The Aflatoxin Risk Indicator is a model framework that looks at the mechanism by which aflatoxins are produced and contaminate of corn, including temperature and water activity, to predict aflatoxin contamination risk where the crop is grown. Gro’s framework has been used by some of the world’s largest food companies. Users can incorporate their own food safety data to assess contamination risk in specific locations during the growing season. This allows food safety teams to flag the risk of aflatoxins in their supply chain and to take preventive measures in sourcing and procurement.
Customers Use the Model to
Why It Matters
Aflatoxins, a virulent type of mycotoxin, are a major risk to the global food supply chain. Some 4.5 billion people globally are chronically exposed to aflatoxin in their food, according to an article in the journal Nature. Damages stemming from aflatoxin and other mycotoxins are estimated to cost US farmers and livestock producers up to $1 billion a year. Global food company Mars Inc. has said it rejects up to 70% of the peanuts delivered to its factory gate in India because of aflatoxin contamination.
There are hundreds of mycotoxin types, but aflatoxin and vomitoxin stand out for their impact on humans and livestock. Aspergillus and Fusarium, the fungi responsible for aflatoxin and vomitoxin, respectively, thrive under different conditions, and can provide clues to enable building predictive models of mycotoxin risk. Plants are most susceptible to Aspergillus under drought conditions. High temperatures and inadequate moisture during grain-fill leave the plants vulnerable while the Aspergillus mold is able to flourish. Later in the season, if precipitation levels increase and humidity is high, the problem can be exacerbated. Fusarium, which produces vomitoxin, thrives under wet conditions, with cool to moderate temperatures. Data available in the Gro platform allows users to monitor weather conditions and environmental factors that increase or decrease the risk of infection from mycotoxin-producing molds. Academic studies looking for such correlations in the past have been limited in scope and region, partly due to the difficulty of processing large amounts of geospatial data and modeling complex relationships. Gro has aggregated numerous global, pixel-level datasets down to the district or county level. With this pre-computed data, users can create machine-learning, algorithm-based models to estimate risk.