VALIDATION OF AN ELECTRONIC SENSOR IN SHEEP BEHAVIOR

INTRODUCTION: Sheep farming has great potential for the Brazilian economy, but there are still limitations. To recognize health changes in the flock, producers must understand behaviors, health assessments, and well-being with the help of precision technology. AIMS: Thus, the objective was to validate a bovine behavioral monitoring sensor (Cowmed) for use in adult sheep with visual observation. MATERIALS AND METHODS: The experiment was conducted at the Gralha Azul Experimental Farm, Paraná, Brazil. Twenty adult sheep of mixed breeds (Texel, Ile de France, or Suffolk) were recorded for 11 days, totaling 24 hours per sheep. The sheep were individually marked with a colored stick and equipped with Cowmed sensors. The observed behaviors were: activity (AT; feeding or drinking water), rumination (RU; regurgitation of ruminal content to the mouth, followed by remastication and swallowing), and idleness (OC; idleness, walking, defecating, urinating, and interacting with other sheep). For statistical analysis, R 4.3.2 was used for Pearson correlation tests, mixed linear regression, concordance correlation coefficient (CCC), and Bland-Altman (BA) plot. RESULTS: The sensor was considered accurate for behavior if the linear regression slope was not different from one and if CCC was moderate (>0.90); and it was considered precise if the correlation coefficient was greater than 0.70 and if the mean BA bias included zero in the 95% agreement interval. The behaviors of AT, OC, and RU showed positive correlations (0.91, 0.80, and 0.78, respectively). For CCC, activity achieved an acceptable result (0.91), while idleness and rumination had weaker results (0.78 and 0.75, respectively). The BA showed agreement between the sensor and visual observation for all behaviors. The linear regression was 1.00 (confidence interval from 1.00 to 1.08) for AT, 0.88 (confidence interval 0,82 to 0,95) for OC, and 1.10 (confidence interval 1,03 to 1,17) for RU. FINAL CONSIDERATIONS: The sensor was accurate and precise for AT, precise for RU, but was not considered accurate and precise for OC. Therefore, further studies are needed to improve the accuracy of the device for sheep.

KEYWORDS: Monitoring; Precision Technology; Sheep farming.