Researchers predict 'vaccine scares' using Google and Twitter trends

What do Google searches and tweets tell us about disease outbreaks? As it turns out, analyzing search and tweet trends could give warning signs for when a disease outbreak may happen due to reduced vaccinations.

An international team of researchers analyzed searches and tweets related to measles and the measles-mumps-rubella vaccine using artificial intelligence and a mathematical model, and detected warning signs of a "tipping point" two years before the Disneyland outbreak happened.

In early 2015, there was a measles outbreak that was traced to Disneyland in California. Many of the people who fell ill in Disneyland were not immunized some too young for the vaccine and others had personal reasons for refusing shots. The outbreak was declared months later in the spring.

Chris Bauch, one of the researchers and a professor of applied mathematics at University of Waterloo, told CBC News the "tipping point" is where there begins to be"enormous decline" in vaccinations due to fear of "vaccine risk."

Bauch said while there were warning signs for California that showed it was approaching a tipping point, it didn't actually cross over.

"You had this outbreak which seems to have made people more scared of the disease againand that pushed the population back away from the tipping point."

The study was conducted in collaboration with researchers from Dartmouth College in the United States andcole polytechnique fdrale de Lausanne in Switzerland. It was recently published in theProceedings of the National Academy of Sciences.

What warning signs look like

As to what those warning signs look like in real life, outside of the data, Bauch said there tends to be a lot of "variability" and "change in the population's opinion in pro or against vaccines."

Bauch is hoping his team's work can be used for public health units and other organizations to focus their resources on educational campaigns on vaccines, to ensure the rates don't go low enough to set off a disease outbreak.

He made an analogy using people who are at higher risk for heart disease due to smoking, where physicians could target those patients and advise them to stop smoking.

"In the same way, we hope to target populations where they're showing a lot of variability," Bauch said."And we can therefore say, well if we don't do something here, the population might have a vaccine scare, an outbreak, so we should try to do something to stop that."

For now, his research team will work on expanding the research to analyze trends in searches and tweets that are in languages other than English.