Influenza-like illness (ILI) remains a significant public health burden to both the general public and the U.S. Department of Defense. Military personnel are especially susceptible to disease outbreaks owing to the often-crowded living quarters, substantial geographic movement, and physical stress placed upon them. Currently, the military employs syndromic surveillance on electronic reporting of clinical diagnoses. While faster than traditional, biologically-focused monitoring techniques, the military surveillance system proved inadequate at detecting outbreaks quickly enough in a recent study conducted by the CDC. Recently, research has included novel data sources, like social media, to conduct disease detection in real-time and capture communities not traditionally accounted for in current surveillance systems. Data-mining techniques are used to identify influenza-related social media posts and train a model against validated medical data. By integrating social media data and a medical dataset of all ILI-related laboratory specimens and doctor visits for the entire military cohort, a more comprehensive model than presently exists for disease identification and transmission will be possible.
Objective
To integrate existing influenza surveillance data sources and social media data into an accurate and timely outbreak detection model embedded into dashboard biosuveillance analytics for the Department of Defense.