Department of Farm Animal Health, Faculty of Veterinary Medicine, University of Utrecht, Utrecht, the Netherlands.
J Dairy Sci. 2010 Jan;93(1):234-41. doi: 10.3168/jds.2009-2580.
An accurate prediction of the average somatic cell count (SCC) for the next month would be a valuable tool to support udder health management decisions. A linear mixed effect (LME) model was used to predict the average herd SCC (HSCC) for the following month. The LME model included data on SCC, herd characteristics, season, and management practices determined in a previous study that quantified the contribution of each factor for the HSCC. The LME model was tested on a new data set of 101 farms and included data from 3 consecutive years. The farms were split randomly in 2 groups of 50 and 51 farms. The first group of 50 farms was used to check for systematic errors in predicting monthly HSCC. An initial model was based on older data from a different part of the Netherlands and systematically overestimated HSCC in most months. Therefore, the model was adjusted for the difference in average HSCC between the 2 sets of farms (from the previous and current study) using the data from the first group of 50 farms. Subsequently, the data from the second group of 51 farms were used to independently assess this final model. A null model (no explanatory variables included) predicted 48 and 59% of the HSCC within the predetermined range of 20,000 and 30,000 cells/mL, respectively. The final LME model predicted 72 and 81% of the HSCC of the next month correctly within these 2 ranges. These outcomes indicate that the final LME model was a valid additional tool for farmers that could be useful in their short-term decisions regarding udder health management and could be included in dairy herd health programs.
准确预测下一个月的平均体细胞计数（SCC）将是支持乳房健康管理决策的有价值的工具。使用线性混合效应（LME）模型预测下一个月的平均 herd SCC（HSCC）。LME 模型包括 SCC、 herd 特征、季节和管理实践的数据，这些数据来自先前的一项研究，该研究量化了每个因素对 HSCC 的贡献。LME 模型在 101 个农场的新数据集上进行了测试，该数据集包含了 3 年的连续数据。农场被随机分为两组，每组 50 个和 51 个农场。第一组 50 个农场用于检查预测每月 HSCC 的系统误差。最初的模型基于荷兰不同地区的旧数据，在大多数月份系统地高估了 HSCC。因此，使用第一组 50 个农场的数据，根据两组农场（来自之前和当前的研究）之间平均 HSCC 的差异对模型进行了调整。随后，使用第二组 51 个农场的数据来独立评估这个最终模型。一个空模型（不包括解释变量）预测了 48%和 59%的 HSCC 在 20,000 到 30,000 个细胞/ml 的预定范围内，分别。最终的 LME 模型在这两个范围内正确预测了下一个月 72%和 81%的 HSCC。这些结果表明，最终的 LME 模型是农民的一个有效的附加工具，可以在他们关于乳房健康管理的短期决策中使用，并可以包含在奶牛群健康计划中。
