Despite numerous successes in using social media to detect food borne illness and to predict influenza trends, the use of social media as a public health tool has yet to gain widespread adoption. While social media data cannot directly diagnose illness, aggregate trends in symptom proliferation may readily be observed. Such trends may allow a health agency to watch for signs and symptoms related to target conditions within its jurisdiction. Further, social media surveillance offers a distinct advantage in immediacy and sensitivity as it is not dependent upon infected individuals seeking care for reportable illnesses and as such information is not delayed by the handling, transfer, and processing of reports. These advantages may enable the earlier preparation and initiation of scaled response sequences during public health emergencies. Such data may also yield additional evidence through shared symptoms, rumors, and observations crucial to an epidemiological investigation.
Objective
To formally introduce ChatterGrabber, an open source, natural language processing based toolset for public health social media surveillance. ChatterGrabber is designed to collect and categorize a high volume of content at a low cost, providing a readily deployable solution for Epidemiologists to track emergent outbreaks in the field and a signal for syndromic surveillance.