Prediction model identifies nursery pigs at risk of health treatments using feeding and drinking patterns, suggesting improvements in swine management.
Key Points
Models showed some ability to rank nursery pigs by health treatment probability, indicating potential for early identification.
Accuracy was assessed using area under the curve and correlation metrics, with Random Forest demonstrating the highest performance.
Dynamic disease challenges limited predictive power of treatments across different batches, emphasizing variability in outcomes.
Incorporating additional data features may enhance early identification of pigs in need of health interventions.