Researchers Using Twitter Extend Syndromic Surveillance Beyond Flu Tracking

2-11-2013 11-00-24 PM

Researchers at the University of Rochester expand on earlier success tracking the spread of flu with Twitter by adding data elements and correlating that data to overall population health.

Using tweets collected in New York City over a month, researchers have been exploring factors such as how often a person goes to the gym, visits a particular a bar or restaurant, or takes public transportation. They are determining which of 70 data elements are associated with a positive, negative, or neutral impact on the each user’s health.

By looking deeper into the aggregate data available on Twitter and matching it up with location tags, researchers have been able to pinpoint real-time hotspots within cities where sick people appear to be convening.

The analytics engine is built on machine learning principles. Algorithms are continually run and fine tuned to teach the machine how to differentiate between whether a tweet is or is not from a person who is actually sick. As new data elements are added to the algorithm, additional insight can be extracted.

"It’s like teaching a baby a new language. We need the algorithm to understand that someone who tweets ‘I’m sick and have been in bed all day’ should be characterized as sick, but ‘I’m sick of driving around in this traffic’ shouldn’t be." – Adam Sadilek, postdoctoral researcher at the University of Rochester

After succeeding with initial efforts to track flu, researchers are now working with faculty in the Department of Psychiatry and the School of Nursing to develop the capacity to measure and understand factors that impact depression and other psychological disorders.

↑ Back to top

Founding Sponsors

Platinum Sponsors