In the first days of 2020, before most of the world had heard of the new coronavirus, the Toronto company BlueDot was already modeling where it would go. BlueDot, founded by infectious-disease physician Kamran Khan, runs software that scans news reports, official bulletins and animal-disease alerts in dozens of languages to flag emerging outbreaks, then layers global airline-ticketing data on top to estimate where infected travelers are likely to land. The company has said it alerted its clients to a cluster of unusual pneumonia in Wuhan on December 31, 2019, ahead of the World Health Organization’s public statement.
The analytical core of that forecast was published on January 14, 2020 in the Journal of Travel Medicine, in a paper by Isaac Bogoch, Alexander Watts, Andrea Thomas-Bachli, Carmen Huber, Moritz Kraemer and Khan titled “Pneumonia of unknown aetiology in Wuhan, China: potential for international spread via commercial air travel.” Using 2018 International Air Transport Association data on passenger volumes out of Wuhan’s airport, the authors quantified which cities were most connected to Wuhan and therefore most exposed. The model pointed to Bangkok, Hong Kong, Tokyo and Taipei as the highest-volume destinations - a list that closely matched where the first exported cases actually appeared.
The episode became one of the most-cited examples of AI for public health, though BlueDot itself has always been careful to describe the system as a human-in-the-loop tool: machine learning surfaces candidate signals, and trained epidemiologists decide whether to send an alert. The forecasting power came less from any single clever algorithm than from fusing unstructured outbreak chatter with structured mobility data.
Why business readers should care: BlueDot is a clean example of value created not by a novel model but by joining two ordinary data sources - news text and airline tickets - and putting expert judgment in the loop, a pattern that travels far beyond epidemiology.