A Topological Data Analysis Approach to Visualizing Ebola Tweets

Abstract

 The role of social media in health and medical information in general, and during an epidemic in particular, has been reported. Data from social media such as tweets provide vast opportunities and potential benefits for health and medical information communication. However, openly-available social media network visualization tools focus on the online connections between social media users which may not be of utmost importance for health and medical information practitioners. This paper takes the topological data analysis (TDA) approach to render a visual representation of these large, unstructured, and highly complex data. Utilizing a TDA and machine learning platform, distinct features of these Ebola tweets were visually and statistically identified.

 Topological locations of relevant keywords (virus, epidemic, Africa, Sierra Leone, blood, saliva, fever, misinformation, etc.) contained in these tweets, as well as the topological locations of news and health-related organizations are presented.