How AI Tactics May Spare Your GI Tract

Keep meat and dairy foods refrigerated, wash your hands, cook thoroughly—these are some of the best ways to stave off foodborne illnesses, which each year sicken an estimated one in six Americans and cause about 3,000 deaths—but there’s only so much even the most cautious consumer can do.

Now, supported by a $500,000 National Institute of Food and Agriculture grant, University of Maryland researchers are looking to enhance food safety risk assessment models, starting with that reliable “stomach flu” pairing—salmonella and chicken.  

Led by Abani Pradhan, associate professor of nutrition and food science, the project is taking advantage of the branch of artificial intelligence known as machine learning, as well as whole-genome sequencing of pathogens like salmonella, to improve public health through more specific food safety risk assessments better able to predict pathogen outbreaks and guide risk management decisions at the policy level. 

Big data analytics have gained momentum in other fields, including transportation, manufacturing, healthcare, and even finance, but lagged in food safety risk assessment; now, the simultaneous progress of bioinformatics and genomic data has changed that, Pradhan said.

“AI as an emerging technology can take advantage of big data available in the agriculture and food sectors and has the potential to integrate food production, processing, food safety risk factors and genomic data that can transform public health strategies to prevent foodborne diseases and rapidly respond to outbreaks,” he said. 

Pradhan is working with co-investigators Jianghong Meng, director of the UMD Joint Institute for Food Safety and Applied Nutrition (JIFSAN) and the Center for Food Safety and Security Systems, Hector Corrada Bravo, associate professor of computer science at UMD, and Marc Allard of the U.S. Food and Drug Administration.

The process known as quantitative microbial risk assessment (QMRA) uses mathematical and statistical models to understand, predict and prevent risks presented by foodborne pathogens like salmonella, E.coli and listeria. It can predict the behavior and transmission of pathogens across the food production, processing and supply chains, identify areas in the chain that could allow contamination, and estimatethe probability and consequences of adverse public health effects. 

Integrating genomic data using big data analytics techniques can be a game changer for QMRA. “The sheer abundance of information by including molecular and genomic data available should increase the robustness of disease risk estimates by reducing the sources of uncertainty and variability in the QMRA model,” Pradhan said.

He and his team are focusing on salmonella with this particular work to develop the framework to be applied to other foodborne pathogens, since current risk estimates for salmonella in chicken do not take into account the possibilities given rise by the many variations in the pathogen’s genetic profile.

“The idea is to connect that genetic information with the characteristics of the pathogen to bridge the gap between the genes and the food safety aspects for consumers,” he said.