Applying topic modeling to investigate social media discussions on prescription opioids in the context of urbanization

Authors: Moying Li*, University of Maryland, Kathleen Stewart, University of Maryland
Topics: Medical and Health Geography, Spatial Analysis & Modeling, United States
Keywords: Prescription opioids, social media, topic modeling, urbanization
Session Type: Paper
Day: 4/10/2018
Start / End Time: 4:40 PM / 6:20 PM
Room: Grand Ballroom C, Astor, 2nd Floor
Presentation File: No File Uploaded

The nonmedical use of prescription opioids is a national epidemic across the U.S., with the US Centers for Disease Control and Prevention reporting that deaths attributable to prescription opioids have increased more than fourfold since 1999. As this national health crisis continues to gain public attention, there’s an increasing need for better identifying underlying misuse behaviors, trends, and risk factors in order to optimize efforts at improving access to treatment, preventing overdoses, and ensuring community interventions are effective. Understanding the impact of place on health is a key element of epidemiologic investigations, and numerous tools are being employed for the analysis of spatial health-related data. In this study, we analyze the geographical patterns of social media discussions that refer to prescription opioids. We filter geo-located tweets collected from the Twitter public application program interface stream for commonly misused prescription opioid drug keywords. We then apply unsupervised machine learning using Single Topic Latent Dirichlet Allocation (ST-LDA) models to identify topics associated with the mention of prescription opioids. Spatial and statistical methods are applied to reveal the differences or similarities between topics in urban and rural counties from the states along the east coast that intersect the I-95 as well as Ohio and West Virginia, i.e., states that have high age-adjusted prescription opioids poisoning death rates. The results will provide insights into relationships between references about prescription opioids that could be useful in understanding risk factors, especially in the context of urbanization.

Abstract Information

This abstract is already part of a session. View the session here.

To access contact information login