The first warning about the Wuhan outbreak did not come from the US Centers for Disease Control and Prevention or the World Health Organization. It came from a Canadian company’s software.
BlueDot, an AI-powered health monitoring platform, sent its users an alert about the outbreak on December 31. The CDC reported the news on January 6. The WHO followed on January 9.
How it works
BlueDot is run by a team of physicians and programmers. Its surveillance program scours news in 65 languages and combs airline data, looking for trouble spots and telling users which areas, Wuhan among them, to avoid.
The platform’s founder and chief executive, Kamran Khan, described the process to Wired. Machine learning sorts through the data for trends and parallels. Then epidemiologists check that the conclusions and predictions hold up before any alert goes to users, government officials or businesses.
Language processing is the tricky part. Khan said it trains the engine to tell “whether this is an outbreak of anthrax in Mongolia or a reunion of the heavy metal band Anthrax.”
He also explained why the company looks beyond official channels. “We know that governments may not be relied upon to provide information in a timely fashion,” he said, adding that the system “can pick up news, little murmurs or forums or blogs of indications of unusual events.”
BlueDot’s website sums up what it sells: “We help governments protect their citizens, hospitals protect their staff and patients and businesses protect their employees and customers.”
Not its first forecast
The company launched in 2014. It had called an outbreak ahead of time before: it forecast a Zika virus outbreak in Florida and published the results in The Lancet.
The warning arrived as Chinese officials were working out how to stop the spread. They had put Wuhan under quarantine, which affected the movement of more than 56 million people.
Sources
Sources
- A Canadian robot sent coronavirus warnings days before officials
- Coronavirus Threat Was Detected By Artificial Intelligence …





