Using Twitter for diabetes community analysis

Abstract

Social media platforms have become a common venue for sharing experiences and knowledge about health-related topics. This research focuses on examining social media-based communication patterns related to diabetes on the Twitter platform. Specifically, we apply an updated methodology to examine changes in the current use of hash-tags, trending hash-tags, and the frequency of diabetes-related tweets using a previous study as a baseline. Our results show significant growth in the diabetes community on Twitter over time and also evidence that this community is increasing in its capacity to spread awareness regarding diabetes- related health topics. Our methodological contributions include an improved framework for collecting, cleaning and analyzing Twitter data related to diabetes as well as the application of regular expressions to categorize subsets of tweets. We have also developed a model based on word-embedding and long short term memory to identify tweets of diabetic patients.