Erkennen des Beginns der klinischen Mastitis durch Auswertung der täglichen Aktivitäts- und Milchmengendaten von Milchkühen

Detection of the start of clinical mastitis by analysing daily activity and milk yield data in dairy cows

Mastitis is the disease with the highest economic impact in dairy production. As it is a multifactorial disease, different factors influence the emergence of mastitis and the clinical occurrence can be divers. Digitalization in the dairy barn, known as precision dairy farming, has the potential to serve as a syndromic surveillance system. The detection of a beginning clinical mastitis by changes in milk yield and behavior may allow early recognition and treatment of the cow at risk.
In this study, individual milk yields and activity variables of holstein frisian dairy cows housed in freestall barns in two dairy farms (farm 1 hold 1100 and farm 2 hold 400 milking cows) were investigated for deviations of this variables in the days, before clinical mastitis was diagnosed. Documented mastitis events in the herd management software Dairy Comp305 were used to identify mastitis periods. The cows were equipped with activity sensors (IceQubes of IceRobotics LtD, South Queensferry, UK).
First, the absolute values of daily milk yields and activity values (lying time, standing time, steps, motion index and changes of standing to lying) were examined ten days before the documented mastitis event. After that, the individual relative deviation five days before mastitis was calculated for all listed variables.
Due to the indirect proportional relationship of standing time and lying time and the lack of statistical significance of changes between standing and lying concerning the development of mastitis, the number of variables were reduced to the relative deviation in milk yield, lying time and motion index. On day two before the documented mastitis, the chosen variables were significant compared to the control periods.
The relative deviations of milk yield, lying time and motion index were used to distinguish between animals with mastitis and a control group. In farm 1, a combination of cut-off points led to a sensitivity of 70 % and a specificity of 40 %. In farm 2, the sensitivity of 70 % and specificity of 63.6 % was reached.
When the effect of the lactation phase on the deviation in milk yield and the individual activity to the group activity was taken into account, it became evident that this was no meaningful advantage in distinguishing mastitis periods from a control periods by relative deviations.
So far, no comparable number of mastitis events over such a long period was used for analysing this topic. It has to be noted however, that further analyses may be necessary, which could utilize more detailed information about clinical mastitis. It will also be important to check the insights of this thesis during the daily work in dairy farms.