       |
Media |
|
Press Releases |
|
Potomac Institute Leads Pioneering Biosurveillance Evaluation October 18, 2004 News Release Contact: Arlington, VA -- The Potomac Institute was selected by the Department of Homeland Security to perform the independent evaluation role for the Bio-ALIRT (Bio-event Advanced Leading Indicator Recognition Technology). This is one of the first large quantitative evaluations of medical surveillance systems, whose usage is increasing due to concerns about the need for early detection of bioterrorism. The Potomac evaluation showed that advanced biosurveillance systems could detect natural outbreaks of influenza-like and gastrointestinal illness often on the same day they started at reasonable false alarm rates (1 every two to four weeks). Human experts, however, were more likely to detect the outbreaks days later, over ten days later for seasonal respiratory outbreaks. The leading detection systems were supplied by the Johns Hopkins University Applied Physics laboratory (“ESSENCE”), and the University of Pittsburgh/Carnegie Mellon University (Realtime Outbreak Detection System [RODS]). “The biosurveillance systems did very well, well enough to justify deployment in metropolitan areas. In addition to providing early warning of a bioterrorist incident, they will almost surely assist Public Health in managing naturally-occurring outbreaks that may kill over 35,000 Americans per year,” said David Siegrist, Potomac Senior Fellow and Principal Investigator for Evaluation of the Bio-ALIRT program. “This is especially important with the current lack of flu vaccine.” Influenza and GI outbreaks were chosen because their early symptoms mimic those of several Class A pathogens that the CDC lists as risks for terrorism. Mr. Siegrist presented his results at a Syndromic Surveillance conference at the New York Academy of Medicine last fall. His article (along with co-author Dr. Julie Pavlin of the Walter Reed Army Institute of Research) describing the evaluation was chosen for a special edition of the CDC’s referred publication Morbidity and Mortality Weekly Report Special Supplement. You can view the report at http://www.cdc.gov/mmwr/preview/mmwrhtml/su5301a29.htm. Mr. Siegrist was also a peer reviewer of the special edition, and has served on CDC’s “Expert Panel” that helped establish CDC’s evaluation framework for early detection surveillance systems. In summary, the evaluation considered 5 metropolitan areas for which de-identified medical encounter records could be obtained for both military and civilian populations covering a period of over two years. Epidemiologists examined the records for signs of outbreaks that they would consider significant, including estimating when local Public Health might be expected to make the “call” that an outbreak was indeed taking place. The unlabeled data was released to the investigators for analysis. Participants included the Bio-ALIRT teams, as well as a CDC team and one from a large metropolitan Department of Health. The detection algorithms generated a probability (“p value”) of outbreak for each day’s data, which was later run against different false alarm rates to determine relative performance. The best algorithms were able to detect all the outbreaks, often on the first day that the human experts had determined retrospectively that they had begun. This was typically well before the human experts would have detected them prospectively, that is, as the daily data were received, the way the systems had done.
The Potomac Institute for Policy Studies is an independent, 501(c)(3), not-for-profit public policy research institute, and is dedicated to the development and implementation of policies that advocate and manage the increasing role of science and technology in our evolving world.
|
|
| © Potomac Institute for Policy Studies 2004 • Privacy Statement • Email comments to Webmaster • |