How Sask.'s COVID-19 modellers stay a step ahead of the virus

"People mistake models as crystal balls. They’re actually tools for more quickly pointing out our mistaken understanding."

Zak Vescera, Saskatoon StarPhoenix 3 minute read November 15, 2021

Saskatchewan’s COVID-19 modelling team has helped jurisdictions across the country and the world prepare for and prevent the worst of COVID-19, giving policy-makers prescient insight into how the unseen foe might strike.

Models produced by Dr. Nathaniel Osgood’s team at the University of Saskatchewan are used in every province in the country, at the federal level and even in Australia, to map how the virus might spread given population behaviour, vaccination rates, public health restrictions and a slew of other variables and parameters projecting the pandemic’s next phase.

Osgood said he’s honed this art over more than 30 years of mapping more than a dozen diseases, but it’s not well understood by the general public.

“People mistake models as crystal balls,” Osgood said in a recent interview. “They’re actually tools for more quickly pointing out our mistaken understanding, cluing us in to more reliable understanding, and for sharing and more effectively making sense of our collective knowledge.”

Osgood began work on Saskatchewan’s COVID-19 models in February 2020, before he was seconded to assist the province’s health authority the following month as the pandemic reached his home province.

Since then, the computer simulations have been updated plenty. Vaccines made the virus easier to stop, new variants made it harder, and new sources of information like waste water epidemiology helped produce more reliable results. But the core of the model has remained much the same and produced results that are markedly accurate, Osgood said.

By adjusting parameters, modellers are also able to project what may occur if certain steps are taken — if people reduce their social mingling, for example, or a government imposes measures like masking or gathering restrictions.

“Essentially, models get better over time as your information gets better over time,” U of S epidemiologist Dr. Cordell Neudorf said. “So you start with what you know about how the virus spreads … but as new studies come out, you get to see in real time what happens.”

Models produced in June, for example, almost perfectly charted Saskatchewan’s devastating fourth wave, fuelled by the extra-infectious Delta variant.

“As the summer went on, the scientific evidence on Delta became clearer and clearer in ways that were more worrisome and not less,” Osgood said. Other modellers also noted and warned their partners of what the spread could mean. At the time, Saskatchewan was preparing to drop its last public health measures in July. The province would eventually reintroduce measures in September after intensive care unit hospitalizations shot to unprecedented levels.

“Many jurisdictions across the country took action that was reflective of this … it is a matter of public record that our province did not,” Osgood said.

His favourite models are the ones whose projections never come true, precisely because their very creation prompts action, he said.

“In my view, the most successful models are ones that invalidate their predictions because they motivate action in response that will lessen this burden. It’s an irony — and I can say this from presenting to clients other than the (health) minister and the health authority — that what we anticipated early in June held eerily prescient, almost to the point of prophecy.”

The news seems to be flying at us faster all the time. From COVID-19 updates to politics and crime and everything in between, it can be hard to keep up. With that in mind, the Saskatoon StarPhoenix has created an Afternoon Headlines newsletter that can be delivered daily to your inbox to help make sure you are up to date with the most vital news of the day. Click here to subscribe.