אסיף מאגר המחקר החקלאי

Ketosis detection in early lactation of dairy cows by behaviour and performance sensing

Year:

2013

Source of publication :

Authors :

Volume :

Co-Authors:

Steensels, M., Institute of Agricultural Engineering, Agricultural Research Organization (ARO), Volcani Center, PO Box 6, Bet-Dagan 50250, Israel, Division Measure, Model and Manage Bioresponses (M3-BIORES), KU Leuven, Kasteelpark Arenberg 30, 3001 Heverlee, Belgium
Maltz, E., Institute of Agricultural Engineering, Agricultural Research Organization (ARO), Volcani Center, PO Box 6, Bet-Dagan 50250, Israel
Bahr, C., Division Measure, Model and Manage Bioresponses (M3-BIORES), KU Leuven, Kasteelpark Arenberg 30, 3001 Heverlee, Belgium
Berckmans, D., Division Measure, Model and Manage Bioresponses (M3-BIORES), KU Leuven, Kasteelpark Arenberg 30, 3001 Heverlee, Belgium
Antler, A., Institute of Agricultural Engineering, Agricultural Research Organization (ARO), Volcani Center, PO Box 6, Bet-Dagan 50250, Israel
Halachmi, I., Institute of Agricultural Engineering, Agricultural Research Organization (ARO), Volcani Center, PO Box 6, Bet-Dagan 50250, Israel

Facilitators :

Link :

Abstract:

The aim was to develop and validate an automated detection model for post-calving ketosis based on rumination time, activity and milk yield. Data were collected in four commercial dairy farms. In total, 203 cows were diagnosed with ketosis and 503 cows were healthy. Rumination time, activity and milk yield were measured online by commercial sensors. A logistic regression model was (i) calibrated on the large farm and validated on other farms, (ii) calibrated on a percentage of all farm data and validated on all remaining farm data and (iii) calibrated and validated on individual farm level. When calibrated on the large farm, validation specificities ranged from 0.54 to 0.85 and sensitivities ranged from 0.45 to 0.82 in the different farms. When calibrated on all farm data, validation specificities ranged from 0.53 to 0.85 and sensitivities ranged from 0.55 to 0.96 in the different farms. The best model performance was obtained when calibration and validation dataset came from the same farm, with model specificities ranging from 0.74 to 0.84 and sensitivities ranging from 0.68 to 0.82 in the different farms. A general model gave reasonable results in all farms. A better model performance was, however, obtained with a farm-specific calibration.

Note:

Related Files :

Related Content

More details

You may also be interested in