Early Detection of Bioterrorist Threat or Epidemic Is Goal of Software System for Medical Emergencies

BUFFALO, N.Y. -- Computer scientists at the University at Buffalo who developed handwriting recognition software systems for the U.S. Postal Service and the U.S. Census Bureau are developing a system to flag suspicious patterns in emergency medical reports and make them available to public-health authorities within days, if not hours.

"By automating the collection of data on all patients who enter the emergency medical system, patterns of public-health emergencies that might be related to a terrorist attack or an epidemic would be revealed right away," said Venu Govindaraju, Ph.D., UB professor of computer science and engineering and associate director of UB's Center of Excellence for Document Analysis and Recognition (CEDAR.)

Features of the system, Govindaraju said, would have application to other sectors where prompt data processing and analysis also are critical, such as in the processing of application forms filled out for the U.S. Immigration and Naturalization Service by travelers at ports-of-entry.

Govindaraju, principal investigator on the project, said valuable information could be gleaned from an automated analysis of patient data. Whether it's West Nile virus or a bioterrorism threat, he stressed that the sooner medical personnel receive information about patterns of medical emergencies, the more likely they are to respond effectively. While time is the critical factor, it can be months before data are keyed in manually at a regional processing center and years before completed analyses are available.

When the first few cases of anthrax surfaced in the Washington, D.C., area last fall, some postal workers may not have received the proper diagnoses because health-care workers had not yet been notified to be on the alert for anthrax cases, particularly among postal workers. The consequences of those delays may have been tragic.

"If an automated analysis of pre-hospital care reports shows that many patients from the

same geographic area are reporting the same symptoms in a short period of time, this critical information -- which may not be obvious to the ER staff, given the volume of patients that go through the system each day -- could be disseminated easily and quickly to the appropriate authorities and to health-care workers," said Govindaraju.

"The sooner public health officials know that a pattern is emerging, the sooner they can act to contain it."

Automated tools that gather and analyze patient data also allow policymakers to track such cases once treatment has been administered, improving the accuracy of post-crisis assessments.

"The database itself would then become a valuable resource for enabling data mining and knowledge discovery for the entire medical community," said Govindaraju.

The system he envisions features a software program that can turn into digital data the handwriting, check marks and circled responses used to describe a patient's presenting problem, vital signs and other symptoms. The UB researchers will use a lexicon, or vocabulary, of medical terms and keywords that they will construct using sources from the National Institutes of Health.

The program will feature data-mining tools designed to automatically analyze data on the form once it has been processed.

UB researchers at CEDAR are working on the project in consultation with New York State Department of Health officials and with Western Regional Emergency Medical Systems, Inc. of Buffalo.

CEDAR is the largest research center in the world devoted to developing new technologies that can recognize and read handwriting. Over the past decade, CEDAR has worked with the U.S. Postal Service developing and refining the software now in use in postal distribution centers across the nation that allow up to 70 percent of the handwritten addresses on envelopes to be read by sorting machines.