Mapping Individual Differences on the Internet: Case Study of the Type 1 Diabetes Community
Bibliographic record
Abstract
BACKGROUND: Social media platforms, such as Twitter, are increasingly popular among communities of people with chronic conditions, including those with type 1 diabetes (T1D). There is some evidence that social media confers emotional and health-related benefits to people with T1D, including emotional support and practical information regarding health maintenance. Research on social media has primarily relied on self-reports of web-based behavior and qualitative assessment of web-based content, which can be expensive and time-consuming. Meanwhile, recent advances in natural language processing have allowed for large-scale assessment of social media behavior. OBJECTIVE: This study attempts to document the major themes of Twitter posts using a natural language processing method to identify topics of interest in the T1D web-based community. We also seek to map social relations on Twitter as they relate to these topics of interest, to determine whether Twitter users in the T1D community post in "echo chambers," which reflect their own topics back to them, or whether users typically see a mix of topics on the internet. METHODS: Through Twitter scraping, we gathered a data set of 691,691 tweets from 8557 accounts, spanning a date range from 2008 to 2020, which includes people with T1D, their caregivers, health practitioners, and advocates. Tweet content was analyzed for sentiment and topic, using Latent Dirichlet Allocation. We used social network analysis to examine the degree to which identified topics are siloed within specific groups or disseminated through the broader T1D web-based community. RESULTS: Tweets were, on average, positive in sentiment. Through topic modeling, we identified 6 broad-bandwidth topics, ranging from clinical to advocacy to daily management to emotional health, which can inform researchers and practitioners interested in the needs of people with T1D. These analyses also replicate prior work using machine learning methods to map social behavior on the internet. We extend these results through social network analysis, indicating that users are likely to see a mix of these topics discussed by the accounts they follow. CONCLUSIONS: Twitter communities are sources of information for people with T1D and members related to that community. Topics identified reveal key concerns of the T1D community and may be useful to practitioners and researchers alike. The methods used are efficient (low cost) while providing researchers with enormous amounts of data. We provide code to facilitate the use of these methods with other populations.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
How this classification was reachedexpand
Full frame machine prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. The Gemma side is a direct model label for every work in the frame, read from the title-only record. The Codex side is a classifier learned from the 10,348 direct Codex labels and calibrated to design-weighted sample rates; fields without enough sample support carry no Codex call. Candidate is the union of the two sides; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels.
Distilled classifier scores by category (both heads)
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.002 | 0.010 |
| Meta-epidemiology (narrow) | 0.000 | 0.000 |
| Meta-epidemiology (broad) | 0.000 | 0.000 |
| Bibliometrics | 0.003 | 0.004 |
| Science and technology studies | 0.006 | 0.001 |
| Scholarly communication | 0.002 | 0.003 |
| Open science | 0.001 | 0.004 |
| Research integrity | 0.002 | 0.002 |
| Insufficient payload (model declined to judge) | 0.003 | 0.000 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from itClassification
machine, unvalidatedMachine predicted; a candidate call from one source (direct Gemma or distilled Codex), not a consensus.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".