We found a match
Your institution may have access to this item. Find your institution then sign in to continue.
- Title
Detection of susceptibility of dairy cows to clinical mastitis by artificial neural networks based on selected genotypes and milk production records.
- Authors
Zaborski, Daniel; Wojdak-Maksymiec, Katarzyna; Grzesiak, Wilhelm
- Abstract
The aim of this study was to verify the applicability of artificial neural networks to the detection of dairy cows susceptible to clinical mastitis based on veterinary records, milk recording data and selected genotypes (lactoferrin, lysozyme, tumor necrosis factor alpha and combined defensin genotypes). Moreover, we wanted to determine the effects of complete and reduced sets of predictors (input variables) on the detection performance of the neural models. A total of 24712 test-day records from 990 Polish Holstein-Friesian Black-and-White cows were analyzed. Eight continuous and eight categorical predictors (including proportion of Holstein-Friesian genes, calving age, milk yield and composition, four genotypes, lactation number and stage, calving season) were used. Health state (mastitis vs. healthy) was an output variable. Multilayer perceptrons and radial basis function networks were trained and tested, yielding the percentages of correctly detected cows susceptible to clinical mastitis and resistant ones in the range of 57.8 to 63.3 % and 60.3 to 66.6 %, respectively. The most significant factors affecting mastitis occurrence were: lactation number and stage, calving age, the season of calving and mastitis diagnosis, tumor necrosis factor alpha and combined defensin genotypes. Also, Lactroferrin genotype was quite significant for two neural models, whereas lysozyme genotype had a much smaller effect on the health status of cows. After reducing the initial set of 16 predictors to only five, decreased performance of the networks was observed. It can be concluded that an indication of cows susceptible to clinical mastitis may facilitate the application of preventive measures and consequently reduce mastitis incidence.
- Subjects
MAMMARY glands; UDDER; DIAGNOSIS; CATTLE disease prevention; INFLAMMATION; CATTLE diseases
- Publication
Landbauforschung, 2016, Vol 66, Issue 2, p145
- ISSN
2194-3605
- Publication type
Article
- DOI
10.3220/LBF1474381685000