The nation’s pandemic modelers are reshaping
At the start of last year, Australia’s national modeling was awarded to the Doherty Institute in Melbourne, a move that has since generated some dissatisfaction within the modeling community at large. The problem does not lie with the Doherty Institute, which works with 30 experts from 11 institutions in five states. The problem is with the nondisclosure agreements and the exclusivity designation, to have this collaboration under one banner.
Many believe the details of the national model should be subjected to a full review before being presented as a fait accompli. To support this view, some point to a crucial mistake made early in the pandemic when anxiety across the country was high.
Charts from the Doherty Institute estimated that, without restrictions, the daily demand for new intensive care beds could reach 35,000 – a figure beyond Australia’s capacity. It was scary and it was wrong. The 35,000 were for hospital admissions, not intensive care beds.
The Doherty Institute was alerted to the error in April by Professor Emma McBryde, an infectious disease physician with a doctorate in mathematical and statistical modeling of infectious diseases. He acknowledged the mistake, alerted the government and corrected the mistake in writing in August.
“Sometimes, even if we would like to, because of confidentiality, we can’t ask colleagues to criticize our work – and that’s how you make it better,” says Hellard, who is also a member of the consortium of the Doherty Institute. “The Doherty works very hard and does a good job. Anything where we have the opportunity to support each other can only help us do a better job.
“The way forward is to get the government to recognize the benefits of independent modeling groups criticizing each other’s work. We need a mechanism that allows this to be done under the confidentiality agreement, before the work is turned over to the government or the community, ”she said. “I suspect this would improve the quality and robustness of the work. “
Hellard says Australia has many good role models and “it is extremely important” that many voices are heard in the future. Last month, she spoke at a forum on the state of modeling, hosted on Zoom by the Australian Academy of Health Sciences and Medicine. It was a private meeting attended by between 30 and 40 professionals.
Top doctors and health workers from Canberra and the states were also invited. The central point of the discussion was whether Australia should adopt the same type of modeling process that is used in Britain to make important decisions regarding COVID-19.
“The UK requires at least three different groups to do this modeling,” said Tony Blakely, professor of epidemiology at the University of Melbourne, who also addressed the forum. “If all three of them use the same inputs but different model structures, assumptions, and methods, and return the same answer, they’re pretty confident they’re on the right track.
“They would be worried with three different answers and would ask the groups to compare their notes, see where they agree, where they differ and understand what is going on,” he says. “As we deal with important issues, such as decreased vaccine immunity and the need for boosters in Australia, we should use the UK model. It is better to have a policy based on a diversity of models rather than a single modeling group.
It is the main driver of a new online site for experts, decision makers and the public looking for quick access to modeling studies in an understandable way. The Australian COVID-19 Modeling Initiative is supported by well-known institutes and academic departments, and is an attempt to create more transparency.
“It’s like a showcase where, for example, a chief medical officer of health can come and see who is doing what and find the people who are doing it,” says Blakely. “Right now they’ve all been pretty much tied to what’s going through the peak Australian Senior Health Protection Committee, which created an unfortunate conflict of interest.
“The two people who provide expert epidemiological modeling advice to the committee are also the two people who lead the consortium that provides the modeling,” he says. “This structure was put in place in haste during the pandemic. Now is the time to reassess and remove some conflicts. If we could go back, we would strongly advise the government to follow the UK process rather than the AHPPC process. “
Improve all the time
Still, he says Australia has improved a lot in modeling over the past two years and peaked with the Victorian roadmap in September. “It was underpinned by the Burnet modeling which was immediately released to the public in enough detail for the modelers to understand and for the public to read. It was impressive.”
McBryde agrees that the modeling is improving. “At the start of the pandemic, knowledge about the virus was limited and decisions were based on analogies with other viruses and prediction models.
“As real world data has been acquired in epidemiological studies, better predictive models have become possible. “
McBryde, of the Australian Institute of Tropical Health and Medicine at James Cook University, admires the transparency of the British model, even though it is “more practical for governments to do things quietly and use the models that they trust ”.
“When modeling is done in the dark, it creates a potential perception that there may have been undue influence during the process. Highlighting it would allow modelers to deliver results without fear and have the opportunity to examine commonalities between their models, ”she says. “Consensus statements, like those from the UK, are very valuable. They reveal the commonalities and the differences. Some differences are useful and others need to be eliminated.
To order and manage multiple models, the government would need to increase its capacity. Working with a single model maker requires a large investment of time. It is an iterative process and repeating it three times in the emergency of a pandemic would require considerable effort.
Inspiration from overseas
The UK could do it because it is a global powerhouse in modeling and, compared to Australia, already had a high level of established expertise in this area, says Professor Jodie McVernon, director of epidemiology at the Doherty Institute, which co-directs the National Modeling Program. team. Britain also has an independent intermediary body which evaluates the work of the three groups and submits it to the government.
In Australia, the two leaders of the Doherty group, McVernon and James McCaw, professor of mathematical biology at the University of Melbourne, act as intermediaries between the consortium and the government.
“In Australia, only one product goes to government, but it draws deeply on the country’s expertise and is shaped by many people,” says McVernon. It is multi-part and uses a network approach to answer multiple questions from government. Every week, he provides the government with an assessment of the situation. A multiple model principle is also used with three distinct groups – Adelaide, Melbourne and Defense – producing an ensemble model for pandemic prediction.
“We’ve had long collaborations with Defense, Health and DFAT (Department of Foreign Affairs and Trade) – that’s how we started in the room in February of last year,” says McVernon. She says the Doherty Institute is not an isolated agent and has strong global links with leaders in the field in the UK and other groups through the formal and informal networks of the World Organization. health.
“We do not act in isolation. Our access to discussion and reflection has been more international than national, and this is how we compare our work.
For academics working with government, she says the issue of transparency is a challenge. “But when you’re involved in a political process, it’s very participatory, and we work iteratively on the issue with decision-makers. So we are not free to go out and talk about it until a decision is made.
“It’s not that unusual because in academia most people don’t openly discuss development work because of the competition. “
Despite some tensions, one of the bright spots of this pandemic is that Australia has revealed its enormous capacity and growing expertise in pandemic modeling.