Use of animal-mounted accelerometers to identify positive welfare in dairy cattle.

Background: Methods for assessing dairy cattle welfare, such as the gold standard Qualitative Behaviour Assessment (QBA), require training and are time consuming in a sector facing labour constraints and staffing shortages. This research explores whether automated measurements from existing animal-mounted sensors can be used as a basis for objective welfare assessment in dairy cattle, supporting consumer, processor and farmer needs by demonstrating positive welfare on farm. Data (mean standing and lying duration, mean standing and lying bout frequency, mean standing and lying bout duration, mean maximum standing and lying bout duration, mean lying and standing bout frequency, mean step count) were acquired from commercially available ankle-mounted accelerometers intended for oestrus detection and compared to 20 QBA metrics from 107 animals at pasture and during housing on four dairy farms. Results: Our analysis showed that sensor data was significantly correlated with QBA data, suggesting that sensor data can be used to accurately characterise animal behaviour. Although sensor data was influenced by location (housed and pasture), data obtained from ankle-mounted pedometers showed high accuracy (61%) in predicting an animal's mood as positive or negative. In addition, step count and standing time were strongly correlated with positive behaviour QBA scores, suggesting that increased step count with decreased standing time may be an indicator of positive welfare. QBA results showed that animals at pasture displayed more positive behaviours, with 70.2% of pasture cattle exhibiting QBA scores associated with positive behaviours/mood, in comparison to 34.0% of housed cattle. Results also showed that skewedness of sensor data from cattle at pasture was an accurate indicator of behavioural synchrony (i.e. animals exhibiting the same behaviour at the same time), a known measure of positive welfare, although more granular time data is needed to investigate further. Conclusions: This study demonstrates the potential of using automated animal-mounted sensor data to assess positive welfare in dairy cattle, with sensor-derived behavioural features enabling classification of mood states (positive or negative) in 61% of observations. Supplementary Information: The online version contains supplementary material available at 10.1186/s44363-025-00018-6.