Related Experiment Video
Updated: Jan 9, 2026
04:56
An Efficient Single—Person Technique for Milk Sampling from Laboratory Mice
Published on: March 28, 2025
1.5K
Machine learning-based clinical mastitis detection in dairy cows using milk electrical conductivity and somatic cell
Lihong Pan1,2, Xiao Chen1,3, Ding Han1,4
1State Key Laboratory of Transducer Technology, Aerospace Information Research Institute, Chinese Academy of Sciences, Beijing, China.
Frontiers in Veterinary Science
|December 1, 2025
Summary
Machine learning models accurately detect bovine mastitis using somatic cell count (SCC), outperforming electrical conductivity (EC) methods. SVM achieved 95.6% accuracy, while FNN showed superior pattern recognition for improved dairy herd health.
Area of Science:
- Veterinary Medicine
- Dairy Science
- Machine Learning Applications
Background:
- Bovine mastitis significantly impacts dairy production economics.
- Current detection methods like electrical conductivity (EC) lack specificity.
- Somatic cell count (SCC) is a reliable biomarker but faces implementation challenges with existing sensors.
Purpose of the Study:
- To develop and evaluate machine learning models for accurate bovine mastitis detection.
- To compare the performance of logistic regression (LR), support vector machines (SVM), and feedforward neural networks (FNN).
- To assess the utility of EC, SCC, and their combination as input features.
Main Methods:
- Collected data from 93 cows across four dairy farms.
- Trained and tested LR, SVM, and FNN models using EC and SCC data.
- Evaluated model performance using accuracy, sensitivity, and Area Under the Curve (AUC).
Main Results:
- SCC-based models significantly outperformed EC-based models.
- SVM achieved 95.6% accuracy and 100% sensitivity with SCC.
- FNN model demonstrated the highest AUC (0.981), indicating strong diagnostic capability.
- Combined EC and SCC data showed potential for enhanced robustness in specific scenarios.
Conclusions:
- Somatic cell count (SCC) is a superior indicator for bovine mastitis detection compared to EC.
- Machine learning, particularly FNN and SVM, offers powerful tools for mastitis monitoring in dairy settings.
- Future research should focus on real-time deployment with high-precision SCC sensors.
