Digesting AI for Food Safety in Restaurants and Grocery Stores

Approximately 48 million people every year get sick from foodborne illnesses, according to the U.S. Centers for Disease Control and Prevention (CDC), which says the most common causes of food poisoning are foodborne pathogens like Salmonella, Listeria and E. coli. While some incidents are the result of poor food preparation, hygiene and handling at home, many cases are caused by kinks in the food system itself.

Edge computing and new artificial intelligent (AI) food safety applications can help manage those events by leveraging big data in food safety practices across the food system, from the farms that produce food and the trucks that transport it to the restaurants that serve it and the retail stores that sell it.

How Farmers Are Finding an Edge

Farmers and food manufacturers can use edge computing near food production areas. This real-time data processing in agriculture can improve efficiency and quality control.

Dr. Nadia Sabeh, president and founder of Dr. Greenhouse, works with Forever Feed Technologies to develop automated sprouted grain (ASG) to feed more than 5,000 cows. Instead of in a field, which would require more horizontal space and a larger footprint, the ASG is grown on shelves. The organizations use sensors in the feed mill to identify which specific plant groups need water, light or cooling.

Though AI is beneficial for tracking, researchers must account for multiple variables. If a model is trained on what happened in February, for example, and March traditionally has different temperatures and precipitation levels, can the model adjust efficiently?

“What’s most important to me is the consistency,” Sabeh said. “I’d be interested in a computer model that said, ‘This is what you need to do not just to increase your yield, but to have the most consistent yield.’”

Visual models benefit production, too. Ben Miller, executive vice president of regulatory and scientific affairs at The Acheson Group, said crop researchers can use microscopy to check for pathogens. These visual models could help prevent foodborne illnesses, such as recognizing Salmonella or E. coli on a lettuce leaf.

“You can pull that leaf off, put it under a microscope and run your visually trained model to say, ‘I see something. How does that compare to what I’ve been trained on?’” Miller said. 

Organic Valley manages a network of more than 2,000 family farmers. The company uses Nutanix software to simplify operations and improve performance, offering ERP solutions, warehouse management systems and supply chain analyses. With the extra support, it can work on other initiatives.  

“Our network, storage and compute teams were all working weekends and doing updates because there was a fear of breaking something in a complex operation,” said Nicholas Korte, director of technology operations at Organic Valley. 

“When we went to Nutanix, it was like, ‘Hey, we have an opportunity to simplify and take that time back for ourselves.’” 

AI in the Food Supply Chain

Even after items depart farms and factories, food safety measures remain critical. Traceability can reduce recall impacts by as much as 95%, Food Logistics magazine reported in a 2022 article. Tools like Index Biosystems’ BioTags connect product data to trace the entire supply chain. 

Tyler Williams, CEO of ASI Food Safety, said that visibility is essential for maintaining proper safety standards.   

“If I have milk, for example, I know the temperature when it’s sitting in storage, when it’s in transit and when it’s at the store,” Williams said. “If there are any variances, we’re able to catch those things faster across hundreds of thousands of different SKUs.”

En-route traceability is one way to use AI for foodborne illness prevention. Shelf-life risk for perishable items is another. Much like retailers utilize the edge to enhance customer and employee experience, manufacturers are adopting edge computing in food safety to identify potential issues.

“Larger manufacturers use edge computing to monitor and predict recalls,” Williams continued. “If four people got sick, they can see where they purchased from and be more accurate in identifying the culprit.”

Forecasting in Restaurants and Retail

The COVID-19 pandemic emphasized cloud concepts for restaurants, from data sharing to ghost kitchens. Nathan Jarvis, a former chef and founder of Jarvis Hospitality and Advising, said AI can be used for predictive forecasting.

“In retail grocery settings, automatic sensors provide minute-by-minute data and establish trends,” Jarvis said. “They can predict when a refrigeration unit will go down and catch it before the refrigerator dies.”

This technology also improves food safety for individual items. For example, a human might only periodically check a display of rotisserie chickens. With AI sensors in place, a retailer can more accurately read its products.

“They can know that each chicken hits the right temperature each time, instead of one thermometer hitting the right chicken,” Jarvis said.

Food processing requires critical control points where companies collect data at certain frequencies. Predictive analytics could help identify a faulty piece of equipment or other malady.   

“More immediate opportunities are in engineering failure point analysis, so you can predict when a part of the process might go off the rails from a food safety standpoint,” Miller explained. 

“Anywhere you have a well-trained model looking at a particular problem and relatively high-quality data, something can be done.”

The Future of Food Safety is Sustainable

Food safety and sustainability go hand-in-hand. While the horizon is bright, the industry must maintain realistic expectations.

“People make these big claims,” Sabeh said. “We’re going to reduce water use by 94%, land use by 90%, blah blah blah…” 

“Reducing water usage by 25 or 50% is a big number, but those lofty goals are a disservice to us.”

Emerging categories in food science are also worth watching. Jarvis pointed to mushroom mycelium, which modifies proteins to remove graininess and flavor changes. Fungi-based options require up to 99% less energy, land and water than compared to raising cattle.

“It performs better in a product, tastes better and takes away some of the negative aspects of plant-based proteins,” Jarvis said. “We’re going to find all sorts of interesting new uses and possibilities.”

There’s still plenty of research and development to be done around AI food safety practices. Williams therefore stressed the importance of sound data collection and using AI as a complement to human experience and reasoning.

“The information coming out is only as good as what we’re putting into it,” he said. “AI shouldn’t be your whole program. Tailor it to your processes and operations.”

Editor’s note: Learn more about Nutanix’s AI platform, GPT-in-a-Box, and hybrid multicloud. 

Joey Held is a writer and podcaster based in Austin, Texas, and the founder of Fun Fact Friyay and Good People, Cool Things. Connect with him on Twitter or LinkedIn.

© 2024 Nutanix, Inc. All rights reserved. For additional information and important legal disclaimers, please go here.