Antibiogram use on dairy cattle for bovine respiratory disease: Factors associated with bacterial pathogen identification and prediction of bacterial recovery using machine learning
Journal of Dairy Science · 2025
DOI: 10.3168/jds.2024-26205 ↗Kurz gefasst
Fachartikel in der Zeitschrift Journal of Dairy Science, 2025. Verfasst von Hugo F Monteiro und 9 weiteren Autor:innen. Beteiligte Einrichtung: Department of Population Health and Reproduction, School of Veterinary Medicine, University of California, Davis, CA 95616. Electronic address: hmont…. Hauptthemen laut MEDLINE: Cattle Diseases, Bovine Respiratory Disease Complex. Die Arbeit wurde bisher 3-mal zitiert (Crossref) und nennt 34 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0). Gefördert von: University of California, Davis, Council of Fashion Designers of America, California Department of Food and Agriculture.
Abstract (Originalsprache)
Effective isolation of bacterial pathogens for bovine respiratory disease (BRD) is a critical step for accurate diagnosis of the agent associated with this disease on the dairy. Limited information is available on factors associated with herd-level bacterial pathogen recovery for BRD clinical cases, which are important data to help identify strategies to support the successful collection of a minimum number of each organism over time to generate cumulative antibiogram susceptibility testing reports. Our objective was to evaluate factors associated with the recovery of common pathobionts (Pasteurella multocida and Mannheimia haemolytica) in BRD clinical cases from preweaning calves, heifers, and cows at 3 California dairy farms over 2 yr. A second objective was to test the predictability of isolating these organisms in BRD clinical cases using the factors evaluated in the first objective utilizing machine learning (ML). During monthly herd visits, deep nasopharyngeal samples were collected from calves, heifers, and cows diagnosed with BRD over 2 yr. Samples were cultured in aerobic conditions, and pathogens were isolated through colony morphology and validated with MALDI-TOF MS. Evaluation of factors influencing bacterial recovery was initially tested for independence, followed by a logistic regression model and a stepwise logistic feature selection in SAS, and ML classifiers with leave-one-out cross-validation in Python packages. For our study, samples were collected from a total of 301 BRD clinical cases: 146 samples with a culture-positive for P. multocida, 63 samples with a culture-positive for M. haemolytica, and 3 samples with a culture-positive for Histophilus somni. For factors associated with the culture-positive of P. multocida in BRD clinical cases, an interaction between age and season was identified, where cows had overall lower odds of being culture-positive independently of the season when compared with calves in the spring and summer and heifers in the fall and winter. For factors associated with the culture-positive of M. haemolytica in BRD clinical cases, an interaction was also observed between age and season, but the farm further played a role in the odds of being culture-positive, with one farm having considerably greater odds than the remaining ones. Machine learning models with cross-validation showed a weak ability to distinguish positive from negative cases when using age, season, and farm for all scenarios according to F1-scores and receiver operating characteristic analysis. Differences in predictive abilities, factor importance, and the still limited number of predictors in these ML analyses further indicate a potential for building more robust future models once datasets are expanded and more robust algorithms are considered. Overall, correctly identifying factors that may be associated with the prevalence of BRD pathogens and, therefore, recovery of these pathogens is critical for the development of antibiogram programs for evaluation of the antimicrobial susceptibility of BRD pathogens.
Schlagwörter
MeSH-Schlagwörter (MEDLINE)
Stichwörter der Autor:innen
Autor:innen und Institutionen
- Department of Population Health and Reproduction, School of Veterinary Medicine, University of California, Davis, CA 95616. Electronic address: hmonteiro@ucdavis.edu.
- Department of Population Health and Reproduction, School of Veterinary Medicine, University of California, Davis, CA 95616.
- California Animal Health and Food Safety Laboratory, University of California, Davis, CA 95616.
- Department of Population Health and Reproduction, School of Veterinary Medicine, University of California, Davis, CA 95616; Veterinary Medicine Teaching and Research Center, School of Veterinary Medicine, University of…
- Department of Population Health and Reproduction, School of Veterinary Medicine, University of California, Davis, CA 95616. Electronic address: rvpereira@ucdavis.edu.
Förderung
- University of California, Davis
- Council of Fashion Designers of America
- California Department of Food and Agriculture
Zitieren
APA
Monteiro, H. F., Hoyos-Jaramillo, A., Garzon, A., Fritz, H. M., Miramontes, C. C., Matos, I. M., Lehenbauer, T. W., Aly, S., Lima, F. S., & Pereira, R. V. (2025). Antibiogram use on dairy cattle for bovine respiratory disease: Factors associated with bacterial pathogen identification and prediction of bacterial recovery using machine learning. Journal of Dairy Science. https://doi.org/10.3168/jds.2024-26205
Vancouver
Monteiro HF, Hoyos-Jaramillo A, Garzon A, Fritz HM, Miramontes CC, Matos IM, et al. Antibiogram use on dairy cattle for bovine respiratory disease: Factors associated with bacterial pathogen identification and prediction of bacterial recovery using machine learning. Journal of Dairy Science. 2025. doi:10.3168/jds.2024-26205
BibTeX
@article{Monteiro2025,
author = {Monteiro, Hugo F and Hoyos-Jaramillo, Alejandro and Garzon, Adriana and Fritz, Heather M and Miramontes, Craig C and Matos, Isabelle M and Lehenbauer, Terry W and Aly, Sharif and Lima, Fabio S and Pereira, Richard V},
title = {Antibiogram use on dairy cattle for bovine respiratory disease: Factors associated with bacterial pathogen identification and prediction of bacterial recovery using machine learning},
journal = {Journal of Dairy Science},
year = {2025},
doi = {10.3168/jds.2024-26205}
}
EUVIMED