Automated tools for systematic review screening methods: an application of machine learning for sexual orientation and gender identity measurement in health research
Bibliographic record
Abstract
Objective: Sexual and gender minority (SGM) populations experience health disparities compared to heterosexual and cisgender populations. The development of accurate, comprehensive sexual orientation and gender identity (SOGI) measures is fundamental to quantify and address SGM disparities, which first requires identifying SOGI-related research. As part of a larger project reviewing and synthesizing how SOGI has been assessed within the health literature, we provide an example of the application of automated tools for systematic reviews to the area of SOGI measurement. Methods: In collaboration with research librarians, a three-phase approach was used to prioritize screening for a set of 11,441 SOGI measurement studies published since 2012. In Phase 1, search results were stratified into two groups (title with vs. without measurement-related terms); titles with measurement-related terms were manually screened. In Phase 2, supervised clustering using DoCTER software was used to sort the remaining studies based on relevance. In Phase 3, supervised machine learning using DoCTER was used to further identify which studies deemed low relevance in Phase 2 should be prioritized for manual screening. Results: 1,607 studies were identified in Phase 1. Across Phases 2 and 3, the research team excluded 5,056 of the remaining 9,834 studies using DoCTER. In manual review, the percentage of relevant studies in results screened manually was low, ranging from 0.1 to 7.8 percent. Conclusions: Automated tools used in collaboration with research librarians have the potential to save hundreds of hours of human labor in large-scale systematic reviews of SGM health research.
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 distilled prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. Learned from the 10,348 direct Codex labels and 10,348 direct Gemma labels. Candidate is the union of thresholded teacher heads; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels or direct frontier model labels.
Codex and Gemma teacher scores by category
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.034 | 0.029 |
| Meta-epidemiology (narrow) | 0.000 | 0.000 |
| Meta-epidemiology (broad) | 0.001 | 0.000 |
| Bibliometrics | 0.000 | 0.001 |
| Science and technology studies | 0.000 | 0.000 |
| Scholarly communication | 0.000 | 0.000 |
| Open science | 0.000 | 0.000 |
| Research integrity | 0.000 | 0.000 |
| Insufficient payload (model declined to judge) | 0.000 | 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; both teacher heads agree on what is shown here.
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".