SDSU researcher: Using Twitter to track the flu could save lives

SDSU researcher using tweets to track flu outbreak

SAN DIEGO - The nation is in the midst of a sneezy, achy, miserable flu season, and it has been a deadly one in San Diego County so far. Eight more influenza-related deaths last week brought the local death toll to 20. One researcher at San Diego State University has found a way to use Twitter to identify outbreak zones and he believes it has the potential to save lives.

"We are in the right time, in the right moment, doing the right things," said SDSSU geography professor Ming-Hsiang Tsou.

If you have ever tweeted about feeling under the weather in the last few years, chances are Ming has seen it. His project began in 2010 with a $1.3 million grant from the National Science Foundation. Since then, his data has been astounding -- so accurate, it is submitted to the Centers for Disease Control and Prevention (CDC) every two weeks.

Ming identifies keywords that pop up in tweets, often along with the location of those tweets. For example, keywords like "cold," "sneeze" or "sore throat." When the raw numbers from Twitter are crunched, they can reveal areas where outbreaks of the flu are underway or, in some cases, about to occur.

This is what Ming said excites him about this project. He monitors 31 major cities and his data is faster than traditional reporting methods. If health officials can be clued in to where an outbreak is about to occur, more resources can be funneled to that area before the disease can proliferate.

"By using Twitter to predict the flu outbreak, we will be able to identify potential outbreaks 10 days before the official record from the CDC," explained Ming. "With this information, hospitals and clinics can prepare more -- arm themselves with vaccines -- and save hundreds of people's lives."

The CDC considers the flu season to begin in October and last through May, peaking around February. The existing standard to report the flu is an inefficient one -- health professionals track incoming patients that have flu-like symptoms, record that data and submit it to regional health agencies. From there, it must make its way to the CDC.

Ming said this process takes at least two weeks.

The instantaneous nature of Twitter gives his research the advantage here, as his numbers come out 10 days before the CDC. Additionally, his Twitter data is proving to be very comparable to the official CDC data.

"I was a little bit shocked at how close we are," said Ming.

Ming's data has revealed several interesting patterns. In the last two years, the flu had its first outbreak early in San Diego. Christmas shows a huge spike in the flu here at home.

His data from 2012 also shows a fascinating movement of disease. After Superstorm Sandy hit, flu exploded on the East Coast. From there, it moved westward. Using Twitter, Ming hypothesizes it was the close quarters in the storm shelters that caused the outbreak.

"You see the idea of this spread, of this pattern in ways that we were never able to look at before," said Brian Spitzberg, a Senate Distinguished Professor of Communication at SDSU. "We're not necessarily talking about 'Patient Zero,' but people who have the flu are connected to other people. And sometimes those connections can travel airplane routes, travel routes, job routes."

If tracking your every word in your every tweet at every location seems a little "Orwellian," Ming said it's true. Emerging research like this raises new questions about privacy. Ming believes, in this case, it is for a good cause.

"In Twitter, we are looking through mountains of garbage to find nuggets of gold," said Ming. "We can collect not just one or two people but millions of people, or billions of messages together. We're really able to use social media to monitor human behavior -- that's so exciting."

Print this article Back to Top