The data we’re creating could be key to tracking the spread of the coronavirus

What happens when you feel ill? The chances are you grab your phone, type in your symptoms, and start scrolling through the cornucopia of websites offering advice for whatever ailment you may have. 

Much of the information isn’t useful for anything other than causing panic, turning a minor sniffle into a life-ending cancer or a cough into a deathly illness. But when we’re all hyper-aware of the spread of coronavirus, and taking care to try and check whether our symptoms match up to what may be seen from patients suffering from the pandemic illness, each and every search query could prove a vital lifeline to track the spread of coronavirus.

Google is the most-visited website in the world, and its search box is a place we share surprisingly private thoughts with. Murderers have been convicted thanks to their Google search history demonstrating motives; and in a world which is struggling to keep track of the prevalence of coronavirus due to a lack of testing capacity, search results could provide a strong source of semi-reliable information.

A window into our world

That’s the hunch that a group of transnational researchers had, including some from Harvard Medical School, University College London, and Public Health England. They recognised, in a constantly updated pre-print paper, that tracking online searches to monitor the spread of influenza – which has been commonplace for years – could also provide a glimpse into how coronavirus is moving.

The researchers have used Google’s Health Trends API, a non-public API created by Google and accessible to academics investigating topics around healthcare, to track search trends in the UK, England, US, Canada, Australia, France, Italy and Greece. They’re monitoring search queries relating to covid-19 symptoms and keywords, which they’ve declined to share in case it biases their sample. 

The team have also introduced measures to cancel out the likely search traffic for news stories around coronavirus, knowing that is likely to increase traffic and bias their results.

I search, therefore I am

The research will continue throughout the coronavirus crisis, with the team constantly updating the paper with their new findings as and when they happen. But what they’ve found so far are fascinating. 

As you’d expect, search terms for coronavirus symptoms are increasing significantly as the spread of the disease increases, and public awareness about it ramps up. When we get a sniffle or a cough, we’re turning to Google to find out whether what we have is a common cold or something potentially more serious.

“The mapped trends correspond sufficiently well to confirmed cases data in most countries,” the researchers conclude. A specific look at Italy, which seems to be more advanced in being caught in the grasp of coronavirus than most other countries outside China, shows that the model correlates well with actual cases.

Worrying conclusions

Which makes the conclusions the researchers reach all the more worrying. “The search query frequency data for Italy were best correlated with other countries after shifting them by a number of days; 1for Canada, 2 for the UK and the US, 7for Australia, 8 for England and France, and 18 days for Greece,” they write.

While debates roil about the potential of using smartphone location data to track how people are encountering those who have tested positive for coronavirus, with the risk of infringing on people’s privacy taking a backseat to the potential to slow the spread of the pernicious disease it causes, looking at search data seems like a more fruitful, and less invasive, way of getting access to some real information about how far, fast and wide the coronavirus is spreading.