A new early warning system developed at Canada’s National Research Council (NRC) could help public health officials better detect and respond to outbreaks of diseases and illnesses.
The technology was successfully employed in Ottawa during the fall 2009 H1N1 pandemic to track influenza-like illness (ILI) in patients presenting at the emergency departments of three of the city's hospitals (The Ottawa Hospital, the Children's Hospital of Eastern Ontario, or CHEO, and Queensway Carleton Hospital).
Called ILI Watch, the software program extracted data from the Ottawa hospitals on the chief complaints received from patients. A component known as an ILI-specific classifier was then used to identify symptoms associated with the flu based on key words or phrases, such as "fever," "high temperature" or "severe headache."
Using this data, ILI Watch produced daily reports for Ottawa Public Health officials on the number of ILI cases reported at hospitals as well as the age of the patients. The city's public health authority could then use the data to spot any spikes in the incidence of ILI, and determine which age group was affected. In fact, during the 2009 pandemic, Ottawa Public Health identified several spikes at Ottawa hospitals, particularly at CHEO.
Faster and more accurate than a human
To create the classifier, the NRC Institute for Information Technology (NRC-IIT) collected about 150,000 chief complaints that were received in Ottawa hospital emergency rooms between 2007 and 2008, and that medical professionals classified as ILI or non-ILI symptoms.
"We used those statistics to train a computer to mimic what a human expert would do in identifying symptoms," explains Dr. Joel Martin, who leads NRC's Interactive Information group, which designed ILI Watch.
"Instead of having public health officials spend hours going through that data, the classifier did it in minutes."
The unique automated classification process was "invaluable" during the 2009 H1N1 pandemic, says Dr. Amira Ali, a senior epidemiologist with Ottawa Public Health.
"It didn't just save time, which in itself was very significant. It also improved accuracy by eliminating the possibility of human error associated with having someone manually review the daily intake of hospital cases."
She explains that during the H1N1 pandemic, the ILI-classifier was instrumental in giving Ottawa's public health officials an "overall picture of the daily burden of ILI in the hospital ERs," and to help detect flu outbreaks a lot sooner.
Adaptable for future viruses
Ottawa Public Health will continue using ILI Watch, which has also been deployed in the Grey Bruce Health Unit, northwest of Toronto, as well as in Kingston, Frontenac and Lennox & Addington Public Health in eastern Ontario. The ILI classifier has been modified to provide flu updates every six hours.
In fact, ILI Watch can be adapted to add new symptoms and satisfy any requirements of public health authorities - such as the frequency of reporting - for future strains of flu viruses, an aspect not featured in any other automated syndromic surveillance system, according to Jason Morin, a project officer with NRC-IIT's Interactive Information group, who adds that ILI Watch is also unique in its accuracy of replicating human judgment of ILI syndromes.
The ILI-related tracking system is part of the Advanced Syndromic Surveillance and Emergency Triage (ASSET) project, led by the University of Ottawa Heart Institute and involving several federal government, industry and health-related partners, including NRC, Ottawa Public Health and Ottawa-based software company, AMITA Corporation.
Launched in Ottawa in 2007 and funded by Canada's Chemical, Biological, Radiological-Nuclear, and Explosives Research and Technology Initiative (CRTI), ASSET is also an automated system that monitors data streams, such as health records, to identify medical symptoms associated with any illness or disease that could result from a terrorist attack.
It is also nimble enough to track syndromes related to other occurrences, such as the bacterial contamination of a water supply from E. coli.