This study substituted cheese solids with other dairy ingredients (whey, WPC, NFDM) and determined the botulinal safety of nonstandard process cheese products. Researchers also developed machine learning tools (Bayesian networks, decision tree, support vector machine and artificial neural networks) to supplement the statistical models used to predict toxin production. The study found that WPC or NFDM does not negatively affect the botulinal safety of model process cheese products compared to standard formulas. When the machine learning strategies are coupled with logistic regression, it is possible to separate formulations that are potentially toxic from those that are safe.
Effects of Cheese Solids on Clostridium Botulinum in Processed Cheese Products
More information
Investigators
Johnson, Eric
Institution
University of Wisconsin - Madison
Start date
2001
End date
2005
Funding Source
Categories
Commodities