Predictive model of aflatoxin contamination risk associated with granary-stored corn with versicolorin A monitoring and logistic regression

Predictions of aflatoxin (AF) in grain at post-harvest can be useful for ensuring the safety of stored grain. Versicolorin (Ver) A, a precursor of AFB1, can serve as an early indicator of AF contamination, even when AFs themselves are present at undetectable levels. In the current research, we developed a probabilistic model based on logistic regression and Ver A levels to estimate the risk of AF contamination in stored corn. Moisture content, aflatoxigenic fungal load, and initial and maximum values of Ver A in the first three sampling cycles were experimentally determined as the four important factors for the probabilistic model. Both internal and external model validations were shown to be high at 96.4% and 93.3%, respectively. For high-risk samples, a precise model was developed to predict the maximum period of safe storage, which can be useful for decision-making by the stakeholders in feed and food supply chain. Our findings provide a basis tool for establishing an early warning system for AF contamination in granaries, which can improve global food safety.