Related Experiment Video
Updated: Jun 5, 2025
Application of Long-term cultured Interferon-γ Enzyme-linked Immunospot Assay for Assessing Effector and Memory T Cell Responses in Cattle
Published on: July 11, 2015
Using supervised machine learning algorithms to predict bovine leukemia virus seropositivity in dairy cattle in
Ameer A Megahed1, Reddy Bommineni2, Michael Short3
1Department of Large Animal Clinical Sciences, College of Veterinary Medicine, University of Florida, Gainesville, FL 32610, USA; Department of Animal Medicine (Internal Medicine), Faculty of Veterinary Medicine, Benha University, Moshtohor, Toukh, Kalyobiya 13736, Egypt.
Supervised machine-learning models can predict bovine leukemia virus (BLV) infection in dairy cows. Older cows in southern Florida are more likely to test positive for BLV, according to the best-performing random forest model.
Area of Science:
- Veterinary Medicine
- Machine Learning
- Epidemiology
Background:
- Bovine leukemia virus (BLV) is an infectious disease affecting dairy cattle.
- Screening tools are needed to identify BLV-seropositive cattle efficiently.
- Supervised machine-learning (SML) algorithms offer potential for disease prediction.
Purpose of the Study:
- To compare six SML algorithms for predicting BLV seropositivity in Florida dairy cattle.
- To identify key risk factors associated with BLV seropositivity.
- To evaluate the performance of different SML models in a real-world dataset.
Main Methods:
- Utilized a dataset of 1279 dairy blood samples tested for BLV antibodies from 2012-2022.
- Compared six SML algorithms: logistic regression (LR), decision tree (DT), gradient boosting (GB), random forest (RF), neural network (NN), and support vector machine (SVM).
- Assessed model performance using metrics like Kolmogorov-Smirnov (KS) statistic, AUROC, gain, and misclassification rate.
Main Results:
- Corrected seroprevalence of BLV was 26.0%, with 312 positive samples.
- The random forest (RF) model demonstrated the highest predictive accuracy (KS=0.75, AUROC=0.93, misclassification rate=0.10).
- Logistic regression (LR) performed the worst; key predictors identified were age (≥ 5 years) and geographic location (southern Florida).
Conclusions:
- SML algorithms, particularly RF, show promise for predicting BLV seropositivity in dairy cattle.
- Dairy cattle aged 5 years and older in southern Florida exhibit a higher risk of BLV infection.
- This study highlights the methodological contribution of SML in developing predictive screening tools for BLV and the importance of representative data.
