4min read
PREVIOUS ARTICLE Google pitches 'showcase' as w... NEXT ARTICLE Green shoots after COVID-19 hi...

Your Facebook data could soon help authorities predict and stop the spread of the coronavirus in the early stages of an outbreak.

Harvested by the social media giant with permission, Facebook’s data shows a user’s movements throughout the day.

Using the de-identified information, Australian researchers have built a model that can predict how clusters would spread.

The team, from The University of Melbourne, The University of Adelaide, Monash University and The University of NSW, then applied to model to three real Australian outbreaks, comparing the prediction with actual case numbers.

“It does a pretty good job,” author of the paper, published in the Journal of the Royal Society Interface, Dr Cameron Zachreson told AAP.

The data is particularly useful in the early days of an outbreak, before new cases reach dozens a day, he said.

“Probably the Berala (Western Sydney) cluster is in the stage now where something like this might be useful.”

Other risk factors – such as a person’s job – can also be included in the model.

“We did that with (the Victorian outbreak at) Cedar Meats,” he said.

“We looked at the proportion of meat workers in each local government area, and we used that as a secondary risk factor in our predicted risk map.”

Dr Zachreson says the results are often not surprising, but are hard evidence authorities can base decisions on.

He has already provided some of his COVID-modelling to the Victorian government, and hopes it can be used to help governments focus public health measures – lockdowns, mandatory mask use and so on – more keenly on high risk areas.

“The nice thing about this is that it’s not resource intensive at all. All of the work is basically done by computer programs, there’s no major man hours going into this.”

Dr Zachreson is working on an automated, real-time model he hopes will be available to the public – or at least public health agencies – soon.