By Cavero, D. and Kramer, E. and Krieter, J. and Stamer, E., Livestock Science, 2009
Research Paper Web Link / URL:
http://www.sciencedirect.com/science/article/pii/S1871141309000754
http://www.sciencedirect.com/science/article/pii/S1871141309000754
Description
The aim of the present study was to develop a fuzzy logic model for classification and control of lameness and mastitis in cows using the data of the Futterkamp dairy research farm of the Schleswig–Holstein Chamber of Agriculture. A dataset of about 13,500 records from 125 cows was used. Lameness treatments were used to determine two definitions of lameness; they differed in the length of the corresponding disease block. Mastitis was determined according to the definitions: (1) udder treatments and (2). udder treatment or SCC over 400,000/ml. Disease alerts by the fuzzy logic model were generated using the variables milk yield, dry matter intake, dry matter intake behaviour (number of visits at the feeding trough, time spent at the feeding troughs), water intake, activity and information about preliminary diseases as input data. To develop and verify the model, the dataset was divided into training data (9074 records) and test data (4604 records). The evaluation of the model was carried out according to sensitivity, specificity and error rate. If the block-sensitivity was set to be at least 70%, the specificity for lameness detection ranged between 75.3% and 75.9% and the error rate varied between 98.9% and 99.5% depending on lameness definition. With the mastitis detection models, specificities ranged between 84.1% and 92.1%, while error rates were obtained between 96.2% and 97.9%. The results of the test data verified those of the training data, indicating that the models could be generalized but also are not yet applicable in practice.
The aim of the present study was to develop a fuzzy logic model for classification and control of lameness and mastitis in cows using the data of the Futterkamp dairy research farm of the Schleswig–Holstein Chamber of Agriculture. A dataset of about 13,500 records from 125 cows was used. Lameness treatments were used to determine two definitions of lameness; they differed in the length of the corresponding disease block. Mastitis was determined according to the definitions: (1) udder treatments and (2). udder treatment or SCC over 400,000/ml. Disease alerts by the fuzzy logic model were generated using the variables milk yield, dry matter intake, dry matter intake behaviour (number of visits at the feeding trough, time spent at the feeding troughs), water intake, activity and information about preliminary diseases as input data. To develop and verify the model, the dataset was divided into training data (9074 records) and test data (4604 records). The evaluation of the model was carried out according to sensitivity, specificity and error rate. If the block-sensitivity was set to be at least 70%, the specificity for lameness detection ranged between 75.3% and 75.9% and the error rate varied between 98.9% and 99.5% depending on lameness definition. With the mastitis detection models, specificities ranged between 84.1% and 92.1%, while error rates were obtained between 96.2% and 97.9%. The results of the test data verified those of the training data, indicating that the models could be generalized but also are not yet applicable in practice.
We welcome and encourage discussion of our linked research papers. Registered users can post their comments here. New users' comments are moderated, so please allow a while for them to be published.