h from the time of AI until ovulation (OV). The system recorded a heat index (measure of estrus strength), maximum activity change, maximum rumination change and duration of high activity. The sensitivity was 91.0%, with a false positive and false negative rate of 8.0%, and the positive predictive value to detect true estrus events was 83.5%. Pregnancy per AI to first AI was 67.6% and 97.9% of heifers become pregnant after four inseminations. Most false positive estrus events had a heat index < 45 and a rumination change < -20, while false negative events had a rumination change ≥ -20. Odds of pregnancy was not associated with any estrus characteristics measured by the system. However, pre-ovulatory follicle diameter had a weak correlation (r < 0.25) with all estrus characteristics. The average (range) interval of onset of high activity, peak activity and end of high activity to OV was 28 h (16-46 h), 22 h (10-40 h) and 16 h (0-36 h), respectively. For conventional semen, each hour increase in interval from activity onset or peak activity to AI reduced the predicted probability of pregnancy by 3.8 and 4.2%, respectively. For sex-sorted semen, the relationship between activity onset or peak activity to AI and predicted probability of pregnancy was quadratic, but not significant. Overall, the SCR eSense ear tag AAM performed well and strategies to identify false positive and false negative estrus events, along with optimization of timing of AI, should further improve performance in Holstein heifers.

中文翻译：

---

用于奶牛发情预测的耳标自动活动监测器的性能和优化

本研究的目的是评估 SCR eSense 耳标自动活动监测器 (AAM) 的性能，以检测荷斯坦小母牛的发情行为，并使用性别分类或常规方法确定从发情警报到人工授精 (AI) 的最佳时间精液。总共有 281 头小母牛安装了 AAM，一旦有资格进行育种（>13.5 m 年龄）。对于第一个 AI，使用 500 μg 氯前列醇 (PGF) 同步发情，间隔 14 天，并在第二个 PGF 后给小母牛发情检测贴片 (Estrotect™)。在 AAM 系统发出高活动警报后或如果发情补丁有 ≥ 50% 的颜色变化，小母牛在随机归因的时间受精。大多数小母牛在第一次人工授精时接受按性别分类的精液，在随后的授精时接受常规精液。在 AI 后 30 天进行妊娠诊断，小母牛有 4 次怀孕机会。在小母牛的一个子集 (n = 149) 中，从 AI 到排卵 (OV) 每 12 小时扫描一次卵巢。该系统记录了发情指数（发情强度的测量）、最大活动变化、最大反刍变化和高活动持续时间。灵敏度为91.0%，假阳性和假阴性率为8.0%，检测真发情事件的阳性预测值为83.5%。每次人工授精到第一次人工授精的妊娠率为 67.6%，97.9% 的小母牛在四次授精后怀孕。大多数假阳性发情事件的发情指数<45，反刍变化<-20，而假阴性事件的反刍变化≥-20。怀孕几率与系统测量的任何发情特征无关。然而，排卵前卵泡直径与所有发情特征的相关性较弱（r < 0.25）。高活性开始、峰值活性和高活性结束到 OV 的平均（范围）间隔分别为 28 小时（16-46 小时）、22 小时（10-40 小时）和 16 小时（0-36 小时） . 对于传统精液，从活动开始或活动高峰到 AI 的间隔每增加一小时，预测的怀孕概率分别降低 3.8% 和 4.2%。对于按性别分类的精液，活动开始或活动峰值与 AI 与预测怀孕概率之间的关系是二次的，但不显着。总体而言，SCR eSense 耳标 AAM 表现良好，并具有识别假阳性和假阴性发情事件的策略，