Text Mining of Reddit Data to Explore Priorities and Experiences of Individuals with Cervical Spinal Cord Injury

Research Objectives: To detect topics reflecting patient priorities and experiences in cervical spinal cord injury (CSCI) using text mining of online consumer-generated content; and compare these topics with patient priorities identified from prior research. We hypothesize that text-mining-identified topics will converge with findings from prior work with an emphasis on upper extremity function.

Design: Using Python’s Reddit Application Programming Interface (API) Wrapper (PRAW), we extracted relevant posts and their associated comments from SCI-focused subreddits, filtering content by relevant keywords (e.g., cervical) and user characteristics. After preprocessing to eliminate duplicates and irrelevant text, we applied topic modeling using Latent Dirichlet Allocation (LDA), an unsupervised machine learning approach, to identify topics within the text.

Setting: Four SCI-focused forums on Reddit, an online platform known for anonymity and publicly available data, with over 9,000 combined subscribers.

Participants: A total of 2,148 anonymous Reddit users who contributed content related to CSCI in specified subreddits.

Interventions: Not applicable.

Main Outcome Measures: The identification of dominant topics or themes in discussions related to CSCI, which reflect users’ priorities and experiences.

Results: We analyzed a total of 12,855 text documents from users, including posts and comments. The LDA model identified 11 distinct topics: (1) mobility devices and assistance; (2) pain and spasticity management, and medication; (3) workplace support systems and logistics; (4) pressure wound management; (5) peer support; (6) gaming, leisure, and sexual health; (7) pain, physical sensation, and mobility; (8) time, daily life, and retrospection; (9) bladder and bowel management, and lifestyle changes; (10) assistive technology; (11) injury etiology, characteristics, and surgical interventions. Topics 1, 6, and 10 prominently featured terms associated with the upper extremity, such as ‘hand’ and ‘finger,’ which ranked among the top 15 words in these topic clusters. In addition, many of the topics included keyword collections that reflected previously reported patient priorities for improving quality of life post-injury, such as managing pain and other sensations, bladder and bowel function, and sexual health.

Conclusions: Discussion of upper extremity function and other key functional domains is central to several topic clusters, supporting our hypothesis. Our findings highlight the feasibility of using consumer-generated content and data-driven methods to capture the lived experiences and priorities of individuals with CSCI. Moreover, text mining has the potential to detect underrepresented priorities, potentially informing new patient-reported outcome measures and enhancing support services to address diverse patient concerns and improve quality of life.

Author(s) Disclosures: None.