02/20/2020 - 8:00
- Clarín.com
- World
On the shores of Lake Ontario, a Canadian startup launched one of the first alerts about the risks of an unknown virus that appeared in the Chinese city of Wuhan. As it did? With artificial intelligence
BlueDot developed an algorithm that rummages through thousands of news and air traffic information to detect and monitor the spread of infectious diseases.
The company alerted its customers about the outbreak of a new coronavirus on December 31; few days before the authorities officially did it.
BlueDot, which is based in Toronto, also correctly predicted which countries were most exposed to contagion.
"What we are trying to do is really extend the limits, using data, analysis and technology to move quickly," said Kamran Khan , founder of the company.
"When you deal with an outbreak, timing and timing are everything," he said.
Dr. Kamran Khan, founder and CEO of BlueDot, in Toronto. / AFP
Khan, a 49-year-old doctor, had the idea of launching BlueDot after the SARS epidemic in 2002-03.
At that time, Khan specialized in infectious diseases in a Toronto hospital. I saw helpless as that disease left 44 dead in the largest city in Canada.
"Many health workers were infected, including one of my colleagues. We had several health workers dead," he recalled.
"That experience opened our eyes and was the motivation for everything we are doing in BlueDot," he added.
In 2015 Khan launched BlueDot, which now has 40 employees that include a team of doctors, veterinarians, epidemiologists, data analysts and program developers .
Kamran Khan./ AFP
Together they designed a real-time alert system based on natural language processing and machine learning techniques.
Every 15 minutes the BlueDot algorithm reviews official reports, professional forums and news services for keywords and phrases.
The algorithm can read texts in 65 languages and track 150 types of diseases.
"We call this needles in the haystack," says Khan.
Kamran Khan./ AFP
"There is an overwhelming amount of data. The machine looks for needles and presents them to human experts." They review them and tell the machine if that information corresponds to a real threat.
If credible, the information goes to a database that analyzes the place of the outbreak, nearby airports and itineraries of commercial airplanes around the world.
Climate data , databases of national health systems and even the presence of mosquitoes or any type of animal that transmits diseases are also taken into account.
Once the analysis is completed, BlueDot sends alerts to its customers ; which include government agencies, airlines and hospitals.
The goal is that the authorities can prepare for the worst: a major outbreak of a disease.
Early in December 31, the BlueDot system found an article in the Mandarin language that mentioned 27 people with pneumonia and all of them related to a market in Wuhan City.
The virus was not yet identified but the algorithm found two key expressions: "pneumonia" and "unknown cause."
At 10 in the morning of that day, the first alert was sent to customers; especially to those in Asia.
"While we didn't know it would be a big global outbreak, we noticed some ingredients similar to what we saw during the SARS epidemic ," Khan explained.
BlueDot was able to predict that there was a certain risk that the virus would spread from Wuham to Bangkok, Taipei, Singapore, Tokyo and Hong Kong.
Cases of the new coronavirus that have already killed 2,000 people , almost all in China, have been reported since then.
BlueDot has another milestone: in 2016 it predicted the spread of zica from Brazil to southern Florida, the United States.
AFP