Internal validation of a computerized systems model for health management decision support in growing hogs

A study was designed to conduct an internal validation of a computerized systems model for health management decision support in growing hogs. Evaluation focused on both the alpha-beta tracker, which was employed to predict underlying system variables, and the simulation model itself, which predicted system performance. Both mean absolute prediction error and Theil's u₂ statistic were calculated. Simulation scenarios were designed to highlight specific aspects of the model. In addition, a case example was developed to demonstrate the model's logical consistency and its applicability for assessing the economics of health management decisions. Use of the alpha-beta tracker to project data series, including pig deaths and disease prevalence rates at slaughter, was largely unsuccessful. The model consistently predicted hogs marketed such that 0 < u₂ < 1. Also, decreased disease rates improved both physical and financial performance as expected. Depending on the quarter involved, the maximum bid to achieve a 50% decrease in the prevalence rates of pneumonia at slaughter and a 10% decrease in the prevalence rates of atrophic rhinitis at slaughter ranged from $0.15 to $0.37 per hog marketed for one producer. Future validation efforts should emphasize data quality and the effects of disease on production while seeking system application in a commercial production setting.