Coronavirus prompts scientists to make phone app to track symptoms and predict COVID-19 outbreaks

Image credit: Brendan Esposito, ABC News

Some of Australia's leading astrophysicists have teamed up with public health experts to detect possible outbreaks of COVID-19, even before testing takes place.

About 12,000 people globally are already using a trial version of an app, which they log into daily to answer questions about symptoms and risk factors not currently recorded by health authorities.
The astrophysicists use supercomputers, usually reserved to process data from the world's largest telescopes, to identify geographical clusters of symptoms and detect where COVID-19 could be spreading.

Buoyed by promising early results, the team will this week launch the 'BeatCOVID19Now' app to crowdsource anonymous data from users around the world.

​

Swinburne University astrophysics professor Matthew Bailes, who leads Australia's research into gravitational waves, has put that research on hold to devote his time — and supercomputers — to the project.

"We're used to catching literally hundreds of gigabytes a second of data from the Square Kilometre Array[radio telescope project]," Professor Bailes said.
"And the amount of data we're talking about for this is a small fraction of that."

Supercomputers crunch data to predict outbreaks
Professor Bailes has teamed up with Swinburne public health professor Richard Osborne, who previously worked with the World Health Organisation and the Australian Bureau of Statistics on disease-tracking surveys.

"We need to know where the next epidemic is going to break out … so we can actually beat COVID-19 by getting in there early and supporting people to be safe," Professor Osborne said.
The project is independent of a recently announced Australian Government plan to develop an app that would monitor users' movements and contacts.
That app would require 40 per cent of Australians to sign up to be effective, the Government believes.
For this project, the Swinburne scientists believe they need data from just a few hundred people in a location — as small as a suburb or as large as a city — to produce useful results.

"We'd like to be able to turn back the clock on our survey data and say, 'You know what? Three days before that outbreak, we saw lots of people with sore throats or runny noses or they lost their sense of taste'," Professor Bailes said.
"Those sorts of statistics could be correlated with actual outbreaks and then we could start predicting outbreaks in the area before they happen.
"And that would really help the Government with its planning."

How it works:

• Users are prompted to log in to a mobile phone app each day to answer a series of questions — even if they feel healthy
• Supercomputers analyse information about symptom clusters, and users' travel history and likely exposure to COVID-19, to identify areas of likely outbreaks
• The information will be made available to authorities around the world, so they could target resources like testing kits to those areas

Initial results point to promise
Based on the early data from a small number of users, the researchers said the program indicated there were outbreaks in certain Melbourne suburbs, which matched official Government data.
Researchers say that although the analysis was conducted retrospectively, it demonstrates that the program should be effective in predicting outbreaks.
The program also saw an influx of people logging in from Los Angeles, days before an outbreak was seen there. The researchers say lots of people using the app could have been an early sign that people there were getting sick.
"I think that's what a local pandemic can do to your motivation to want to track your symptoms," Professor Bailes said.
To date, most users have discovered the website through social media, but marketing and IT experts have volunteered their time to help promote the phone app.
"The reason I got in this project is because I wanted to do something to help," Professor Bailes said.
"And I think everybody sitting out there can help by just doing this survey. They can be part of the solution."

Professor Osborne said the team was now working on ways to share the data with health authorities — including Australia's Department of Health — and other coronavirus researchers.
Nancy Baxter, head of the Melbourne School of Population and Global Health, said the app would be a useful tool if the Swinburne researchers could get enough people to use it.
"Having some way of looking and seeing in the population who … likely has COVID-19 can be really helpful in terms of knowing where to target testing," Professor Baxter said.
She said it could also help prevent the disease spreading by finding and isolating patients before they fell very ill — but other measures would still be vital, such as widespread testing, surveillance, contact tracing and quarantine.
"It can add to what we're doing, but it's not the fundamental part of what's what we need to do," she said.

As featured on ABC News