You are here

Mathematical modeling in the fight against COVID-19

Tuesday, November 30, 2021

Q&A with Dr. Caroline Colijn, Canada 150 Research Chair in Mathematics for Evolution, Infection and Public Health at Simon Fraser University and Dr. Sarah Otto, Professor of Zoology at the University of British Columbia.

Dr. Sarah Otto and Dr. Caroline Colijn 

In late December 2020, when the first SARS-CoV-2 Variants of Concern (VOC) were first identified, the Canadian COVID-19 Genomics Network (CanCOGeN) convened a group of mathematical modelling experts to work with public health groups doing genomic sequencing. This modelling group is working with CanCOGeN’s Data Sharing Committee to inform how the data generated through CanCOGeN can be used to model various outcomes in the COVID-19 pandemic—providing vital information to researchers and public health decisionmakers. 

Two vital members of this group are Dr. Caroline Colijn and Dr. Sarah Otto, mathematical modellers with expertise in epidemiology and the evolution of pathogens. In the COVID-19 context, they have been modeling the evolution of SARS-CoV-2 variants, including VOC and variants of interest (VOI). Dr. Colijn is a Canada 150 Research Chair in Mathematics for Evolution, Infection and Public Health at Simon Fraser University. Dr. Otto is a Killam University Professor in evolutionary biology at the University of British Columbia. 

We asked them about the role of mathematical modelling and genomics in the fight against COVID-19.


“Math helps us understand the implications of our logic and our thinking, and it reveals when our intuition is wrong.” - Dr. Sarah Otto

“With the latest variant of concern, Omicron, models can help determine the likely impact of the variant on the pandemic in Canada and other countries.” - Dr. Caroline Colijn


How does mathematical modelling help fight pandemics?

Colijn: Modelling helps us to synthesize the information we have. This information might include what we know about how long an infection lasts, the average number of people one person is likely to infect and what interventions are currently available (e.g., vaccinations). Mathematical modelling is a way of putting all of that information together into a representation we can adjust to see what happens and explore the implications of our assumptions.

Otto: Mathematical models help us project what might happen with the development of the pandemic. Pandemics are classic examples where there are so many factors in play—mutations and changes in behaviour for example—that might impact their progression. Math helps us understand the implications of our logic, and it reveals when our intuition is wrong.

How is mathematical modeling being used to tackle COVID-19?

Colijn: One of the uses that gets a lot of profile is case forecasting. What direction are we going? How high could the number of cases get in six weeks? Modelling is more like having headlights on than having a crystal ball; it helps us see what the path ahead looks like. We can’t predict people’s behavior, but modelling is used to understand what’s happening with cases, hospitalizations, ICU case counts and deaths. It’s used to estimate vaccine efficacy. And it’s also used to estimate the impact of interventions, like distancing and masks—to indicate the direction we are heading in. 

How is SARS-CoV-2 genomic data used to track VOC and VOI in Canada, and globally?

Otto: Not only are we tracking the spread of the disease from person-to-person using math, but we’re also tracking how subtypes of the virus are changing—in particular, variants of concern that increase the transmission rate or other major properties of the disease. The genomic data is needed to tell us if there is anything different about a version of the virus that one person has versus the virus that a different person has. And then that can be used in the models once we can detect differences in transmission rate.

Colijn: Those differences could be differences in transmission rate, which we saw with Alpha and with Delta. Those differences had a profound impact on the pandemic. If we hadn’t had those increases in transmissibility, we’d be in a different place. With the latest variant of concern, Omicron, models can help determine the likely impact of the variant on the pandemic in Canada and other countries.

When it comes to SARS-CoV-2 genomic data and accompanying contextual data, why is it important to increase data access and sharing?

Otto: It’s hugely important. We need to share data from multiple regions for two reasons. One is that we don’t have the power to detect trends if we have a small data set from one region. We have more power to detect changes if we have shared genomic data across provinces and countries. The other reason is that we need to see the same kind of patterns in multiple places to answer questions like: is a rise in cases due to the unique characteristics of a virus variant (like with Alpha or Delta) or is it just by chance (like a super spreader event at a concert).

