Exploring the predictive capability of machine learning models in identifying foot and mouth disease outbreak occurrences in cattle farms in an endemic setting of Thailand

Abstract

Occurrences of foot and mouth disease (FMD) outbreaks in cattle farms in Thailand have been significantly harmful to the cattle industry for the past decade. A prediction of FMD outbreaks based on relevant risk factors with a high prediction accuracy is important for authorities to develop a plan for preventing the outbreaks. Data-driven tools are widely accepted for their prediction abilities, but an application of these techniques to FMD outbreak prediction is very limited. The objectives of this study were to develop prediction models of FMD outbreaks among cattle farms using machine learning (ML) classification algorithms including classification tree (CT), random forests (RF), and Chi-squared automatic interaction detection (CHAID) and to compare the predictive performance of the developed models. Data from 225 FMD and 608 non-FMD outbreak farms from an endemic setting were analyzed using ML methods. The CT, RF, and CHAID methods were utilized to develop predictive models, and their prediction capabilities were compared. The results showed that models developed using ML methods have an acceptable to excellent ability to predict the occurrence of FMD outbreaks. The RF model had the highest accuracy and the value of area under the operating characteristic curve in predicting the occurrence of an FMD outbreak. Meanwhile, the CT and CHAID models delivered comparable results. In this study, we demonstrated the capability of machine learning algorithms to predict FMD outbreaks using actual FMD outbreak data from the endemic setting and provided a new insight into the prediction of FMD outbreaks. The ML techniques demonstrated herein may be used as a prediction tool by the relevant authorities to predict the occurrence of FMD outbreaks in cattle farms.