Machine learning algorithms can predict tail biting outbreaks in pigs using feeding behaviour records.

Area of Science:

• Animal Science
• Machine Learning
• Animal Welfare

Background:

• Tail biting in pigs is a significant welfare concern, leading to health issues and economic losses.
• Early identification of tail biting precursors is essential for implementing timely preventive strategies.

Purpose of the Study:

• To develop and evaluate machine learning algorithms for real-time prediction of tail biting outbreaks.
• To utilize electronic feeder data on feeding behavior for predicting tail biting events.

Main Methods:

• Seven machine learning algorithms were assessed using daily feeding data from 65 pens of grower-finisher pigs.
• Data splitting involved random distribution and pen-based selection for training and testing.
• The K-nearest neighbour algorithm was specifically evaluated for its predictive performance.

Main Results:

• The K-nearest neighbour algorithm achieved 78% prediction of upcoming tail biting events with 96% accuracy when trained on data from the same pen.
• Different data splitting methods influenced model performance, highlighting the importance of data structure for prediction.

Conclusions:

• Machine learning models show promise for real-time tail biting prediction in swine.
• Integration into automatic feeder systems could enable proactive welfare management and reduce tail biting incidents.