# Exploring COVID-19 Related Stressors Using Topic Modeling

**Authors:** Yue Tong Leung, Farzad Khalvati  
**arXiv:** [2202.00476](https://arxiv.org/abs/2202.00476)  
**Date:** 2022-01-12  
**Summary page:** https://www.summarizepaper.com/en/paper/2202.00476v1/

## Key points

- COVID-19 pandemic has led to changes in lifestyles and potential mental health problems
- Healthcare organizations need to identify specific stressors related to COVID-19
- Study uses natural language processing (NLP) techniques to analyze social media data from subreddit rCOVID19_support
- Dataset of 9266 Reddit posts from February 14, 2020, to July 19, 2021
- Latent Dirichlet Allocation (LDA) used for topic modeling along with lexicon methods
- Trends of topics' prevalence during different stages of the pandemic visualized using monthly sum of LDA model output
- Comparison made with development of the pandemic using data on total cases, new cases per day, and vaccinated population from Our World Data for United States, United Kingdom, and Canada
- Word clouds created for each topic group to visualize most frequently mentioned words within each topic
- Study provides insights into prevalence of COVID-19-related stressors at different stages of the pandemic using NLP techniques applied to social media data
- Findings can inform healthcare organizations about specific stressors individuals are facing during different phases of COVID-19
- Previous research has also utilized NLP techniques to analyze COVID-19-related discussions on social media platforms
- Studies have provided insights into public reactions, concerns, and changes in mental health support groups during the pandemic

## Summary

The COVID-19 pandemic has had a significant impact on people's lives worldwide for nearly two years, leading to changes in lifestyles that can result in psychosocial stressors and potential mental health problems. To provide effective mental health support, healthcare organizations need to identify specific stressors related to COVID-19 and understand the prevalence of these stressors over time. This study aims to address this by applying natural language processing (NLP) techniques to analyze social media data from the subreddit \rCOVID19_support. The researchers obtained a dataset of 9266 Reddit posts from February 14, 2020, to July 19, 2021. They used Latent Dirichlet Allocation (LDA), a topic modeling technique, along with lexicon methods to identify the topics discussed on the subreddit. To visualize the trend of these topics' prevalence during different stages of the pandemic, the researchers computed the monthly sum of LDA model output for each topic group. They compared these trends with the development of the pandemic by obtaining data on total cases, new cases per day, and vaccinated population from Our World Data for the United States, United Kingdom, and Canada. Additionally, word clouds were created for each topic group to provide a visual representation of the most frequently mentioned words within each topic. These word clouds helped further understand and interpret the identified topics. This study contributes valuable insights into the prevalence of COVID-19-related stressors at different stages of the pandemic using NLP techniques applied to social media data. The findings can inform healthcare organizations about specific stressors individuals are facing during different phases of COVID-19. Moreover, this research demonstrates how NLP techniques can be leveraged in future studies to analyze event-specific stressors. Previous research has also utilized NLP techniques to analyze COVID-19-related discussions on social media platforms. Some studies have focused on identifying topics discussed in tweets and subreddits related to COVID-19 while others have examined the mental health impacts of the pandemic using NLP techniques on Reddit data. These studies have provided valuable insights into public reactions, concerns and changes in mental health support groups during the pandemic. In summary, this study highlights the importance of understanding psychosocial stressors during the COVID-19 pandemic by analyzing social media data. The use of NLP techniques such as LDA topic modeling and lexicon methods allows for a comprehensive analysis of topics discussed on online platforms. The visualization of topic trends provides a clear overview of prevalent stressors at different stages of the pandemic.