Social Media Mining
Social media is a source of massive amounts of information. Social media is accessible by large populations and it contains large amounts of health-related chatter posted by users themselves. Therefore, social media often contains health-related information that may not be available from any other sources. Our social media mining research covers a range of topics including population health and individual health.
- Build end-to-end NLP pipelines for converting noisy social media text into valuable and actionable knowledge that can be used by domain experts.
- Deriving population-level knowledge regarding topics of interest such as prescription drug use and misuse, drug effectiveness and adverse reactions, drug reliance and addiction, medication assisted treatment for addiction, assessing people’s perception regarding certain drugs, and many other drug/medication-related studies.
- Studying the effectiveness of health service providers and programs such as Medicaid and Medicare.
- Performing longitudinal analysis of user-posted information to study long-term behavioral patterns and associations.
- Unobstrusively studying mental health-related information, including depression, stress, anxiety and loneliness.
We currently have two active funded projects.
Social Media Mining for Toxicovigilance
Our work on prescription medication use and misuse is funded by the NIH/NIDA. We are trying to build the NLP and computational methods that can make use of social media big data to predict future drug-related crises (such as the opioid crisis), study the current state of prescription drug related problems, study the natural histories of individuals suffering from substance use disorder and mining information that are useful to toxicologists who are assisting people with substance use disorder on a daily basis.
Our publications related to the project are available from the NIH: HERE.
Social Media Mining for Studying Medicaid Related Information
We are working, in collaboration with the Oregon Health & Science University and the University of Pennsylvania, to study Medicaid-related information from Twitter. We have two primary objectives: (i) to understand how Medicaid agencies and managed care organizations (MCOs) are using Twitter to provide services, and (ii) to study user perceptions about Medicaid-related services.