Diabetes topics associated with engagement on Twitter

Area of Science:

• Public Health
• Social Media Analysis
• Health Communication

Background:

• Social media platforms are increasingly utilized by the public and health professionals.
• Engagement with health information on social media may influence health behaviors.
• Analyzing the vast amount of social media data presents a significant challenge for researchers.

Purpose of the Study:

• To investigate topics associated with engagement with diabetes information on Twitter.
• To assess the utility of crowdsourcing for analyzing social media health data.

Main Methods:

• A random sample of 100 tweets containing '#diabetes' was collected daily over a week.
• Crowdsourcing via Amazon's Mechanical Turk was employed to categorize tweets and user types.
• Tweedie regression was used to identify characteristics associated with tweet engagement (favoriting and retweeting).

Main Results:

• Tweet topic and user type classification demonstrated reliability.
• Tweets discussing diabetes-related health problems showed a positive association with engagement.
• Tweets on diabetes prevalence, nonmedical resources, and jokes/sarcasm were negatively associated with engagement.

Conclusions:

• Crowdsourcing is a viable, efficient, and cost-effective method for classifying social media data.
• Public health practitioners can leverage these findings to enhance engagement strategies around diabetes by focusing on relevant health topics.