Battling a global pandemic with maths and stats

One of the few bright spots that has arisen out of the COVID-19 pandemic is that the profile of the mathematical sciences has increased significantly.

Leaders at the highest levels of government now understand the importance of mathematical and statistical thinking when making informed decisions.

Before 2020, a lot of people had never heard of mathematical modelling. Or, more specifically, epidemiological modelling. In fact, there were only a small group of researchers around Australia doing epidemiological modelling.

Dr Andrew Black was in that group. He had been studying the spread of epidemics for over a decade.

Early in 2020, Andrew was among a group of Australian epidemiologists who were closely watching the spread of a new coronavirus that had originated in China and was already starting to spread to other countries.

In March, he suddenly found himself thrust into what would be Australia's initial response to the Coronavirus when he and a few select others travelled to Canberra for a meeting with the Australian Department of Health. Australia had already closed its borders to China.

“Obviously, there were risks of people still bringing the virus to Australia from other parts of the world. We were asked to examine those risks and to start looking at the probabilities of seeing outbreaks as time went by,” says Andrew, a Lecturer at The University of Adelaide and an ACEMS Associate Investigator.

“I was quite involved with that modelling. It was a very intense few weeks of doing that work.”

The pandemic modelling team was led by Professor James McCaw from the School of Mathematics and Statistics at The University of Melbourne and Professor Jodie McVernon from the Peter Doherty Institute for Infection and Immunity. Following initial work on scenario analysis and planning, and with cases arriving in Australia, the team leaders established a situational awareness team. Real-time evaluation of the status of the epidemic and future activity forecasts would become an essential element of the response.

Since 2014, James McCaw and Dr Rob Moss, from the Centre for Epidemiology and Biostatistics at The University of Melbourne, had been developing forecasts for seasonal influenza. But the COVID-19 response required further skills. So, they turned to ACEMS Chief Investigator Rob Hyndman.

Rob is the head of the Department of Econometrics and Business Statistics at Monash University. For years, he’s helped companies and government agencies worldwide forecast things like product sales, energy usage, and tourism. But not epidemics.

“Forecasters and epidemiologists haven’t worked together much before. Epidemiologists usually study disease dynamics and measures to reduce transmission. But with a pandemic occurring, we needed to join forces to provide the best advice on how to manage the situation,” says Rob.

Rob’s job was to put together an ensemble forecast that combines forecasts from multiple models to get an overall forecast. One of the models he included was from an ACEMS team at The University of Adelaide. That team was led by Professor Josh Ross, who had also been researching and modelling the spread of epidemics for a decade.

“The pandemic is one of those things we always knew was inevitable, but still didn’t want to happen,” says Josh, an ACEMS Associate Investigator.

To help him, Josh enlisted ACEMS PhD candidate Dennis Liu and ACEMS Associate Investigator Dr Lewis Mitchell. Their model included Google mobility data about how people are mixing and interacting. It also included regular survey responses from people in Australia’s capital cities about their compliance to guidelines, like the 1.5-metre social distancing rule.

Rob also created a model to add to the ensemble. To help him with that, he enlisted the help of ACEMS postdoctoral researcher Dr Pablo Montero-Manso, who had already been doing pandemic modelling for his home country of Spain. The third model for the ensemble forecast was adapted from existing seasonal influenza forecasting models developed by Rob Moss in collaboration with Defence Science Technology Group and Australian states’ public health departments.

By the end of 2020, the models had matured quite a bit because a lot of information had been built in. They were quite nuanced

Rob Hyndman.

For Rob, this has been a very challenging but interesting experience.

“Sitting in on all those zoom calls for months, I’ve figured I’ve learned a lot about epidemiological modelling, how they do it and what they think about,” says Rob.

Rob has also taught them quite a bit about forecasting.

“As a forecaster, I’m not really interested in the average number of cases when we’re talking about something like a pandemic. I am more interested in the extremes, the upper tail of these probability distributions we were creating. To measure their accuracy, we needed probabilistic scoring methods.”

Their work continues in 2021.

In addition to the modelling and forecasting work advising the National Cabinet, ACEMS researchers tackled many different aspects of the crisis.