The North Carolina Bio-Preparedness Collaborative, a partnership between the University of North Carolina, Chapel Hill; North Carolina State University (NCSU) and software analytics company SAS, will analyze data from various organizations in an attempt to better understand past medical emergency patterns, which will in turn help predict and control future outbreaks.
For example, the Emergency Medical Services Corp. receives 1.5 million ambulance calls every year and holds records of those calls from many years past, said David Potenziani, the collaborative’s executive director. “But they don’t necessarily understand patterns,” he said, “and they certainly don’t understand how they can detect anomalies, which we believe are the pathway to being able to detect emergencies and potential threats to human health.”
Potenziani and his “dream team,” as he calls them, of public health and health-care experts have begun to gather that data, along with other files from emergency room visits, hospitalizations, tainted food reports and veterinary records. This amalgam of data will help the team create thresholds that will help the team differentiate between normal health patterns and environmental changes, versus natural or man-made health threats.
“What we’re doing is trying to understand phenomena for data that’s collected from other purposes,” he said, and then use that data for detecting health hazards.
For example, his team is currently trying to get access to school attendance records throughout the state in order to recognize emerging diseases hours before they appear in the reported epidemiological systems — when parents take their children to the doctor. Before it’s officially reported, “that illness is invisible to the health-care and public system,” Potenziani said, which leaves adequate time for the illness to spread.
The team will also have access to poison control data very soon, he added.
The idea for the system sprouted in 2007 when a group of University of North Carolina faculty was talking about coughing, Potenziani said. Questions started to form about how to detect threats like avian flu that originate in the natural world and in nonhuman species. These disease vectors are hard to detect and therefore can spread to thousands of humans without warning.
When discussing how a system could track patterns in individual cases — efforts have tried and failed at the federal and national levels, said Potenziani — the team focused on ways to avoid the pitfalls of previous attempts.
One way is gathering data from the original file location, where it’s the most accurate. That’s preferable to transferring the data to a center where the team would then hold and analyze it.
“The goal is to conduct analytics in real time, but the challenge is getting access to the data in a timely fashion,” he said. Current technology allows the team to scale computational resources of various sizes, and the team is planning on speeding up that process as local data collection advances.
Another unique aspect, Potenziani said, is using technology that was created for another function and repurposing it for this project. For example, the preparedness system runs on a cloud computing technology created by the NCSU and IBM, called the Virtual Computing Lab, which was initially developed for supporting education by providing a configurable platform for instruction.
By adopting it to serve for biosurveillance, it brings down the cost of the project to a fraction of what has been spent on previous similar federal projects, Potenziani said. To date, the project has received $5 million from the U.S. Department of Homeland Security.
Currently the system is operational on a limited basis; new data sources are being added. Potenziani said he hopes to complete the system by this summer. Eventually he’d like to expand the program nationally.
