Masters Thesis

Big data analysis for user health tracking

Data analysis of social media postings can provide a wealth of information about the health of individual users, health across groups, and even access to healthy food choices in neighborhoods. In this thesis, I analyze the messages posted in Twitter that are of 140 characters or less, known as tweets, to infer user health status over time. Tweets and in turn their users' health are scored according to semantic analysis, sentiment analysis, emoticon classification, meta-data analysis, and profiling over time.The purpose of the analysis includes individually targeted healthcare personalization, determining health disparities, discovering health access limitations, advertising, and public health monitoring. The approach is analyzed on over 12,000 tweets spanning as far back as 2010 for 10 classes of users active on Twitter. With this user health scores classification approach, we can say approximately 65%-70% of the users can be accurately classified as healthy or unhealthy.

Items in ScholarWorks are protected by copyright, with all rights reserved, unless otherwise indicated.