Publication Database

SCC predictions using generalized additive models:
can they support mastitis management decisions? (2019)

Art

Vortrag

Autoren

Bartel, A. (WE 16)

Gass, E.

Onken, F.

Baumgartner, C.

Querengässer, F. (WE 16)

Doherr, M. G. (WE 16)

Kongress

IDF Mastitis Conference 2019

Copenhagen, Denmark, 15. – 16.05.2019

Quelle

IDF Mastitis Conference 2019 : 14-16 May, Copenhagen, Denmark : Abstract Book — International Dairy Federation (IDF) ; SEGES ; Danish Udder Health Center (DUHC) (Hrsg.)

Copenhagen, Denmark, 2019 — S. 24

Sprache

Englisch

Verweise

URL (Volltext): https://www.researchgate.net/publication/333506813_The_IDF_Mastitis_Conference_2019_in_Copenhagen/link/5cf0c983a6fdcc8475f8c634/download

DOI: 10.13140/RG.2.2.21103.28324

Kontakt

Institut für Veterinär-Epidemiologie und Biometrie

Königsweg 67
14163 Berlin
+49 30 838 56034
epi@vetmed.fu-berlin.de

Abstract / Zusammenfassung

Mastitis is one of the most common diseases in dairy cows. Somatic cell counts (SCC) obtained from monthly DHI testing, are helpful to monitor udder health at individual animal and herd level. In Germany, farm-level parameters using DHI based SCC results were established. In this study, we complemented these key figures by establishing statistical models that reliably predict udder health in the future. Individual cow risk predictions for sustained cell count increases can support treatment or culling decisions. Identification of animals that have a low probability of elevated cell counts in the future allows deriving comparable udder health parameters at cow and farm level.
We combined DHI data from over 900,000 cows in 3 German federal states over the course of 2 years, resulting in more than 10 million measurements. We fitted 2 generalized additive models (GAM), one for chronic SCC elevations and one for stable good udder health. Chronic elevations were defined as cell counts above a defined threshold in the next 2 DHI measurements. In order to meet different farmers’ needs, results were derived for SCC thresholds between 200,000 - 700,000 cells/ml. Cows with stable good udder health had SCC values below 100,000 cells/ml in the next 2 DHI measurements. We tested both models using internal and external validation data. By using GAMs, we can identify biases in the underlying data.
The predictions of the models accurately reflect the real probability, independent of region, size and breed composition of a farm. The AUC of the chronic and good udder health models were 0.868 and 0.779 with a Calibration Slope of 0.989 and 0.992. Using predictions for the risk of chronic cell count elevations, farmers could identify animals in need of intervention and prioritize resources by ranking the animals. The proportion of animals with a high probability of stable low cell counts can serve as a general indicator of udder health and for effective management.