Detecció precoç i predicció de malalties en bovins utilitzant un sistema de monitorització d'activitat
- Autores: Mohammed Anouar Belaid
- Directores de la Tesis: Sergio Calsamiglia (dir. tes.)
- Lectura: En la Universitat Autònoma de Barcelona ( España ) en 2020
- Idioma: español
- ISBN: 9788449094859
- Tribunal Calificador de la Tesis: Alfred Ferret i Quesada (presid.), Maria Devant Guille (secret.), Diego Moya Fernández (voc.)
- Programa de doctorado: Programa de Doctorado en Producción Animal por la Universidad Autónoma de Barcelona
- Materias:
- Enlaces
- Tesis en acceso abierto en: TDX
- Resumen
Three experiments were conducted to determine if an activity-monitoring systems could be used for an early predictor of the risk of sickness in cattle. In the first experiment, Friesian male calves were monitored from 30 to 90 d of life. Calf health status was controlled. A matched pair design was conducted from d -10 to +10 relative to the diseases diagnostic to compare sick versus healthy calves, and a multivariate logistic regression was performed on the days before the disease event to develop a prediction model. Healthy calves did daily 1,476 ± 195 steps, spent 185 ± 32.5 min at the feed bunk, did 10 ± 1.1 meals, 19.5 ± 1.8 lying bouts and spent 978 ± 30.5 min lying. Sick calves did fewer steps, had 18% less meals on d -1 and 0, spent less time at the feed bunk on d -10 and -1 and had 15% less lying bouts from d -2 to +9 compared with healthy calves. The prediction model developed for d -10 had a sensitivity of 67%, a specificity of 67%, and accuracy of 67%. The false discovery rates and the false omission rates were 60% and 14%, respectively. Results indicate that the occurrence of diseases can be predicted in advance and a preventive treatment can be applied only to animals at risk. In the second experiment, young bulls were monitored during the first three months after their arrival to the feedlot. Bulls were examined daily for health status. To compare sick and healthy bulls, a matched pair design was performed in the 20 d around the diseases event and a multivariate logistic regression model was built to develop a prediction model. Bulls did on average 2,422 ± 128.3 steps/d, attended the feed bunk 8 ± 0.15 times/d for a total of 95 ± 8.2 min/d, had 27.8 ± 0.76 lying bouts/d and spent 889 ± 12.5 min/d lying. Sick bulls did fewer steps, less meals, spent less time in the feed bunk, had less lying bouts and spent less lying time compared with healthy bulls. The best prediction model was able to predict sick bulls at 9 d before the clinical symptoms with a sensitivity and specificity of 79.2 and 81.3%, respectively. The validation of the model resulted in a 50% false discovery rates and 7% false omission rates. Results suggest that activity monitoring systems may be useful in the early identification of sick bulls. However, the high false positive rate may require further refinement. In the third experiment, Holstein dry cows were monitored from d –21 to the day of calving. Cows postpartum health status was monitored until 30 DIM. A multivariate linear mixed model was built from d –21 to the day of calving to compare sick vs. healthy cows. A multivariate logistic regression model was developed to predict metritis. On average, healthy cows did 1,627 ± 56 steps, spent 184 ± 10.6 min at the feed bunk, did 8.5 ± 0.3 meals, did 10 ± 0.5 lying bouts and spent 743 ± 18.4 min lying per day. Sick cows did 1,644 ± 89 steps, spent 183 ± 10 min at the feed bunk, did 8 ± 0.4 meals, did 11 ± 0.6 lying bouts and spent 740 ± 40 min lying per day. A prediction model for metritis was developed. This model at the highest sensitivity (73%) and specificity (86%), had 83.7% accuracy, 48.4% false discovery rates and 6.1% false omission rates. Results indicate that the occurrence of metritis can be predicted in advance and preventive treatment can be applied only to animals at risk.
