Software to predict food dangers

In an effort to prepare for a possible bioterrorism attack, the Food and Drug Administration plans to develop a modeling tool to predict scenarios and outcomes of food contamination.

FDA officials want to project the possible result and cause of a food-borne threat using several variables such as symptoms of an illness and characteristics of a public health response.

"This system should be able to model the flow of food, but also let us manipulate variables to create what-if scenarios," said Morris Potter, FDA's project officer and lead scientist for epidemiology in the Center for Food Safety and Applied Nutrition.

For example, the tool would notify FDA officials that they should examine the public health and economic consequences if a food-borne hazard is introduced during processing or distribution. By predicting these eventualities, officials would be able to focus on intervention efforts, Potter said.

"It helps us predict what happens when certain things go wrong so we can be better prepared, and it helps us prioritize," he said.

The FDA was accepting proposals for the project through Aug. 8. Once a contract is awarded, FDA officials expect to have a basic model within a year. "The model should at the end of the year be a fine model, but it might not have all the bells and whistles," Potter said.

The tool would outline where to find data, such as food production and distribution processes, human consumption patterns, and illness symptoms and outcomes. It will also demonstrate how to extract that data. Officials expect to be able to run the system in reverse by entering outcomes to predict the origin.

Much of the tool's value lies in the scope of what the FDA intends to model, such as contamination of the water supply as well, said Harold Reetz, Midwest director of the Potash and Phosphate Institute, which focuses on agriculture research and education.

"Each [situation] is going to have its own model of responses," he said. "Identifying some of the vulnerabilities and understanding how they operate is the first step."

Another element is having the right people in place to carry out the response dictated by the model, Reetz said. Unless the proper people are trained to deal with the outbreak or contamination, the response modeling will not succeed. "That's useful not just for terrorism, but also for natural disasters," he said.

John Corby, director of public policy for the Association of Food and Drug Officials, said officials could tailor their response depending on where the contamination occurs and the scope of the event. The tool would also take into account the entire life cycle of a food product, giving a broader picture of the industry and the possibilities for contamination.

"It always seemed to be the restaurant, grocery store or processor who was blamed for [an illness]. We never incorporated things at the grower, producer or even the consumer level," Corby said. "Now we recognize we are all involved. If this system can be useful in bringing 'the farm to the table' concept into action, it has to be useful."

Jim Mock, president of CropVerifeye LLC, which develops solutions for food traceability, said the modeling tool could be one of many approaches to detecting and responding to a food-borne outbreak. But the tool needs a system for food traceability to support the efforts.

"It sounds like they're looking at a predictive tool. If you had that with a traceability tool, you would be [doubly] covered," Mock said. "If you relied only on a predictive tool, I would be worried about that. They're never perfect."

The FDA does have plans to create a food-tracking system to monitor where food has been and where it's going as it travels from ports, factories and distributors within the United States. FDA officials could avoid unreliable paper invoices and quicken the response to an outbreak. The development of that system is one of several provisions mandated in the Bioterrorism Act of 2002.

***

Forecasting bad apples

The Food and Drug Administration expects vendors to complete three tasks, as outlined in its request for proposals for an analytical modeling tool to track food contamination:

* Create a PC-based predictive modeling tool that maps the flow of foods from production to consumption.

* Identify the necessary data requirements, such as consumer practices and illness symptoms.

* Provide descriptions of where to find and how to gather the data.