The use of sensor data before parturition as an indicator of resilience of dairy cows in early lactation

The transition period is a critical phase in the life of dairy cows. Metabolic andinfectious disorders occur mostly in the first weeks after calving. These disorderscan be considered as critical transitions for which early-warning indicators mightbe available following the theory of resilience of biological systems. Sensor datamight be useful to notice early-warning signals like slower recovery fromperturbations, increased autocorrelations and increased variance. Sensor data(measuring activity and behaviour) and extensive reference data were collectedfor a group of 22 dairy cows during a period from 2 weeks prior to expectedparturition until 6 weeks after parturition. During this period the cows werescored daily for health status. The number of days of diminished health (DDH)were used a health measure of a cow. The correlations of the log-transformedDDH with several sensor quantities were determined. Correlations with averagevalues were significant (*) for inactive time and eating time. Correlations withvariances were significant (*) for ear temperature and number of steps.Correlations with autocorrelations were not significant. Correlations withnonperiodicity were significant for eating time (*), number of steps (**), motionindex (**) and lying time (***); where nonperiodicity was defined as the meansquared error of the correlogram with a sinusoid with a 24h cycle and anamplitude of 0.25. The high correlations before parturition of some sensor datawith nonperiodicity might be used as indicator for critical transitions afterparturition. Further research is needed to validate whether a regular life mayprevent disorders in dairy cows.