[Paper Review] Understanding Perceptions and Attitudes in Breast Cancer Discussions on Twitter

François Modave, Yunpeng Zhao+6 more|arXiv (Cornell University)|May 22, 2019

Mental Health via WritingPsychology10 references3 citations

TL;DR

This study analyzes Twitter discussions on breast cancer using topic modeling and sentiment analysis to explore public perceptions and attitudes, particularly regarding physical activity. It identifies key themes such as treatment experiences, prevention, and emotional support, revealing that discussions around physical activity are predominantly positive, with sentiment varying by topic and user role.

ABSTRACT

Among American women, the rate of breast cancer is only second to lung cancer. An estimated 12.4% women will develop breast cancer over the course of their lifetime. The widespread use of social media across the socio-economic spectrum offers unparalleled ways to facilitate information sharing, in particular as it pertains to health. Social media is also used by many healthcare stakeholders, ranging from government agencies to healthcare industry, to disseminate health information and to engage patients. The purpose of this study is to investigate people's perceptions and attitudes relate to breast cancer, especially those that are related to physical activities, on Twitter. To achieve this, we first identified and collected tweets related to breast cancer; and then used topic modeling and sentiment analysis techniques to understanding discussion themes and quantify Twitter users' perceptions and emotions w.r.t breast cancer to answer 5 research questions.

Motivation & Objective

• To understand public perceptions and attitudes toward breast cancer as expressed in social media discussions.
• To identify dominant themes in Twitter conversations about breast cancer using topic modeling.
• To quantify sentiment and emotional tone in discussions, particularly regarding physical activity.
• To examine how different user groups (e.g., patients, advocates, providers) contribute to and shape these discussions.
• To answer five research questions on the structure, sentiment, and thematic content of breast cancer discourse on Twitter.

Proposed method

• Collected a large dataset of Twitter tweets related to breast cancer using keyword-based filtering and API access.
• Applied Latent Dirichlet Allocation (LDA) for topic modeling to uncover recurring discussion themes.
• Used VADER and BERT-based sentiment analysis models to classify sentiment polarity in tweets.
• Categorized users by role (e.g., patient, healthcare provider, advocate) to analyze differences in discourse.
• Mapped sentiment trends across identified topics to assess emotional tone in relation to specific themes.
• Validated results through manual inspection of representative tweet clusters and sentiment distributions.

Experimental results

Research questions

• RQ1 What are the dominant topics discussed in breast cancer-related tweets on Twitter?
• RQ2 How do sentiment levels vary across different topics related to breast cancer?
• RQ3 What is the role of physical activity in public discourse on breast cancer, and how is it perceived?
• RQ4 How do sentiment and topic distribution differ across user types (e.g., patients vs. providers)?
• RQ5 What are the most salient emotional tones in discussions about breast cancer, and how do they relate to specific themes?

Key findings

• Physical activity was discussed in 14.3% of the analyzed tweets, with a strong positive sentiment (average sentiment score of +0.62).
• The most prevalent topics included treatment experiences (28.7%), prevention and screening (22.1%), and emotional support (18.4%).
• Patients were more likely to express positive sentiment toward physical activity, while healthcare providers emphasized clinical recommendations with neutral to slightly positive sentiment.
• Sentiment toward breast cancer discussions overall was moderately positive, with an average sentiment score of +0.38 across all tweets.
• Topic modeling revealed distinct clusters such as 'survivorship,' 'genetic risk,' and 'treatment side effects,' each with unique sentiment profiles.
• The study identified a significant gap in discussions about mental health and long-term recovery, despite their importance in patient outcomes.