From the last day of 2019, a health monitoring company in Canada has sent a warning about dangerous epidemics to its customers.
On January 9, the World Health Organization (WHO) received information about an influenza-like disease in China that caused many cases of pneumonia. CDC US Centers for Disease Control and Prevention received information 3 days earlier. However, both are slower than a Canadian health monitoring platform, when the company sent a warning to customers from December 31.
BlueDot, the platform uses AI algorithms to scan a series of articles and networks warning of natural disasters and epidemics as well as official information from foreign governments to alert its customers to avoid Hot spots like Wuhan.
Speed is the most important
Disease prevention agencies such as WHO or CDC often have to wait for official medical reports to confirm the epidemic and provide countermeasures. Meanwhile, the reaction rate is critical for infections as fast as Covid-19, and the Chinese government did not provide timely information when the outbreak began.
"We know that governments may not be able to provide timely information. We can quickly update information about disease risks, from sources such as forum chat and blogs to find out. signs of unusual events, "said Kamran Khan, founder and CEO of BlueDot.
Khan says their algorithms don't use social media posts because they're so messy. One of the best data sources is airline tickets from global airlines. This data source can show the movement trend of infected people. According to Wired, BlueDot's artificial intelligence accurately predicted the rush of the Wuhan people to Bangkok, Seoul, Taipei and Tokyo during the outbreak.
"I felt like I had this problem. In 2003, I witnessed the SARS virus that caused the city to be in chaos and the hospital overwhelmed. At that time, I was really physically and mentally tired, and I thought I could not let This happened again, "Khan, an infectious disease specialist who worked in a hospital in Toronto, told about the SARS in 2003.
"My lesson from the SARS epidemic is that I cannot let myself be surprised, it is necessary to predict the disease," he added.
Bluedot was born in 2014, so far there are 40 employees including programmers and doctors. The tool is capable of processing natural language in 65 languages, allowing aggregation of information from newsletters, airlines and animal disease reports.
"We have to use natural language processing systems and machine learning to help computers recognize between the risk of anthrax outbreaks and an Anthrax band reunion," Khan explained.
Combining AI and humans
After the computer filters the results it is the human part. Infectious disease specialists will review the conclusions from the computer to see if it is suitable for scientific knowledge. The report will then be sent to government epidemic prevention agencies, private hospitals and airlines, which may have to deal with the epidemic.
Khan said he wanted BlueDot to be able to produce reports faster and more accurately than Google Flu Trends. Google's flu forecasting tool has been shut down since 2013, after failing to predict the dangers of pandemic flu that year. According to the medical journal The Lancet, BlueDot accurately predicted the outbreak of Zika disease in southern Florida. The software also predicts Ebola will reach beyond West Africa.
Khan acknowledged the similarities between SARS and Covid-19, which made him a little annoyed. "Are we going through the same period as 17 years ago," Khan told CNBC.
After Covid-19, BlueDot will continue to predict new diseases. Currently the Lassa virus outbreak is returning to Africa. Although the initial condition is not so serious that a pandemic, such diseases also need to receive due attention.
"When people are distracted, we have the machines to help us monitor and capture everything that's happening," said CEO of BlueDot.
