This article was published in Scientific American’s former blog network and reflects the views of the author, not necessarily those of Scientific American
Vegetable farmers across Europe were affected during the summer of 2011 when a foodborne illness outbreak caused widespread concern about the safety of salad ingredients grown in Europe. Russia even banned the import of all fresh vegetables from the European Union. Sprouts were declared the culprit around 21 days after the beginning of the outbreak investigation by Germany’s Robert Koch Institute, but economic damage was already done, not to mention the illness and death that occurred. Now, a group of IBM researchers and their colleagues have developed an approach that might provide a faster way to identify a guilty food, with the hope of saving both lives and money.
Cases of foodborne illness are tracked by public health departments, but identifying the responsible food and where it came from can be challenging. The method developed by the IBM researchers combines public health case reports with food product sales data to identify potentially contaminated products.
“The question was could we leverage the data that’s already collected by, for instance, retailers like supermarkets,” said James Kaufman, an IBM scientist who led the research collaboration between IBM, Johns Hopkins and the Federal Institute for Risk Assessment in Germany.
On supporting science journalism
If you're enjoying this article, consider supporting our award-winning journalism by subscribing. By purchasing a subscription you are helping to ensure the future of impactful stories about the discoveries and ideas shaping our world today.
Kaufman and his colleagues tested the method using a real set of food sales data and simulated outbreak information. Using only 50 foodborne illness case reports, the average success rate for identifying a specific guilty product was over 80 percent. Even fewer case reports were needed to generate a small group of products very likely to contain the guilty product. In a real situation, this group could then be tested to hone in on the guilty product.
“It’s just going to give investigators more information, and it’ll allow them to hopefully intervene more quickly,” Kaufman said.
Next steps in improving the method include studying the effect of "noise," or unrelated case reports, which were not present in the study simulation. In a real situation, some reports of foodborne illness will be related to the outbreak and useful for identifying the guilty product, while others may be unconnected cases of illness.
Employing the approach to real outbreak scenarios would also likely require a public-private partnership, according to Kaufman. Supermarkets and other retailers would have to be willing to provide their sales data to public health officials so that the officials could combine it with foodborne illness case reports. Since outbreaks can cause economic loss to companies beyond the guilty party (as 2011 showed), maybe private industry will find such a partnership worthwhile.