Ahead of the curve | National Center for Supercomputing Applications at the University of Illinois
Ahead of the curve
09.17.10 - Permalink
by Vince Dixon
Typically, local clinics send reports of illnesses and symptoms to state and national organizations like the Centers for Disease Control. With this method, "the local people who are collecting the data often don't see the results until it is too late to take any action," says Ian Brooks, creator of INDICATOR and director of the Health Sciences Group at NCSA. INDICATOR is an open-source biosurveillance program that analyzes data of local reported illnesses and school absences to detect unusual numbers.
"Although there are existing systems that collect information on infectious diseases, they tend to be top down," Brooks says.
That is why INDICATOR is driven by the local community. It operates from the local-out rather than the standard regional-down approach. This ensures a quicker response time, Brooks says.
It works by collecting data from area schools, hospitals, and clinics. When unusual patterns are detected, the system alerts monitors. This method stresses surveillance to better understand how pandemic diseases can spread and be controlled.
"What we want to do with INDICATOR, because we have computational capability behind it, is be able to link up the surveillance with modeling," Brooks says.
When surveillance and modeling are combined, responses can be planned using actual data as a guide rather than using models alone.
Wendy Edwards is a research programmer at NCSA and member of the Health Science Group. She and Brooks wanted to design INDICATOR so that communities could not only detect disease outbreaks, but also develop possible strategies to address them. Linking surveillance with modeling "would allow practitioners and healthcare providers to decide the best approach at dealing with a disease outbreak," she says.
INDICATOR combines both traditional and syndromic surveillance and uses what its creators call a "cyberenvironment," a system that allows users access to a set of tools that are designed around a specific discipline. The goal is to help researchers design their own workflow and specify the data and algorithms they prefer, Edwards says.
INDICATOR also analyzes data from effectively any kind of source. While other systems require preconceived formats that users input, Brooks and Edwards have written scripts that allow INDICATOR to automatically parse any kind of data format and retrieve the information the team wants. They hope this will encourage more organizations to use the program.
"We want to make it easy for them to work with us," Edwards says.
When local schools, hospitals, and the public health district send data, the analysis algorithm WSARE checks for unusual events. If it detects unusual and meaningful patterns, it sends an alert to the INDICATOR team members and to the public health district. The health district then plans accordingly.
When the H1N1 virus reached Champaign, INDICATOR helped the public health district to better keep track of the number of students absent from schools. With a shortage of vaccinations, the system alerts helped the organization plan for the prioritized distribution of the vaccines. Seeing the number of student absences, the agency revised its distribution strategy to address the apparent increase in outbreak among school children, says Awais Vaid, the epidemiologist for the Champaign-Urbana Public Health District.
"Our goal is that by the time all the data sources start coming in and are analyzed well, it will buy us at least a few days of time before the actual outbreak hits the community," Vaid says. "This system will give us some time to alert community providers to prepare ourselves for something abnormal so that we can plan and respond better."
Once INDICATOR starts to receive heavier amounts of data, the group members hope to run the system on supercomputers.
"When we start getting enough data to justify it, we definitely want to add high-performance computing capabilities to this," Edwards says.
Other plans for IDICATOR include making the system nationally available, Brooks says. And given the close relationship between the spread of human diseases and outbreaks in animal disease, someday INDICATOR will be able to track data from local veterinarians and the University of Illinois College of Veterinarian Medicine, Brooks says. The system's data could also be used for research.
"One of the things you realize quickly when you start working, particularly in the public health area, is that the data you are collecting to make operational decisions is the same data you want to use for research," Brooks says. "It doesn't have to be collected twice."