If a Chinese newspaper publishes a small, seemingly insignificant story on a strange illness, Canada’s health authorities will find out about it immediately via its Global Public Health Intelligence Network (GPHIN).
The network, which was recently updated to version two, is now able
to monitor global media sources such as online newspapers and Web sites in six languages — English, French, Spanish, Arabic and Russian, as well as simplified and traditional Chinese — for information on natural disasters, disease outbreaks, infectious diseases, contaminated food and water, bio-terrorism and exposure to chemical-nuclear agents.
“”There are a few similar products but no one is providing the service we are right now and that is the ability to scan as many sources as we do a day and translate them into the different languages,”” said Peter Uhthoff, chief, counter-terrorism co-ordination and health information network. “”You could ask why people don’t use such things as Google news, but the problem (with such offerings) is they only search the front page of these newspapers.
“”We have found the valuable articles aren’t on the front page, they’re on the inside, where there’s only a small column about this incident that happened at a hospital, or a chemical plant or wherever.””
As well, added Uhthoff, Google searches at the national newspaper level only. “”We have found not only are the interesting articles within the paper but they’re in the local small papers.””
Current subscribers to the system, which is maintained by the Public Health Agency of Canada, include the World Health Organization and the U.S. Center for Disease Control.
According to Uththoff, GPHIN II is now not only multilingual but much more robust than it used to be.
“”(GPHIN I) did work and we were using it, which perhaps as a prototype should not have been. GPHIN II is a lot more robust and … during a crisis we can easily ramp up. The prototype would crash about three to four times a day because it wasn’t meant to handle the capacity we had.””
The system works by constantly searching for words, combinations and arrangements of words using a complex algorithm.
“”We also search for numbers but it depends on the disease, because with certain diseases one is sufficient for an alert, while others require more,”” he said.
GPHIN II was developed with financial assistance from the Nuclear Threat Initiative, a Washington-based foundation whose goal is to keep the world safe from nuclear and other threats.
“”They thought supporting our program, which is an early warning surveillance system, would allow the public health authorities early warning of potential catastrophes, and the sooner the warning the better the response,”” explained Uhthoff.
Laurent Proulx, senior vice-president and chief technology officer at Nstein, said the company’s technology natively analyzes unstructured information such as e-mails or other documents for 12 potential threats.
“”We receive news and from that we will detect potential threats and we will alert a gatekeeper according to a certain threshold or automatically send e-mail around the globe to people responsible for handling that kind of threat,”” explained Proulx.
If gatekeepers, as health authorities are called, want to know more about an article in another language, they can activate the translation on the fly, he said.
Nstein beat out other contenders for the contract due to its linguistic capabilities.
“”Translation is not really an issue for us because we analyse natively, so if we find something in Chinese, because we analyse the text or the news feed natively, we’re not using translation to analyse,”” said Proulx.
Although version 2 of GPHIN has just been rolled out, Uhthoff is already planning ahead for version 3.
“”One of the driving triggers for the GPHIN I prototype was the ability to get information from around the world, so we thought we’ll go to newspapers, which has been very successful,”” said Uhthoff.
“”However, there are many parts of the world where newspapers don’t exist, and it’s still the AM radio that is the means of communication, so we’re thinking about being able to tap into that, not in the spy sense but in the electronic sense, and have software that converts speech to text.””