Predictive modeling of bovine respiratory disease outcomes in feedlot cattle: A narrative review

Abstract Predictive modeling, which aims to predict a current or future event, is a potential way to improve identification of cattle at-risk for new onset bovine respiratory disease (BRD) and other BRD-related outcomes, including mortality, treatment response, relapse, and disease severity. Although recently there has been increased interest in predictive models for BRD-related outcomes, this research area remains largely unexplored. Many known risk factors and biomarkers for BRD have yet to be tested for their predictive value for BRD-related outcomes. Of the risk factors and biomarkers that have been assessed for predictive ability, many have not been externally validated in new samples (e.g. other feedlots or in new cohorts of cattle) or implemented in multivariable predictive models. This narrative review aims to discuss 1) special considerations for selecting and implementing predictive models for BRD-related outcomes in feedlot cattle, 2) the current research on predictive modeling for BRD-related outcomes in feedlot cattle, and 3) potential predictor variables. This narrative review highlights the need for research that focuses on developing accurate predictive models for BRD-related outcomes that are both practical and cost-effective for use at the feedlot. Potential predictor variables and methods are suggested.