Analysis of dairy cows’ activity using a hybrid modelling approach for the early detection of health problems

Masoomeh Taghipoor (1) , Séverine Bord (2) , Quentin Bulk (3) , Joon Kwon (2)

1 MoSAR - Modélisation Systémique Appliquée aux Ruminants
2 MIA Paris-Saclay - Mathématiques et Informatique Appliquées
3 Ferme expérimentale d'AgroParisTech

Early detection of disease in animals allows farmers to refine the use of medicines and enables early intervention, reducing the potential pain an animal will experience. Animal activity is one indicator of the state of wellbeing of the animal, and one of the first indicators of the degradation of health and welfare in animals. The objective of this work is to study the potential of individual times-series activity data to forecast degradation in health state of dairy cows. Conclusion and perspectives ➢ The moderate value of AUC is related to the small number of mastitis cases in the dataset. ➢ The larger is the size of training sample the more robust is the predictor. ➢ Using deviations of milk yield as objective measures of environmental disturbances.

Domaines: Systèmes dynamiques [math.DS]

Masoomeh Taghipoor, Séverine Bord, Quentin Bulk, Joon Kwon. Analysis of dairy cows’ activity using a hybrid modelling approach for the early detection of health problems. UFAW 2022 - Advancing Animal Welfare Science, Jun 2022, Edinburgh, United Kingdom. ⟨hal-03776962⟩