On-farm evaluation of a predictive model for Australian beef and sheep producers’ vulnerability to an outbreak of foot and mouth disease

To explore Australian sheep and beef producer vulnerability to an emergency animal disease outbreak, Bayesian Network models have been developed, with the ultimate goal of creating risk management tool for outbreak preparedness. These models were developed using multiple stakeholder elicitation including modelling experts, epidemiologists and on-farm stakeholders, including on-farm/survey data. An evaluation of the model's predictive capacity was conducted, using independent, blinded on-farm vulnerability assessments. Nine properties were visited, four each with sheep and beef enterprises, and one mixed enterprise. There were some discrepancies between the model predictions and on-farm assessment in the beef enterprises, with greater disparity with the sheep properties. Discrepancies between the model predictions and on-farm assessments have created opportunities for examination of the data collection process for the model development, the model itself and the on-farm assessment process. Bayesian Network approaches that allow for the inclusion of both continuous and discrete variables may improve the usefulness of these models, avoiding the loss of nuanced data by the need for discretisation of continuous variables, as will the inclusion of input from on-farm stakeholders in model development. Future work includes more data collection to improve the sensitivity of the model predictions, and a deeper, systemic exploration of the factors that may impact Australian producers’ vulnerability to an emergency animal disease outbreak.