A Machine Learning Model for Identifying Sexual Health Influencers to Promote the Secondary Distribution of HIV Self-Testing Among Gay, Bisexual, and Other Men Who Have Sex With Men in China: Quasi-Experimental Study
Bibliographic record
Abstract
BACKGROUND: Sexual health influencers (SHIs) are individuals actively sharing sexual health information with their peers, and they play an important role in promoting HIV care services, including the secondary distribution of HIV self-testing (SD-HIVST). Previous studies used a 6-item empirical leadership scale to identify SHIs. However, this approach may be biased as it does not consider individuals' social networks. OBJECTIVE: This study used a quasi-experimental study design to evaluate how well a newly developed machine learning (ML) model identifies SHIs in promoting SD-HIVST compared to SHIs identified by a scale whose validity had been tested before. METHODS: We recruited participants from BlueD, the largest social networking app for gay men in China. Based on their responses to the baseline survey, the ML model and scale were used to identify SHIs, respectively. This study consisted of 2 rounds, differing in the upper limit of the number of HIVST kits and peer-referral links that SHIs could order and distribute (first round ≤5 and second round ≤10). Consented SHIs could order multiple HIV self-testing (HIVST) kits and generate personalized peer-referral links through a web-based platform managed by a partnered gay-friendly community-based organization. SHIs were encouraged to share additional kits and peer-referral links with their social contacts (defined as "alters"). SHIs would receive US $3 incentives when their corresponding alters uploaded valid photographic testing results to the same platform. Our primary outcomes included (1) the number of alters who conducted HIVST in each group and (2) the number of newly tested alters who conducted HIVST in each. We used negative binomial regression to examine group differences during the first round (February-June 2021), the second round (June-November 2021), and the combined first and second rounds, respectively. RESULTS: In January 2021, a total of 1828 men who have sex with men (MSM) completed the survey. Overall, 393 SHIs (scale=195 and ML model=198) agreed to participate in SD-HIVST. Among them, 229 SHIs (scale=116 and ML model=113) ordered HIVST on the web. Compared with the scale group, SHIs in the ML model group motivated more alters to conduct HIVST (mean difference [MD] 0.88, 95% CI 0.02-2.22; adjusted incidence risk ratio [aIRR] 1.77, 95% CI 1.07-2.95) when we combined the first and second rounds. Although the mean number of newly tested alters was slightly higher in the ML model group than in the scale group, the group difference was insignificant (MD 0.35, 95% CI -0.17 to -0.99; aIRR 1.49, 95% CI 0.74-3.02). CONCLUSIONS: Among Chinese MSM, SHIs identified by the ML model can motivate more individuals to conduct HIVST than those identified by the scale. Future research can focus on how to adapt the ML model to encourage newly tested individuals to conduct HIVST. TRIAL REGISTRATION: Chinese Clinical Trials Registry ChiCTR2000039632; https://www.chictr.org.cn/showprojEN.html?proj=63068. INTERNATIONAL REGISTERED REPORT IDENTIFIER (IRRID): RR2-10.1186/s12889-021-11817-2.
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 | 0.012 | 0.017 |
| Meta-epidemiology (narrow) | 0.001 | 0.000 |
| Meta-epidemiology (broad) | 0.001 | 0.002 |
| Bibliometrics | 0.001 | 0.000 |
| Science and technology studies | 0.001 | 0.001 |
| Scholarly communication | 0.001 | 0.001 |
| Open science | 0.001 | 0.001 |
| Research integrity | 0.001 | 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".