Las Cruces – U.S. Food and Drug Administration inspectors have doubled the number of risky products they can find among the millions of imports arriving at U.S. ports each year, thanks to an innovative system developed at New Mexico State University with assistance from two small business consulting companies.
An open house at NMSU June 15 will showcase the PREDICT project, which has earned a 2010 FDA Honor Award.
The PREDICT (Predictive Risk-based Evaluation for Dynamic Import Compliance Targeting) system allows the FDA to focus inspections on the riskiest shipments of food, medicines, pharmaceuticals and veterinary products, while speeding up entry for products that are safe.
The FDA officially began using PREDICT at the Port of Los Angeles in September 2009. The system is now also being used by the New York Import District and will be rolled out to all FDA import districts across the U.S. by the end of 2010.
"The PREDICT screening system is innovative because it attaches a risk score to import companies and their products the higher the score, the greater the risk," said Robert Silver, assistant director of NMSU's Center for Animal Health, Food Safety and Bio-Security, College of Agricultural, Consumer and Environmental Sciences. Working with the FDA, NTELX of McLean, Va., and Natural Selection, Inc. of La Jolla, Calif., Silver has led the system's development since its inception in 2003.
The system has greatly improved FDA's screening process, said Scott Witt, director of PREDICT's Open-Source Intelligence Team, referred to as the POSIT team, which uses intelligence analysis methods to help the FDA identify risky products.
"We have certainly doubled FDA's success rate in terms of finding the bad stuff," Witt said. "And, we've vastly improved 'the may-pass rate' for products we know are safe and we are not going to hold up. Those products can now get into the U.S. much more quickly."
With the number of individual imports doubling every few years from 9.4 million in 2003 to 18.5 million in 2009 expediting the import process is crucial for all concerned.
"While enabling FDA inspectors to focus on shipments posing the highest risk, PREDICT's risk-scoring system also allows the FDA to expedite the entry of safe products, which means that large quantities of goods will no longer linger on docks and at border crossings waiting for clearance," Silver said.
PREDICT creates risk scores based on two broad categories of information: historical data that FDA collects on importers and their products, and a new, dynamic set of intelligence data compiled by the POSIT team. The POSIT component makes the PREDICT system immediately responsive to international economic, political and environmental events that might affect the safety of FDA-regulated goods produced for import to the U.S.
"What we do in simple terms is look for trouble," Witt said. "We look into the future instead of looking at past data for analysis, and we look for things that are likely to introduce risk to any imported product that the FDA has cognizance over.
What POSIT brings to the problem is a human dimension to what would otherwise be primarily a computer-based risk assessment."
The POSIT team is a small group of dedicated, multilingual sleuths working in a windowless room in Gerald Thomas Hall. "Our staff of 14 includes 12 intelligence analysts trained in a variety of collection and analysis methods," Witt said. "They search daily through foreign and domestic resources looking for information that raises red flags about import companies and products. The information is then added to the PREDICT risk-scoring system to be used for screening when these imports arrive at U.S. ports and are offered for import.
"We comb through newspapers, websites, scientific journals, books, reports, trade data and other open-source materials in English and several other languages and also contact subject-matter experts via telephone and e-mail," Witt said.
"The FDA gives the team direction about what to focus on. For example seafood is one of the riskiest imports, so we focus on anything that might affect its quality. A typhoon hitting Vietnam might damage the fishing infrastructure on shore and wipe out refrigeration facilities. We would flag products coming from those areas, their overall PREDICT scores would go up, and FDA inspectors would focus screening and inspections on those products," he said.
The PREDICT computer system even incorporates machine-learning capabilities, enabling the program to "learn" from new data and self-adjust in certain cases, Witt said.
NMSU's PREDICT program has received more than $3.5 million from the FDA, which just provided additional funding to support it through September, when Silver expects to receive more funding for the 2011-12 fiscal year.
Margaret A. Hamburg, M.D., Commissioner of Food and Drugs, formally announced the launching of PREDICT in February.
PREDICT "considers everything from whether a product is intrinsically risky raw seafood falls into this category to information we've acquired from previous examinations of shippers or producers," Hamburg said. "We can even add information on things like floods, hot weather or market conditions that suggest whether a particular shipment is at risk of being spoiled or shoddy."
The open house will be from 3:30 to 5 p.m. in Gerald Thomas Hall, Room 297, and in the POSIT lab, Room 288. For more information, contact Silver at (575) 646-7038 or firstname.lastname@example.org.
To watch an FDA video describing the PREDICT system, visit http://www.fda.gov/NewsEvents/PublicHealthFocus/ucm199940.htm.