Food poisoning: Machines learn about it

Skip to Navigation


  • Published: Feb 15, 2019
  • Author: David Bradley
  • Channels: Chemometrics & Informatics
thumbnail image: Food poisoning: Machines learn about it

Pinpointing poisoning

UGA researchers Xiangyu Deng (front) and Shaokang Zhang led a team of scientists who have trained an algorithm called Random Forest to predict certain animal sources of S. Typhimurium genomes.

A machine-learning approach could be used to pinpoint the animal source of a food-poisoning outbreak, according to US researchers. Specifically, the team has demonstrated efficacy with certain Salmonella enterica outbreaks.

Xiangyu Deng of the University of Georgia Center for Food Safety in Griffin and colleagues published details of their machine-learning approach to food poisoning in the journal Emerging Infectious Diseases. The team used more than a thousand genomes to predict the animal sources, in particular livestock, for Salmonella enterica Typhimurium.

Break out science

“We had at least three outbreaks of Typhimurium, or its close variant, in 2018, explains Deng."These outbreaks were linked to chicken, chicken salad, and dried coconut,” he adds. However, there are more than 2,600 serotypes of Salmonella with "Typhimurium" being just one of them, which makes pinpointing the source difficult. That said, since the 1960s, approximately one in four Salmonella isolates linked to outbreaks reported to US national surveillance were the Typhimurium serotype.

Deng and postdoctoral researcher Shaokang Zhang, worked with colleagues from the US Centers for Disease Control and Prevention, the Food and Drug Administration, the Minnesota Department of Health, and the Translational Genomics Research Institute on the research. According to the Foodborne Disease Outbreak Surveillance System, some 3000 outbreaks of foodborne illness were reported in the USA from 2009 to 2015. Almost a third of them, 900, were caused by different serotypes of Salmonella, including Typhimurium.

The researchers trained a Random Forest algorithm with more than 1300 genomes from samples of S. enterica Typhimurium that were from known sources. After this training process, the algorithm could then predict certain animal sources of novel S. enterica Typhimurium genomes which it was fed.

Genome learning

“With so many genomes, machine learning is a natural choice to deal with all these data," Deng adds. "We used this big collection of Typhimurium genomes as the training set to build the classifier.” He says that “The classifier predicts the source of the Typhimurium isolate by interrogating thousands of genetic features of its genome.” The system had an accuracy of 83 percent. It worked best in predicting poultry and swine sources, followed by bovine and wild bird sources. The system can also give an estimate of the precision of its own predictions, meaning that when the confidence level is high accuracy can be up to about 92 percent.

“We retrospectively analyzed eight of the major zoonotic outbreaks that occurred in the USA from 1998 to 2013,” Deng says. “The classifier attributed seven of them to the correct livestock source.” This work represents a proof of concept and will be improved by the addition of more genomes from disparate sources as they become available. Tracing a strain of foodborne illness more rapidly to its source is critical to cutting an outbreak dead in its tracks.

“Using our method, investigators can better link cases of the same outbreak and better match isolates from food or food processing environments to isolates from sick people,” Deng explains. “This will give investigators more confidence to implicate a specific source that is behind the outbreak.”

Related Links

Emerging Infect. Disease 2019, online: "Zoonotic Source Attribution of Salmonella enterica Serotype Typhimurium Using Genomic Surveillance Data, United States"

Article by David Bradley

The views represented in this article are solely those of the author and do not necessarily represent those of John Wiley and Sons, Ltd.

Follow us on Twitter!

Social Links

Share This Links

Bookmark and Share


Suppliers Selection
Societies Selection

Banner Ad

Click here to see
all job opportunities

Copyright Information

Interested in separation science? Visit our sister site

Copyright © 2019 John Wiley & Sons, Inc. All Rights Reserved