Survival analysis and prediction of cow under mastitis disease in Rwanda using K-nearest neighbors: a retrospective study
Tropical Animal Health and Production · 2026-04 · 58 (3)
DOI: 10.1007/s11250-026-05013-w ↗Kurz gefasst
Fachartikel in der Zeitschrift Tropical Animal Health and Production, online erschienen am 07.04.2026. Verfasst von Maximillian Manzi und 5 weiteren Autor:innen. Beteiligte Einrichtung: Rwanda Agriculture and Animal Resources Development Board, P.O Box 5016, Huye, Rwanda. manzimax2002@yahoo.co.uk. Hauptthemen laut MEDLINE: Mastitis, Bovine. Die Arbeit nennt 21 Literaturangaben.
Abstract (Originalsprache)
This study investigated the prevalence, risk factors, and predictive modeling of mastitis among 337 lactating cows in Rwanda, revealing a high overall mastitis rate of 73.3%. District-level analysis showed slightly higher prevalence in Nyagatare (74.9%) than Gicumbi (71.4%). Mastitis was influenced by farmer demographics, grazing system, hygiene practices, milking frequency, and calf suckling. Kaplan-Meier survival analysis highlighted longer mastitis-free periods in cows under better management practices. Based on the cox regression results, the study revealed that the sex of farmer (HR = 0.091, p-value = 0.0013), sex of household head (HR = 0.048, p-value < 0.001), age of farmer (HR = 1.987, p- value = 0.0405), category of person who cares (HR = 0.185, p- value = 0.024), grazing system (HR = 3.564, p- value < 0.01), frequency of milking per day (HR = 2.827, p- value = 0.0001), washing hands before milking (HR = 2.387, p- value = 0.0011), hygiene of cow (HR = 0.234, p-value = 0.0003), washing milking area (HR = 0.225, p- value = 0.0448), hygiene of cow (HR = 0.323, p- value < 0.02), type of breed (HR = 0.344, p- value < 0.006), calf suckling (HR = 0.441, p- value = 0.001) and lactating cow size (HR = 0.562, p- value = 0.0178) were found to be significant survival risk factors associated with the prevalence of mastitis disease in Rwanda. Finally, machine learning analysis using the K-Nearest Neighbors (KNN) algorithm found that using three teats as input features yielded optimal predictive performance (accuracy: 0.73, AUC: 0.69). These findings underscore the critical role of hygiene, management, and demographic factors in mastitis control and suggest the potential of machine learning in early detection.
Schlagwörter
MeSH-Schlagwörter (MEDLINE)
Stichwörter der Autor:innen
Autor:innen und Institutionen
- Rwanda Agriculture and Animal Resources Development Board, P.O Box 5016, Huye, Rwanda. manzimax2002@yahoo.co.uk.
- Rwanda Agriculture and Animal Resources Development Board, P.O Box 5016, Huye, Rwanda.
- College of Agriculture Animal Science and Veterinary Medicine, School of Food Science and Technology, University of Rwanda, Food Science and Technology Department, P.O Box 210, Musanze, Rwanda.
Zitieren
APA
Manzi, M., Gasana, M., Ndayisenga, F., Munyemana, J., Nkurunziza, F., & Niyonzima, E. (2026). Survival analysis and prediction of cow under mastitis disease in Rwanda using K-nearest neighbors: a retrospective study. Tropical Animal Health and Production, 58(3). https://doi.org/10.1007/s11250-026-05013-w
Vancouver
Manzi M, Gasana M, Ndayisenga F, Munyemana J, Nkurunziza F, Niyonzima E. Survival analysis and prediction of cow under mastitis disease in Rwanda using K-nearest neighbors: a retrospective study. Tropical Animal Health and Production. 2026;58(3). doi:10.1007/s11250-026-05013-w
BibTeX
@article{Manzi2026,
author = {Manzi, Maximillian and Gasana, Method and Ndayisenga, Fabrice and Munyemana, Jacques and Nkurunziza, Florien and Niyonzima, Eugene},
title = {Survival analysis and prediction of cow under mastitis disease in Rwanda using K-nearest neighbors: a retrospective study},
journal = {Tropical Animal Health and Production},
year = {2026},
volume = {58},
number = {3},
doi = {10.1007/s11250-026-05013-w}
}
EUVIMED