Import Screening Pilot Unleashes the Power of Data and Leverages Artificial Intelligence
By: Stephen M. Hahn, M.D., Commissioner of Food and Drugs
I frequently emphasize the importance of data in the U.S. Food and Drug Administration’s work as a science-based regulatory agency, and the need to “unleash the power of data” through sophisticated mechanisms for collection, review and analysis so that it may become preventive, action-oriented information.
As one example of this commitment, I would like to tell you about cross-cutting work the agency is undertaking to leverage our use of artificial intelligence (AI) as part of the FDA’s New Era of Smarter Food Safety initiative. This work promises to equip the FDA with important new ways to apply available data sources to strengthen our public health mission. The ultimate goal is to see if AI can improve our ability to quickly and efficiently identify products that may pose a threat to public health.
One area in which the FDA is assessing the use of AI is in the screening of imported foods. Americans want to enjoy a diverse and available food supply. They also want their food to be safe, whether it’s domestically produced or imported from abroad.
So we launched a pilot program in the spring of 2019 to learn the added benefits of using AI, specifically machine learning (ML), in our import-screening processes. Machine learning is a type of AI that makes it possible to rapidly analyze data, automatically identifying connections and patterns in data that people or even our current rules-based screening system cannot see.
The first phase of this pilot was a “proof of concept” to validate the approach we’re taking. We decided to test this approach on imported seafood to assess the utility of using AI/ML to better target seafood at the border that may be unsafe.
Why seafood? Because the U.S. imports so much of it. Upwards of 94 percent of the seafood Americans consume each year is imported.
Strengthening Our Predictive Capabilities, a Proof of Concept
We embarked on the proof of concept by training the ML screening tool, using years of retrospective data from past seafood shipments that were refused entry or subjected to additional scrutiny, such as a field exam, label exam or laboratory analysis of a sample. This gave us an idea of how much our surveillance efforts might be improved using these technologies.
The results are exciting, suggesting that this approach has real potential to be a tool that expedites the clearance of lower risk seafood shipments, and identifies those that are higher risk. In fact, this is great news. The proof of concept demonstrated that AI/ML could almost triple the likelihood that we will identify a shipment containing products of public health concern.
The implementation team is now working to apply the AI/ML model algorithm to field conditions as part of the second phase of this work, an in-field pilot again focusing on imported seafood, and that’s where we are now. As part of the in-field pilot, the model will be applied to the screening methods used to help FDA staff decide which shipments to examine and will then provide information about which food in the shipment to sample for laboratory testing. We will then compare the results to the recommendations made by our current system.
We see this opportunity as a critical step in the FDA employing the power of AI across the spectrum of product and process challenges facing the agency. Our initial proof of concept results indicate that such innovative approaches hold great promise in further strengthening protections for consumers.
Unleashing the Power of Data to Keep Americans Safe
The pilot taps into two important new initiatives at the FDA. In addition to the New Era of Smarter Food Safety, it also reflects the priorities embodied in our Technology Modernization Action Plan – or TMAP.
On July 13, the FDA released a blueprint for the New Era of Smarter Food Safety outlining how the agency plans to leverage new technologies and approaches to create a more digital, traceable and safer food system.
When we developed the blueprint, we knew that AI technology could be a game changer in expanding the FDA’s predictive analytics capabilities, enabling us to mine data to anticipate and mitigate foodborne risks. The pilot is revealing the specific, immediate benefits that this technology could have in helping us ensure the safety of imported foods.
The TMAP describes important actions we are taking to modernize our technology information systems — computer hardware, software, data, analytics, advanced technology tools and more — in ways that accelerate the FDA’s pursuit of our public health mission.
Additionally, the plan lays out how the agency intends to transform our computing and technology infrastructure to position the FDA to close the gap between rapid advances in product and process technology and the technology solutions needed to ensure those advances translate into meaningful results for American consumers and patients. The TMAP provides a foundation for the development of the FDA’s ongoing strategy around data itself — a strategy for the stewardship, security, quality control, analysis and real-time use of data — that will illuminate the brightest path and the best tools for the FDA to enhance and promote public health.
While both of these initiatives were well underway before the COVID-19 pandemic, lessons learned during this time of crisis have underscored the need for more real-time, data-driven approaches to protecting public health.
Scaling a Mountain of Data
The pilot also gives us the opportunity to learn how to untether the knowledge we need from the huge volume of data we have from screening millions of import shipments every year. In 2019, the FDA screened nearly 15 million food shipments offered for import into our country for sale to American consumers. Last year, the U.S. imported about 15% of the food we consume and that percentage continues to increase.
The FDA has a massive amount of data about these shipments and about the companies that are producing and processing the food, offering it for import, and selling it in the U.S. marketplace. In fact, every year the FDA collects tens of millions of data points on imports alone, and we screen all the data associated with every shipment of food against the information in our internal databases. One of the major goals of our pilot is to assess the ability of AI/ML to more quickly, efficiently, and comprehensively take advantage of all the data and information residing in our systems.
In fact, we believe that we can use the knowledge that ML provides to know where best to concentrate our resources to find potentially unsafe products. In addition to improved import surveillance resources, the intelligence that ML can extract from the stores of data the FDA collects can also inform decisions about which facilities we inspect, what foods are most likely to make people sick and other risk prioritization questions.
The bottom line is this: times and technologies change, and the FDA is changing with them, but the goal remains the same – to do everything in our power to strengthen the way we protect public health.