Automated detection of health disorders in lactating dairy cattle on pasture: a preliminary study.

Since previous health monitoring systems have shown themselves to be unsuccessful in predicting health disorders in dairy cows managed on pasture, the aim of this study was to evaluate the performance of automated health monitoring integrated in an accelerometer-based oestrus detection system (ODS) for dairy cows on pasture. Mixed-breed lactating dairy cows (n=109) in a seasonal-calving herd managed at pasture were fitted with an ODS that provided automated health monitoring. The ODS performed multimetric analysis of behavioural patterns to generate health alerts. Data were collected during the artificial insemination period of 66 days. Clinical examinations and farmer's observations were used to evaluate the performance of automated health monitoring. During the insemination period, the farmer generated two health alerts, which were classified false positives (2/2; 100%). The ODS generated 31 automated health alerts. Of all automated health alerts, 3/31 (9.7%) were confirmed as true health disorders and 28/31 (90.3%) alerts were classified as false positives. The positive predictive value (PPV) of automated health monitoring was 9.7 (95% CI=2-25.8) %. The ODS was able to alert lactating dairy cows on pasture suffering from health disorders. True health disorders were alerted by the ODS before the farmer noticed them, which could provide early and successful treatment when using the system on-farm for automated health monitoring. The evaluated accuracy of automated health monitoring is opposed to a targeted use of the system for on-farm health monitoring. For further validation, testing on other farms and during the transition period would be of interest.