Quality Analysis of Stroke-Related Videos on Video Platforms: Cross-Sectional Study
Bibliographic record
Abstract
Background: Stroke has become a global public health problem due to its high incidence, disability, and mortality. In China, TikTok and Bilibili, as mainstream video-sharing platforms, serve as key sources of getting stroke-related information for people, yet their videos' content quality and reliability remain insufficiently evaluated. Objective: This cross-sectional study aimed to analyze the content and quality of stroke-related videos on Chinese video-sharing platforms. Methods: In March 2025, stroke-related videos were retrieved from TikTok and Bilibili using the search term "" (Chinese for stroke). Eligible videos were analyzed for metadata and content indicators. Researchers assessed video quality using validated tools: the Global Quality Scale (GQS), modified DISCERN (mDISCERN), and Patient Education Materials Assessment Tool (PEMAT). Statistical analyses were performed with Python, including descriptive statistics, group comparisons (Kruskal-Wallis tests), and Spearman's rank correlation to evaluate variable associations, with all P values adjusted for multiple comparisons using the Bonferroni method. A binary classification predictive model was developed using the random forest algorithm, accompanied by feature importance analysis. Results: Among the stroke-related videos from Bilibili (n=157) and TikTok (n=149), popular science education content predominated (204/306, 66.7%). Bilibili videos were primarily categorized as professional lectures (83/157, 52.9%), while most TikTok videos were popular science education (139/149, 93.3%). TikTok videos demonstrated significantly higher median likes and comments (P<.001) and shorter durations compared to Bilibili (P<.001). No significant differences were observed in median GQS (4) or mDISCERN scores (3) between platforms (P>.05). Videos produced by professional teams exhibited significantly higher GQS and PEMAT-A/V scores than those created by independent content creators (P<.05). Popular science education videos achieved higher PEMAT-A/V actionability scores than professional lectures (P<.001), while videos addressing treatment options scored lowest in GQS (P<.05). Strong positive correlations were identified among user engagement parameters (likes, shares, comments; ρ=0.81-0.90, P<.001), but only weak correlations were found between engagement and quality scores (ρ<0.3). Machine learning modeling (AUC=0.58) identified video duration (importance score: 0.15) and uploader subscriber count (importance score: 0.13) as key predictors of content quality. Conclusions: The quality of stroke-related videos on TikTok and Bilibili remains suboptimal. Content uploaded by certified physicians and institutions received higher GQS/mDISCERN scores, confirming that medical authority is a key quality indicator. Our exploratory random-forest model, which used only basic metadata (duration, likes, subscriber count), achieved an area under the curve of 0.58, indicating that surface engagement metrics alone are insufficient to discriminate high- from low-quality material. Consequently, future screening algorithms should incorporate content-based features (eg, captions, medical keywords, visual cues) and creator credentials rather than relying solely on readily available interaction parameters. Multi-platform, larger-scale datasets are warranted to develop clinically useful prediction tools.
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 | 0.007 |
| Meta-epidemiology (narrow) | 0.000 | 0.000 |
| Meta-epidemiology (broad) | 0.000 | 0.001 |
| Bibliometrics | 0.002 | 0.002 |
| Science and technology studies | 0.001 | 0.000 |
| Scholarly communication | 0.001 | 0.001 |
| Open science | 0.000 | 0.001 |
| Research integrity | 0.000 | 0.000 |
| Insufficient payload (model declined to judge) | 0.002 | 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".