Can Google Search Terms Predict COVID-19 Hotspots?

The coronavirus pandemic hit most of us by surprise. One moment, we’re enjoying brunch and a matinee with friends. And the next? Lockdown—and things haven’t been normal ever since.

However, as a recent article published in the Mayo Clinic Proceedings suggests, paying attention to search terms in Google Trends may have provided clues to our current predicament. Used with proper analytical techniques and in conjunction with other disease surveillance methods, search queries may help predict infectious-disease hotspots in the future, too.

How Google Trends Works

Google Trends is a feature in Google that lets you see what people are searching for. You can discover which queries are most popular on Google over a period of time, and you can also explore popular keywords being used in a particular geographic area. 

“Google Trends offers a normalized value that represents how popular a particular query is, based off a sample of Google searches,” explains Ashlynn Daughton, PhD, an information scientist at Los Alamos National Laboratory in Los Alamos, NM (who wasn’t involved in the Mayo Clinic research). “Google uses some process to represent the values as '<1'—meaning it’s a very unpopular query—to 100, which means that it’s very popular. But the numbers themselves don’t mean anything concrete. So, a value of 95 for a particular search just tells you that it's very popular, but it doesn’t say anything about how many searches there actually were for it.”

For the past several years, “web-based analytics” has become a promising tool for helping scientists make public health predictions, particularly in the hopes of staying ahead of the spread of infectious diseases. This can be helpful on the population level.

“Traditionally, disease surveillance can be time consuming and complicated,” Daughton tells VerywellHealth. Widespread testing and public health reporting can cause a lag—for instance, people may not get tested or check themselves into a hospital until after they’ve Googled the symptoms that they were suffering from.

But as the Mayo Clinic researchers found, by analyzing Google Trends results, you can find information about areas that may be headed for an outbreak.

“This data can be used for better allocating personal protective equipment, medications, and more,” says Mohamad Bydon, MD, a Mayo Clinic neurosurgeon and principal investigator at Mayo's Neuro-Informatics Laboratory, in a news release last month.

What Google Trends Might Tell Researchers About COVID-19

When Mayo Clinic researchers analyzed 10 Google Trends search terms very early in the coronavirus pandemic—from January 22 to April 6—certain terms were prominent at certain times. The 10 terms included:

• COVID symptoms
• coronavirus symptoms
• sore-throat+shortness of breath + fatigue + cough
• coronavirus testing center
• loss of smell
• Lysol
• antibody
• face mask
• coronavirus vaccine
• COVID stimulus check

The researchers performed a search query for each keyword for each state in the U.S., and they plotted their findings against the number of new COVID-19 cases. Perhaps not surprisingly, “COVID symptoms” peaked in search toward mid-March. The terms “coronavirus testing center,” “loss of smell,” and “face mask” also had strong correlations with increases in COVID-19 cases, both at a national level and on a state-by-state basis. 

How Web-analytics Might Help Public Health Officials Manage COVID-19

The “real time” data collected by Google Trends is not a moment-by-moment picture. For most people, it delivers data at the monthly level, though researchers with special agreements may have access to weekly data. Regardless, it can offer an interesting snapshot.

“With the right keywords, scientists can get a picture of what people are searching for, which we think relates to what is happening or might happen in the future more generally,” Daughton says.

While everyone has access to the basic offerings of Google Trends, tracking COVID-19 cases isn’t a simple matter of tallying up keywords. Information scientists work with statistical equations to understand the relevance and timing of search queries and to figure out what particular terms may be relevant to their goal. 

Certain terms may have different correlations. For instance, “sore throat” may be a symptom of COVID, Bydon tells Verywell, but it is a “difficult search term because it applies to so many other diseases outside of COVID.” 

Daughton also points out there is such a thing as the “Oprah effect,” where queries don’t correlate to cases because people are searching for terms out of curiosity—it’s mentioned in the news or by someone prominent, and queries shoot up—but these queries doesn’t necessarily reflect what the searchers are experiencing themselves.

What’s more, Google Trends isn’t yet able to dig down into, say, the infection rates in your specific community, which may be more relevant to your everyday life—for instance, whether you’d like to take part in an in-person event or do so remotely.

However, search terms can tell researchers features about aspects of a particular disease that they may not know about.

“Back in March, it wasn’t clear that a loss of smell was a symptom of COVID,” Daughton says. But when you look at search terms retroactively, you might glean that people were experiencing it because they were Googling it. 

As of now, web analytics alone can’t provide a complete picture of a pandemic or predict what’s to come, but it has proven to be a robust tool. “This should be part of an overall strategy that uses multiple metrics to track and predict the spread of the coronavirus pandemic,” Bydon says. Currently, web analytics are part of the Mayo Clinic’s coronavirus tracking tool. This online resource enables users to click on a county within a state and find out the number of new COVID-19 cases each day, as well as the rate and total number of cases.