A clever new model could change the way officials respond to new Covid-19 outbreaks, by predicting how quickly the virus can spread from city to city – or even suburb to suburb.
The model, just completed by Te Pūnaha Matatini researchers, brings together an impressive array of new data, capturing how many people are likely to live in certain households in certain places, and where they travel to each day.
Traditionally, scientists tracking disease outbreaks have used what are called compartment models.
While these offer a general picture of how an infection can travel across a population, they also assume that anyone in a population can meet and infect anyone else on a given day.
Over the Covid-19 crisis, Te Pūnaha Matatini experts have been using a sophisticated version of a compartment model, and which factors in various pieces of social information to produce more accurate predictions.
It's been invaluable in efforts to stamp out community clusters, but a completely new type they have been quietly working on with Government officials is expected to prove much more useful.
Its main ingredient is an individual level interaction network built from a Statistics NZ-run dataset covering most of New Zealand's five-million-strong population, called the Integrated Data Infrastructure, or IDI.
It brings together Census data and tax records that detail what industries people work in – something particularly useful for predicting movements of essential workers over lockdown.
"We can ask how the amount of inter-generational living varies from place to place, or how likely it was that people of a particular age or sex, worked in a particular industry sector, in a particular region of New Zealand," one of the project's leader's, Dr Dion O'Neale, said.
"But rather than trying to build one single network that's an exact copy of the people in Aotearoa – and there are lots of reasons why this would be a bad idea - we use this data to build multiple representative versions to capture real-world properties of how people interact."
Other ingredients like commuting data, and electronic payment records from MarketView, were crucial in simulating movement.
"Since people will often be infectious for three or four days before they have any symptoms, knowing about these movement patterns means we can calculate how far the infection might have spread before they are first confirmed as being infected," he said.
"We also use data from sources like the National Contact Tracing Service to make sure we have accurate parameters for things like the length of time it takes tracers to get in touch with the close contacts of someone who has been infected."
Epidemiological case data, too, fed in information like the fraction of infected people most likely to be asymptomatic – and therefore tough to track through contact tracing.
O'Neale, who's been leading a team of eight alongside fellow researchers Dr Oliver Maclaren and Dr Emily Harvey, said previous models like compartment models had the benefit of being quick to develop and run.
But they couldn't match the depth and detail of what his team had created.
"We can better address issues like the spatial aspect of spread - or questions about relative risk and equity factors that are relevant to disease transmission."
Already, the researchers had used it to learn lockdowns at post-code or suburb level were unlikely to be effective, given the large amount of movement between areas every day through commuting and travel, and the fact that the virus was able to spread for days before symptoms appeared.
"We can however ask, for a new case in a certain area, what sort of spatial spread might we expect to see, or how soon we might expect to detect a case in a new outbreak, as results will differ depending on rates of testing, along with other local factors like access to healthcare."
Situations like Victoria's current wave, which has killed nearly 770 people, and infected nearly 18,500, have highlighted how inequity could make Covid-19 much more devastating.
"Transmission can get out of control when people continue to work when they are infectious – and people on lower incomes, or who have jobs where they are unable to work from home, are less likely to be able to self-isolate.
"They are also more likely to be working multiple jobs and to live in over-crowded dwellings, potentially with more intergenerational living.
"All of these factors can mean they are more likely to be infected, more likely to infect multiple other people and more likely to infect people who will have more severe symptoms and worse health outcomes."
And without building them into models, any interventions to head off Covid-19 were left all the more poorer.
Collaborating directly with the Department of the Prime Minister and Cabinet and Ministry of Health had been especially useful, O'Neale said.
"While we have been providing them with results from the modelling and data analysis, they have been helping us by facilitating access to some of the information that we need to make sure that the models are accurate."
In future, O'Neale envisaged the new model could be used to help distribute vaccines and combat other diseases.