As the COVID-19 pandemic wreaked havoc and lives were at stake, the advice experts gave to decision-makers became indispensable. At the same time, heated debates arose when researchers’ calculations led to differing conclusions on everything from face masks to school closures. A new handbook on mathematical models has now been produced jointly by Chalmers University of Technology and the University of Gothenburg, Sweden, and several Swedish government agencies. The handbook is intended to pave the way for better decision-making and greater preparedness for the next pandemic.

A mathematical model is a simplification of reality that can help us navigate a complex world. During the COVID-19 pandemic, mathematical models were used to simulate the spread of the virus, predict healthcare needs and assess the impact of various measures – from lockdowns to handwashing routines and face coverings.

By translating various factors into mathematical terms – such as data on risk groups and demographics, or information on those infected, those who recovered, and those who died – researchers were able to use mathematical tools to produce forecasts and advise decision-makers on key decisions.

Torbjörn Lundh is a professor of biomathematics at Chalmers University of Technology and the University of Gothenburg. During the pandemic, he helped Sahlgrenska University Hospital in Gothenburg, Sweden, to estimate the demand for intensive care beds on a week-by-week basis using mathematical modelling.

He is now one of the authors of a new handbook produced jointly between Chalmers University of Technology, the Public Health Agency of Sweden, the Swedish Defence Research Agency (FOI) and the Swedish Armed Forces. It provides practical guidance on how mathematical models can be used to inform decision-making, and how the results can be communicated in times of crisis when most things are uncertain – and time is of the essence.

‘This is a book I would have loved to have had myself during the COVID-19 pandemic. Then I could have been even more effective and confident in my work,’ says Torbjörn Lundh.

A model is not a definitive answer

Philip Gerlee, professor of biomathematics at Chalmers University of Technology and the University of Gothenburg, is the lead researcher for the handbook. He hopes that it will raise awareness of different models and how best to deal with them – and thereby pave the way for better preparedness for future pandemics.

‘No model can provide a definitive answer, but they can still be very useful. For us, the handbook arose out of frustration at the misconceptions and, at times, the harsh tone of exchanges between different groups that emerged in Sweden during the pandemic – and which also occurred in other countries. We want to show that all models are simplifications, but that with the right assumptions they can be helpful to decision-makers and that different models can complement one another. Hopefully, this will lead to better collaboration between experts so that we can provide better advice, more effectively, to decision-makers during the next pandemic,’ says Philip Gerlee.

Anders Tegnell, senior adviser at the Public Health Agency of Sweden and former Chief Medical Officer, is one of the co-authors. He recalls the challenges faced during the COVID-19 pandemic, when many organisations wanted to assist in what was a chaotic situation.

‘As everything happened so quickly and many people wanted to contribute their expertise, there was a certain amount of confusion over terminology and even mistrust between different groups. One example of how this played out was in opinion pieces in the Swedish media that were not particularly constructive,’ says Anders Tegnell.

Different models provide a broader picture

A chemist, a mathematician or a biologist will often use completely different models in their work, based on, for example, AI, differential equations or various data models. But, according to Torbjörn Lundh, the wide range of tools is not a problem. Quite the opposite, in fact.

‘Different models and results can provide a broader picture and a deeper understanding. It is rarely a good idea to rely solely on one model, and not all of them work as well across all stages. For example, AI models were difficult to use at the start of the COVID-19 pandemic when there was not yet enough data,’ he says.

If several models point in the same direction, the reliability of the results increases. Another important conclusion is that there are risks associated with relying on overly complex models. Torbjörn Lundh gives an example: the controversial report from Imperial College London in March 2020, which predicted hundreds of thousands of deaths and an overwhelmed healthcare system unless strict restrictions were introduced. Since then, several researchers have criticised the way in which the model underlying the report was used.

‘The more complex a model is, the harder it is to explain and understand. In addition, the results can vary greatly based on even very minor changes to the parameters you set,’ he says.

Swedish data modellers are gearing up for future pandemics

It is also important to ‘rehearse’ together during periods when the virus is not spreading, something that is currently taking place in Sweden as part of the national SEMAFOR – Swedish Epidemic Modelling and Force network.

‘We are a group of pandemic preparedness modellers from government agencies and universities in Sweden who meet to carry out realistic training exercises. For example, we held a mock press conference about dengue fever reaching Stockholm, in which Chief Medical Officer Anders Tegnell played himself. This network has broadened our perspective on all the tools available in pandemic preparedness modelling, and on how we can improve together,’ says Lundh.

More information about the handbook can be found in a news article on chalmers.se: New handbook aims to strengthen Sweden’s preparedness for future pandemics. The text may be used in its entirety.

More about the handbook:

The Handbook of Mathematical Modelling of Infectious Diseases for Decision-Making is part of a larger project funded by the Swedish Civil Contingencies Agency (MCF) (formerly MSB).
The authors are Philip Gerlee and Torbjörn Lundh from Chalmers University of Technology and the University of Gothenburg, Sweden, Lisa Brouwers and Anders Tegnell from the Public Health Agency of Sweden, and Oscar Björnham from the Swedish Defence Research Agency (FOI).
The project also organises exercises with a group of modellers from the SEMAFOR – Swedish Epidemic Modelling and Force network.