Assessment of productive anomalies in dairy cows

In the panorama of precision livestock farming animal welfare, which the quality of products inevitably depends on, is increasingly important. Nowadays, automatic milking systems allow a more detailed monitoring of individual animals and the customized modelling of the productivity trend of each cow, as well as of the herd. It is already known that a warm, humid environment is the main cause of heat stress for dairy cows, and this is becoming even more serious due to climate change. Data from environmental sensors in the barns together with productivity and activity data enable the study and assessment of production loss due to heat stress. In this work, a new method for identifying production anomalies by modelling the lactation curve is presented. The model allows us, on the one hand, to study the residuals (difference between the observed data and the corresponding prediction) and, on the other hand, to examine the production deficit at the end of the lactation cycle. Furthermore, the use of a machine learning model applied to the data obtained from the first analysis shows it is possible to predict the milk yield loss due to heat stress. The training of the model on several animals in similar conditions (e.g. lactation, age) can be a valuable support for the farmer to predict the potential milk yield losses of the herd and to introduce the necessary preventive or mitigative measures.