Related Experiment Video
Updated: Aug 9, 2025
11:02
The Use of an Automated System GreenFeed to Monitor Enteric Methane and Carbon Dioxide Emissions from Ruminant Animals
Published on: September 7, 2015
22.1K
Analysis of sequential ruminal temperature sensor data from dairy cows to identify cow subgroups by clustering and
Eri Furukawa1, Yojiro Yanagawa2, Akira Matsuzaki3
1Laboratory of Theriogenology, Graduate School of Veterinary Medicine, Hokkaido University, Sapporo, Hokkaido 060-0818, Japan.
The Journal of Reproduction and Development
|February 21, 2023
Summary
A new machine learning model using ruminal temperature (RT) data can predict dairy cow calving with 87.5% sensitivity. The model shows potential but requires refinement for specific cow subgroups with varying prepartum temperature changes.
Area of Science:
- Animal Science
- Veterinary Medicine
- Machine Learning Applications
Background:
- Accurate calving prediction is crucial for dairy herd management and animal welfare.
- Prepartum changes in physiological parameters, such as ruminal temperature (RT), may indicate impending parturition.
- Identifying distinct patterns in these physiological changes could improve predictive accuracy.
Purpose of the Study:
- To evaluate a supervised machine learning model for calving prediction using ruminal temperature (RT) data in dairy cows.
- To investigate the existence of cow subgroups based on prepartum RT changes.
- To compare the model's predictive performance across identified subgroups.
Main Methods:
- Collected RT data from 24 Holstein cows using an RT sensor system at 10-minute intervals.
- Calculated average hourly RT and expressed data as residual RTs (rRT).
- Developed a support vector machine model using five features indicative of prepartum rRT changes.
Main Results:
- A mean rRT decrease was observed starting approximately 48 hours before calving.
- Two distinct cow subgroups were identified based on the timing and magnitude of rRT decrease.
- The calving prediction model achieved 87.5% sensitivity and 77.8% precision for predicting calving within 24 hours.
- Model sensitivity varied significantly between the two identified cow subgroups (66.7% vs. 100%).
Conclusions:
- Supervised machine learning of ruminal temperature data shows significant potential for efficient calving prediction in dairy cows.
- The model's predictive performance is influenced by prepartum ruminal temperature change patterns.
- Further model refinement is necessary to enhance prediction accuracy for all cow subgroups.
