Regional Level Influenza Study with Geo-Tagged Twitter Data

Area of Science:

• Public Health
• Data Science
• Epidemiology

Background:

• Social media generates vast, real-time data reflecting real-world events.
• Twitter data has been utilized for diverse applications, including public health trend analysis.
• Traditional public health surveillance can be slow to detect emerging outbreaks.

Purpose of the Study:

• To design, implement, and evaluate a system for real-time collection and analysis of influenza data from Twitter streams.
• To investigate the correlation between Twitter-derived influenza counts and official statistics from the CDC.
• To assess the potential of Twitter data as an early warning system for influenza epidemics.

Main Methods:

• Development of a prototype system to process real-time Twitter streams for influenza-related keywords.
• Geographical analysis of tweet data to monitor influenza prevalence across different regions.
• Statistical correlation analysis comparing Twitter flu counts with Center for Disease Control and Prevention (CDC) data.
• Proposal and validation of a dynamic mathematical model for forecasting Twitter flu counts.

Main Results:

• Real-time Twitter flu counts demonstrated a strong correlation with official CDC influenza statistics.
• The system successfully captured influenza dynamics at both national and regional levels.
• Twitter data showed potential as a timely indicator for influenza epidemics, often preceding official reports.
• The proposed dynamic mathematical model achieved high accuracy in forecasting Twitter flu counts.

Conclusions:

• Real-time social media data, particularly from Twitter, can accurately reflect real-world public health trends like influenza.
• Twitter flu surveillance can serve as a valuable, rapid early warning system for influenza epidemics.
• Advanced mathematical modeling can enhance the predictive power of social media data for public health surveillance.