York U profs develop COVID-19 prediction tool for Canada

A new analytics dashboard from York University’s Schulich School of Business predicts that Ontario will have 5,346 confirmed COVID-19 cases by Sunday, appearing as though the virus is accelerating its rate of spread in the coming days.

“Nobody has a tool like this,” Dr. Murat Kristal, director of Schulich’s master of management in artificial intelligence and master of business analytics programs tells NOW. He developed the COVID-19 Dynamics dashboard alongside Dr. Ikjyot Singh Kohli, senior data scientist and mathematical physicist, and David Elsner, a visual analytics and modelling instructor.

Their project draws on data from Johns Hopkins University’s global COVID-19 map, which publishes numbers of confirmed cases, and GitHub, while using machine learning algorithms, database management programming language SQL and cloud computing service Azure to predict how quickly the virus will spread in every country and region.

The Schulich COVID-19 dashboard predicts the next day’s confirmed cases with 92 per cent accuracy. Though the accuracy diminishes as the platform continues to forecast over the following four days.

As of March 30, there were 1,966 known cases of COVID-19 in Ontario, according to provincial public health officials.

“The idea is not that we are bang on but the order of magnitude is there,” Kristal explains. The tool is meant to assist policymakers, decision makers and the medical field to anticipate the spread and how many intensive care units units will be occupied in the immediate future.

The purpose is similar to the Flatten.ca tool developed by University Of Toronto students. That platform maps COVID-19 spread based on data input by users, using answers regarding symptoms and travel history to predict how many probable COVID-19 cases an area has.

“Our tool is based upon published numbers,” says Kristal. “We are making predictions based on what the government is announcing. This is more for policy and decision makers assuming the government are publishing the correct numbers.”

According to the COVID-19 Database, Canada will have 10,045 confirmed COVID-19 cases by April 5. The tool can also zero in and make predictions for cities as soon as governments start reporting locally.

“As long as those cities start releasing their numbers and we can pull them from somewhere on a consistent basis, we can even start predicting for Toronto.”

**This story has been updated with new dashboard figures made available on April 1. A previous version of this story indicated that the COVID-19 spread in Ontario is predicted to slow down and confirmed cases on the weekend would be under 3000. 


Stay In The Know with Now Toronto

Be the first to know about new and exclusive content