Test, test, test … or click, click, click?
A few weeks ago we published a blog post where we point out the importance of publishing patient data to stop the COVID-19 pandemic. The WHO also urged countries to test, test, test but official data shows that testing strategies among countries are still patchy. Having reliable data is the best way to understand how the disease is progressing and design policies accordingly, but when official information is not available, is it there another way to get the data?
There is a plethora of free, easy-to-use and constantly updated data on Google Trends. Although Google cannot quite be used to determine whether a person is infected or not, trends on search patterns may provide valuable information. Seth Stephens-Davidowitz recently wrote an article for the New York Times where he proposes to forecast local outbreaks by studying surges in COVID-19 symptoms search for a certain area. Moreover, by analysing US state-level data for the weeks prior to the outbreak, he found that people told Google that their eyes hurt more frequently than ever, which may indicate that “eye pain” could be an ignored symptom of the disease. If Seth’s finding is true, we may be missing valuable information about the disease. Moreover, search patterns can also be used to feed more intricate and efficient models and help with tracking the epidemic in real time.
If we want to go a step further in our data collection efforts, we may also ask people to help us doing so. A group of doctors and scientists created the C-19 symptom tracker app which can be downloaded to your phone and keeps track of your health condition every day – even if you are well. More than 1 million users downloaded and reported their symptoms during the first day or so. Similarly, another research team created a questionnaire to gather data on how people prepare and cope with the pandemic. More than 100,000 respondents from different parts of the world answered in the past few weeks.
Finally, we can also exploit automatically generated data to improve the measures in place. For instance, governments are using mobile phone location information to determine whether someone left their home and nudge them to go back to isolation. Similar technology may also be used to warn citizens when they have been in close contact to somebody who has tested positive for the virus.
But not everything is rosy in the garden of free data. For example, we should be careful not to draw hasty conclusions from search patterns – for example Google Flu trends a project launched in 2008 aimed at predicting flu outbreaks was cancelled in 2015 accused of overestimation. We should also be careful when collecting sensitive data – we always risk it falling into the wrong hands.
A novel virus requires novel approaches, but is this data sound enough to predict epidemic outbreaks? Do the cons outweigh the pros? During normal times we would consider this more carefully. But these aren’t normal times: the COVID-19 pandemic is here and we need all the tools in our arsenal.