| Autor: |
Pichlbauer B; Center for Veterinary Systems Transformation and Sustainability, Clinical Department for Farm Animals and Food System Science, University of Veterinary Medicine, 1210 Vienna, Austria., Chapa Gonzalez JM; Center for Veterinary Systems Transformation and Sustainability, Clinical Department for Farm Animals and Food System Science, University of Veterinary Medicine, 1210 Vienna, Austria., Bobal M; Center for Veterinary Systems Transformation and Sustainability, Clinical Department for Farm Animals and Food System Science, University of Veterinary Medicine, 1210 Vienna, Austria., Guse C; Center for Veterinary Systems Transformation and Sustainability, Clinical Department for Farm Animals and Food System Science, University of Veterinary Medicine, 1210 Vienna, Austria., Iwersen M; Center for Veterinary Systems Transformation and Sustainability, Clinical Department for Farm Animals and Food System Science, University of Veterinary Medicine, 1210 Vienna, Austria., Drillich M; Center for Veterinary Systems Transformation and Sustainability, Clinical Department for Farm Animals and Food System Science, University of Veterinary Medicine, 1210 Vienna, Austria.; Unit for Reproduction Medicine and Udder Health, Faculty of Veterinary Medicine, Freie Universität Berlin, 14163 Berlin, Germany. |
| Abstrakt: |
Monitoring animal behavior using sensor technologies requires prior testing under varying conditions because behaviors can differ significantly, such as between grazing and confined cows. This study aimed to validate several sensor systems for classifying rumination and lying behaviors in cows on pasture under different environmental conditions, compare the sensors' performance at different time resolutions, and evaluate a correction algorithm for rumination data. Ten Simmental dairy cows were monitored on pasture, each simultaneously equipped with an ear-tag accelerometer (ET), two different leg-mounted accelerometers (LMs), and a noseband sensor (NB). Indirect visual observations using drone-recorded video footage served as the gold standard for validation. The concordance correlation coefficient (CCC) for rumination time was very high for both the ET and NB (0.91-0.96) at a 10 min time resolution. Applying the correction algorithm to 1 min data improved the CCC for the NB from 0.68 to 0.89. For lying time, the CCC was moderate for the ET (0.55) but nearly perfect for both LMs (0.99). In conclusion, both sensors evaluated for classifying rumination are suitable for cows on pasture. We recommend using a correction algorithm for 1 min NB data. For the measurement of lying time, the LMs significantly outperformed the ET. |