Colijn: If you can pull more data from multiple places, you have more data sooner and can understand a situation sooner. It gives you that statistical power. This doesn’t necessarily mean making all data public. There are data governance structures that need to be used, and can be used, to protect individual privacy while still allowing us to rapidly understand what a new variant can do.

What has been the biggest challenge in developing informative models for COVID-19?

Colijn: One of the biggest challenges is not knowing how many infections we are detecting. Estimates vary hugely, with some estimates saying with that symptomatic testing we’re finding more than 100 per cent of the symptomatic infections, because we’re finding some asymptomatic ones too. Other estimates say we’re only finding a fifth to a third of the infections, and that makes a huge difference when we’re thinking about how much immunity there is in the population. Right now, that uncertainty is impacting how we understand the Omicron variant, because it impacts how we interpret Omicron’s rapid growth in South Africa. 

Otto: At the very beginning, we knew nothing and there was a lot of uncertainty. We didn’t know how the virus was transmitted. We didn’t know how long an infection lasted. We didn’t know how many new cases typically arose from each case. And we didn’t even know whether you could get sick from somebody without symptoms. From the very beginning, the models were used to estimate these unknowns and account for that uncertainty.

What are some common misconceptions about how modelling can be used to inform decision-making during a pandemic? Are people assuming predictive models are either right or wrong?

Otto: Models have been our best guide to what we would expect to see in the next few weeks and months to come. I am very happy when our models don’t work because they have sounded the alarm and led to changes in behavior that prevented a surge.

Colijn: Exactly. You wouldn’t say, “the headlights of my car predicted I would drive off a cliff, but I didn’t, so headlights are useless, and I won’t use them next time.”  I think another misconception is that mathematical models might be used to answer ethical and values questions. Models can help define the questions, and can help understand the potential impact of decisions, but they don’t answer ethical or policy questions in context.

How do you foresee the vaccination of children between ages five and 12 changing the way SARS‑CoV‑2 spreads in Canada?

Colijn: It’s another layer of protection for them as individuals, and it has indirect benefits at the population level because it stops transmission. It will help stop the virus from being able to get to the remaining unvaccinated people because of that layer of protection.

There has been a lot of discussion of shifting from a pandemic phase to an endemic phase. What does this mean for Canada and Canadians?

Otto: I think SARS-CoV-2 is here to stay. We’re not going to eliminate it from the human population. We also know there are animal reservoirs beyond bats. Deer studied in the United States have a very high incidence of it, so it’s not going to disappear. But, in the context of immunized individuals, I think that it’s a lot safer now that our immune systems are primed.

Colijn: We won’t get a kind of idealized herd immunity that leads to eradication. But I think we will get a very high level of immunity in our population that gives us a kind of practical herd immunity where we can largely reopen without threatening our healthcare systems.

Genomic sequencing has played a crucial role in the global response to COVID-19, enabling us to track the evolution of the SARS-CoV-2 virus with incredible speed. How do you see genomic surveillance playing a role in future pandemics?

Colijn: Genomic sequencing has become much cheaper and faster. In the future, we may have rapid real-time monitoring of the virus, together with other respiratory pathogens, built into public health operations as well as research systems so that the results are actionable. That’s a vision, rather than a prediction. But I hope that’s where it goes.

Otto: Hopefully the advances that we’ve made will motivate improved surveillance of viruses and wildlife reservoirs and will enable more rapid sharing of data and analysis. Genomics will continue to play a key role in tracking the evolution of viruses like SARS-CoV-2.


The Canadian COVID-19 Genomics Network (CanCOGeN) is on a mission to respond to COVID-19 by generating accessible and usable data from viral and host genomes to inform public health and policy decisions, and guide treatment and vaccine development. This pan-Canadian consortium is led by Genome Canada, in partnership with six regional Genome Centres, the National Microbiology Lab and provincial public health labs, genome sequencing centres (through CGEn), hospitals, academia and industry across the country.