USF College of Public Health’s Dr. Michael Edwin, professor of epidemiology and population ecology of disease transmission, says the future of COVID-19 infections can, contrary to recent debate, be forecasted using data-driven mathematical models.

He and colleagues present their findings in “Combining predictive models with future change scenarios can produce credible forecasts of COVID-19 futures,” which has been published in Plos One.

They examined how their current COVID-19 forecasting models could be combined with scenarios of plausible future behavioral changes, such as changes in uptakes of vaccinations and compliance with social mitigation measures, to estimate the likely future paths the virus may take in a setting.

According to Michael, data-driven models can provide useful predictions of the pandemic future.

“Our contribution in this paper is to show that provided the parameter space of compartmental models, such as our extended SEIR model, is not entirely depleted by the previous history of the pandemic, and timely information regarding the arrival of new variants is provided through real-time surveillance, then these models can be used together with scenarios outlining plausible future changes in public compliance with vaccination and behaviorally-based protective measures in order to estimate the likely future paths that may be taken by the pandemic in a given setting,” he said.

These findings are significant, according to Michael, who said it indicates that these types of models may have the structure and information in them to produce more credible predictions for the pandemic, making them important informational tools to be used in predictions.

Michael said that while this data-driven modelling approach has allowed him and his colleagues to “faithfully reproduce the temporal evolution of the pandemic, modelling systems, such as this which depend on the entire previous history of the pandemic have been criticized as not being able to predict the future course of a highly uncertain and complex pandemic.”

Therefore, they wanted to see if this could be resolved by ensuring that model parameters were not too “constrained or depleted by past data, data on variants are provided as they emerge, and by modelling plausible future scenarios of changing population adherences to vaccinations and social measures.”

Michael said they fit their model to data from the start of the pandemic to September 24, 2021 to see if forward predictions generated by their modelling system could match the future omicron wave that emerged during the winter of 2021/2022.

“To our pleasant surprise, the results showed that our model was able to predict this future wave provided the omicron virus emerged on November 1, 2021, had a transmissibility rate two times higher than that estimated for the delta variant, and natural immunity waned over 2.5 years,” he said. “Indeed, we also showed that because immunity is not permanent, the pandemic is unlikely to fade away but that its future will be marked by repeat waves with sizes determined by the both the levels of immunity in operation during variant emergence, its transmissibility and immune evasiveness.”

Michael says their modelling framework uses live data as it comes in to update model parameters and add new structures.

“This allows us to incorporate impacts related to changing environmental/behavioral drivers of transmission, such as community trends in compliance with social protective measures and vaccination uptake and the emergence of new variants, on important parameters such as transmission rates and development of both vaccine-induced and natural immunities,” he said.

As for the pandemic future, the key result emerging from this work is that immunity is not permanent and that we should expect to see repeat waves, according to Michael.

“The size and emergence of this wave will depend in a complex fashion on the interplay between levels of natural immunity established and operating at any given future time in the population and the transmissibility/immune evasiveness of emerging variants,” he said. “Provided immune evasiveness does not increase significantly and virulence of the new variants is similar to that observed for omicron, then it is also possible that while we may see repeat waves of rising cases, these will not lead to major hospitalizations.”

The major unknown in this scenario, he says, is whether a more immune evasive variant that can counter the protection afforded by natural immunity may emerge.

“Models can not only uncover key drivers that govern continual transmission and resurgence, but also point to experimental investigations required to resolve the contributions of these drivers more directly. Our finding that the pandemic is unlikely to fade out with waning of immunity but will most likely oscillate with repeat waves forming in the future also indicates a need to continue monitoring the pandemic and to continue to study the factors behind such persistence,” he said.

Michael says their models shows that vaccine-induced immunity is less robust and lasts less than natural immunity.

In fact, he says the pandemic will not fade away, but will instead oscillate with formation of repeat waves going forward due to waning of immunity and arrival of variants.

“This means that repeat waves in the near future will largely be driven by wanning of vaccine-induced immunity, and that it will be high levels of natural immunity, that is immunity derived from infection, that will ultimately allow control of the pandemic. Natural immunity is already around 85 percent in most U.S. communities as a result of past infection and from breakthrough infection of vaccinated individuals. This immunity is also likely more diverse than vaccine-induced immunity meaning that it will be more robustly protective against variants too,” he said.

All of this suggests, according to Michael, that we have arrived in a complex immunity landscape in which ensuring high levels of naturally acquired immunity will be more important than COVID-19 vaccination for influencing viral transmission.

“However, as natural immunity will also wane over the longer term, repeat waves can also be expected from this waning in the longer-term future. In fact, natural immunity will oscillate around herd immunity levels (but will not quite reach it sustainably) as repeat waves of infections rise and fall going forward,” he said.

This suggests that as long as we are able to protect the most vulnerable subpopulations against disease perhaps through repeat vaccinations, natural immunity while not be able to prevent repeat waves from forming (because it will fluctuate) but will act to contain the size of any forming repeat waves, according to Michael.

Both Hillsborough County and the County Department of Health (DOH) are partners in this work.

“We benefited greatly from the questions they [Hillsborough County and the DOH] raised regarding the likely future course of the pandemic. These questions helped shape some of our thinking while undertaking our research, particularly with respect to ensuring that our results also met these policy concerns,” he said.

Michael says they continue to update their model forecasts every two weeks in order to provide updated forecasts to the county and DOH officials.

“Our work helps in allowing health officers and city commissioners to not only appreciate how the pandemic may be evolving but also to gain a deeper understanding of the reasons for the continual transmission of the virus, including the complex integrated impacts of changing levels of vaccinations, social measures and natural immunity on pandemic evolution,” he said. “Further, our simulations of how best to control the pandemic by increasing either or both social measures or vaccinations allow them to inspect the relative importance of either control measure, and when they might have to consider these options.”

Story by Anna Mayor, USF College of Public Health