As Canadians settle into the winter months, flu activity is beginning to pick up. Many will be getting their annual flu shot, but this year comes with a complication: virus strains circulating right now don’t match up with current vaccines.
York University experts explain why this happens and what innovations they are pursuing to help avoid mismatches in the future.
The roots of a mismatch begin long before a vaccine reaches clinics. “The formulation of the flu vaccine, and what strains go inside it, are determined six months ahead of flu season,” says Jane Heffernan, Faculty of Science professor and co-director of the Canadian Centre for Disease Modelling.
Each year, the World Health Organization reviews global influenza data from laboratories, clinics and public health agencies to predict which variants are most likely to dominate. Recommendations for seasonal vaccines are based on that research and once Health Canada gives authorization, manufacturers begin mass-producing doses for distribution six months later.
That long production window gives the virus time to change. As strains circulate through the southern hemisphere and move north, new versions can emerge. “In your body, when you have the flu, your cells are becoming little factories that make new flu virus particles,” says Heffernan. Errors in that process can produce variants with small adaptations, such as better ability to infect cells or dodge immune defences. As these variations spread, they can outcompete other strains.
Those shifts in the virus matter – they shape how well the seasonal flu shot performs. That is what's happening right now, notes Heffernan, as early Canadian surveillance shows the H3N2 strain of influenza now circulating is different – like a cousin – from the one included in this year’s vaccine. What that means for vaccinated people who still get sick comes down to how the immune system responds.
A vaccine teaches the body to recognize key parts of the virus by treating its harmless components as a warning signal. That training allows the immune system to react faster upon exposure, which can prevent infection or lessen symptoms. A mismatch complicates that response because the antibodies produced may not identify the circulating variant as readily.
Due to that, Heffernan says, more Canadians may become ill or face stronger symptoms, particularly older adults and those with chronic health conditions. Even though misaligned flu shots are not uncommon, news about a mismatched vaccine have negative implications in an unexpected way.
"Often, when people hear news about mismatches, vaccine uptake decreases under the assumption that ‘the vaccine is no good this year,’” says Seyed Moghadas, associate dean of research and graduate education and professor of computational epidemiology and vaccine science in the Faculty of Science. That reaction can worsen outcomes, however, because it overlooks a key point. “Though seasons with poorly matched vaccines may be more severe, the important message is often ignored: even a poorly matched vaccine can reduce disease severity and outcomes,” he says.
Even without a perfect match, the vaccine remains close enough to give the immune system a valuable head start. Prior exposure, whether through vaccination or past infection, helps the body respond more rapidly and reduces the risk of complications. Because of that, experts like Moghadas urge that getting a flu vaccine during a mismatch season is even more important.
Still, the ideal scenario is to avoid variant divergence entirely. Despite the challenge of predicting dominant strains months in advance, York experts are developing new approaches to improve accuracy in future seasons. One solution focuses on enhancing the predictions that guide vaccine formulation.
Huaiping Zhu, a professor of mathematics and statistics and Director of the Canadian Centre for Disease Modelling, has led the One Health Modelling Network for Emerging Infections (OMNI / REUNIS), a collaboration of more than 100 researchers across human, animal and environmental health. The network tracks a wide range of viruses, but by combining climate forecasts, wildlife data, animal-health reports and human surveillance, it can spot early signs of seasonal influenza strains that traditional systems might miss.
The network used advanced mathematical models that updated in real time as new information becomes available. This helps account for details simpler models often miss, like incubation periods or testing delays, improving predictions of how influenza spreads and how quickly variants may appear.
Zhu emphasizes the challenge of forecasting influenza variants: “The key is how to predict emerging issues. Every year will be different. That’s where co-ordinated monitoring and modelling really matter.” He adds that his group’s approach is proactive: “Using the information available, we can look at the logic and possible development and evolution to try to predict which strains are most likely to come.” His team has developed mathematical and weather-linked models to forecast outbreak risk and evaluate vaccination strategies under multi-strain and seasonal conditions.
In parallel, Heffernan leads research at York’s Modelling Infection & Immunity Lab, where her team uses mathematical and machine-learning models to track how immune responses change after vaccination. By monitoring antibody levels, immune memory and booster timing, her models show when protection fades and which groups may need extra or earlier doses. Supported by the National Research Council, this work pairs immune-response data with viral-evolution trends to guide seasonal shot timing and vaccine strategies, complementing broader efforts to improve influenza prediction and preparedness.
Beyond modelling, Heffernan highlights another key factor: vaccine production. “If you can make a vaccine in four weeks instead of six months, then your projection of the dominant strain will be more reliable because it's more recent,” she says.
Building on this, Jianhong Wu, distinguished research professor, Canada Research Chair in Industrial and Applied Mathematics, and director of York’s newly created Centre of Excellence for AI in Public Health Advancements, applies artificial intelligence and advanced modelling to improve vaccination strategies. He and his team work with public health agencies and the vaccine industry to evaluate the health and economic benefits of vaccination and turn surveillance and modelling data into guidance for vaccine production.
Looking ahead, Wu suggests these tools could support strategies such as engaging multiple vaccine producers to shorten production times, helping vaccines better match emerging strains. Modelling which influenza strains are likely to dominate could give public health systems and manufacturers more flexibility to respond as viruses evolve.
He also envisions a more personalized approach, where individuals with different health profiles could access vaccines tailored to their risk, such as high-dose options for seniors. “The ultimate goal is personalized vaccination, so people with varying susceptibility and health status can receive the most appropriate protection,” he says. AI and modelling would allow emerging surveillance and immune-response data to be integrated, guiding these strategies effectively.
Zhu emphasizes the importance of integrating these approaches. “While no method can completely predict how influenza will evolve, combining modelling with faster production techniques could allow vaccines to better match emerging strains,” he says.
By merging advanced prediction, immune-response modelling and AI-driven strategies, York researchers are expanding the scope of vaccine design. Although mismatches highlight the complexity of influenza, even an imperfect flu shot provides crucial protection and helps Canadians stay healthier throughout the winter months.