November 18, 2019

Interpretive Summary: Boosted trees to predict performance of growing-finishing pigs

Interpretive Summary: Boosted trees to predict pneumonia, growth, and meat percentage of growing-finishing pigs

By Anne Wallace

There are many challenges associated with animal production. Some of these include maintaining animal health and welfare, determining optimal feed formulation, and efficient management of the environmental and land on commercial farms. In order to optimize a commercial farm, increasing production efficiency and reducing losses due to illnesses of growing animals, such as pneumonia, are necessary.

The ability to identify problems that may impact production in advance by using predictive models is therefore a helpful tool that may minimize production losses. Tree induction is one such predictive model that can help aid the decision-making process in commercial farms to increase growth and performance. However, as with any predictive model using decision trees, inaccuracy of the models when applied to farms may be problematic.

In this study published in the October 2019 issue of the Journal of Animal Science, researchers evaluated the tree induction model “boosted trees” to see if it could accurately predict performance in three-month-old pigs until slaughter (that is, from the onset of the growing-finishing phase).

The authors reported that the model was able to successfully identify pigs prone to growth aberrations and pneumonia, but not pigs prone to low meat percentage. This study suggests that predictive models can be useful in predicting factors that may affect performance but have limitations.

More studies looking at different models or multiple models may be helpful in order to accurately predict factors that may affect performance on commercial farms.