Sensitivity and specificity estimates for each transparency detector, used by [rt_summary()] to correct an apparent prevalence for detector error (the Rogan-Gladen correction).
Format
A tibble with 8 rows and 5 columns:
- variable
Indicator column name, as returned by [rt_all_pmc()].
- label
Human-readable indicator name.
- sensitivity
Detector sensitivity (true-positive rate), 0-1.
- specificity
Detector specificity (true-negative rate), 0-1.
- source
Where the estimate comes from.
Source
Serghiou S, Contopoulos-Ioannidis DG, Boyack KW, Riedel N, Wallach JD, Ioannidis JPA (2021). Assessment of transparency indicators across the biomedical literature: How open is open? PLOS Biology 19(3): e3001107. doi:10.1371/journal.pbio.3001107 . Data and code values: this package's reproducible benchmark and regression estimates (`inst/benchmark/results_data_code.md`).
Details
For conflicts of interest, funding and protocol registration these are the published, importance-weighted validation values of Serghiou et al. (2021); the detectors for these indicators are essentially those validated in the paper. For data and code sharing the detector is implemented natively in this package (it no longer wraps `oddpub`), so the package's reproducible benchmark and regression estimates are used instead (see `inst/benchmark`). These data/code estimates are not an untouched external validation of the native detector; supply your own values to [rt_summary()] via its `accuracy` argument when you have study-specific or externally validated estimates. Novelty's estimate comes from a maintainer-built hand-labeled gold set (see `inst/benchmark/results_novelty_replication.md`). Replication's sensitivity comes from a 111-positive replication-enriched validation (see `inst/benchmark/results_replication_enriched.md`), with the specificity from the 2023 1000-article sample. AI-use disclosure is not included (its prevalence is too low in unselected literature for a stable estimate), so [rt_summary()] reports it uncorrected.