What Matters the Most to Early- and Late-Stage Patients with Lung Cancer? A Dynamic Topic-Modeling Analysis of Big Data in Online Cancer Communities
Presentations
PO-GePV-M-29 (Sunday, 7/10/2022) [Eastern Time (GMT-4)]
ePoster Forums
Purpose: Our aim is to examine the frequency, patterns, and dynamics of conversation topics regarding beliefs, and thoughts of the patients and their caregivers in popular online lung cancer forums. Further, we have to determine the likelihoods of machine learning- and dynamic topic modeling–based classifiers to categorize text-based public opinion data.
Methods: We extracted datasets based on a collection of keywords associated with lung cancer from 3 different online lung cancer discussion forums: lungcancer.net, lungevity.org, and reddit.com that were posted between June 2016 and June 2017. A snowball sampling approach was employed to generate the list of keywords and process was repeated until no new keywords were found. This dataset provided us with enough samples to compare the impacts (e.g., post volume changes and trends in discussion topics) in early and late-stage lung cancer patients. After cleaning the raw data that include 21,998 cancer-related forum posts, the dynamic topic modeling was performed using the preprocessed data.
Results: According to our findings, cancer communication on lung cancer forums has primarily focused on the following six topics for both stages: mental health, insurance and disability, social support, cancer treatment, cancer prevention, and other topics. Discussions about social support and cancer treatment drew the most users’ attention and generated the most comments. Online lung cancer forum posts displaying public behaviors were collected and analyzed using a machine learning classifier with good accuracy (0.90), precision (0.91), and recall (0.92).
Conclusion: This study of cancer-related discussion across different stages on online health forums provides important insights into public personal experience about lung cancer and supports health campaigns on online health forums despite potential privacy and security issues. The findings in this work are intriguing and highlight the need for additional in-depth research into how lung cancer patients express themselves on online health forums.
Keywords
Not Applicable / None Entered.
Taxonomy
Not Applicable / None Entered.
Contact Email