The Food and Drug Administration (FDA) announced today a “new era of smarter food safety” under which it will test artificial intelligence and machine learning to better assess foodborne illness risks from imports and explore blockchain for tracking all foods before they hit grocery store shelves.
In February, the FDA announced pilot programs focused on tracking the movement of medicines throughout the supply chain undere the authority of the Drug Supply Chain Security Act of 2013. The pilots include the use of blockchain.
The agency’s latest moves are part of an effort to standardize and implement “new and emerging” technologies in food tracking systems in accordance with the Food Safety Modernization Act (FSMA) of 2011, which vastly expanded FDA oversight, including giving it recall authority. FSMA was the direct result of a number of foodborne illness incidents during the 2000s.[ Further reading: 22 free tools for data visualization and analysis ]
When it comes to food traceability, many suppliers rely on a largely paper-based system of taking one step forward to identify where the food has gone and one step back to identify the source. But rarely is food traceable to its origin.
Last year, 210 people from 36 states became ill – and five died – from an E. coli bacteria outbreak traced to romaine lettuce. Because the origin of the lettuce could not immediately be ascertained, the Centers for Disease Control and Prevention (CDC) issued a warning for consumers to avoid all types of romaine lettuce; hundreds of stores threw out millions of dollars worth of produce. It was weeks before the tainted lettuce was eventually traced back to Yuma, Ariz.
“We’re at this inflection point in society where many of these technologies emerging have been used in the private sector around us and now we’re looking at what role they can play to help us address some of the remaining health challenges that we have,” said Frank Yiannas, FDA deputy commissioner for Food Policy and Response. “For blockchain, the area that jumps out at me most quickly is the role it could play to enhance tracking and tracing food.”
Part of the FDA’s job it so assess the risks of food imported into the U.S., something that has increased “exponentially” over the past several years.
“Imported foods are increasing because we’ve gone from a place where you went to a concrete structure [to get your groceries] to the world becoming your grocery store; you’re able to order foods from around the world,” Yiannas said.
The FDA has an electronic tracking system that evaluates risk using a database screening system that combs through every distribution line of imported food, ranking risk based on historical data input by people who classify foods as higher or lower risk. Higher-risk foods get more scrutiny at ports of entry.
Artificial intelligence and machine learning can strengthen the FDA’s predictive capabilities to evaluate which imported foods pose the greatest risks and, thus, require additional inspections.
Turning to new tools, Yiannas said, means the FDA can analyze more data than ever before and make screening and risk predictions more dynamic instead of relying heavily on people with subject-matter expertise.
“The idea is…we’re working a little smarter,” Yiannas said.
Prior to taking on his role at the FDA in December, Yiannas was Walmart’s vice president in charge of food safety where he oversaw the implementation of large-scale pilots of a blockchain-based food tracking system based on IBM’s Food Trust cloud service.
Walmart’s pilots showed the amount of time needed to trace a food item from store to farm was reduced from seven days (using a legacy system) to just 2.2 seconds (with a blockchain ledger).
Last fall, Walmart asked its food suppliers to upload their produce data to a corporate blockchain ledger within a year to enable end-to-end traceability of vegetables back to the farm where they were grown.
Earlier this month, grocery chain Albertson’s announced it’s joining dozens of retailers and suppliers on the blockchain-based IBM Food Trust network to keep track of items in the food supply chain such as romaine lettuce.
Blockchain networks, such as GrainChain, have also been used to increase the efficiency of agricultural settlement payments to farmers and suppliers while providing the immediate availability of tradable commodities to buyers.
The FDA is involved in overseeing the safety of about 80% of food products in the U.S. So while the agency is looking to new technologies to update its own internal systems, it’s also seeking help the private sector standardize on methods that will scale its ability to track and trace food supply chains.
Consumers, Yiannas said, are driving the conversation around traceability as they voice concerns about the origin of what they eat.
The FDA plans to hold a public meeting later this year to discuss smarter food safety, seek industry input and share ideas on its overall strategy and initiatives. The agency will be considering the role standards play in scaling food traceability systems and how those systems can be made more interoperable because there are many blockchain platforms in the market today.
“Our role is to be a guide and a thought partner on how these things scale so that both the public and private sectors, and ultimately consumers, can benefit and trust it,” Yiannas said. “We believe there won’t be a one solution that dominates the entire food system, so for this to scale they will truly have to be interoperable.”
Most critical to expanding food traceability among private growers, shippers and retailers through a blockchain-based network will be setting the rules around self-governance. Blockchains are built around a consensus mechanism where users determine what transactions can be added to a ledger as well as how to address business process issues when they arise.
Getting an agreement on an appropriate governance model is one of the biggest challenges for any enterprise blockchain initiative, according to Martha Bennett, a principal analyst at Forrester Research. ” I know of projects that have been halted or never took off due to failure to agree on one,” she said.
While there’s no set timeline for rolling out a production screening system that employs AI and blockchain, Yiannas said he expects the technology, once piloted, to catch on quickly.
“Over the years, I’ve learned these things can scale more rapidly than what people used to think. We’ve seen how some of these technologies can emerge pretty quickly, whether it’s ride sharing or Airbnb,” Yiannas said. “I think the key is once you figure out how to get ecosystems to collaborate, once we figure out the role of the public sector and the standards of interoperability, solutions can escalate.”