Compares realist extraction, baseline strengthening, and classification floor on healthcare incident reports, indicating potential improvements in reporting accuracy.
Frozen preregistration for experiment E-02 of the MARS+SPIRA research programme: a paired, n=400, three-arm comparison (realist/CMO extraction; strengthened generic baseline; classification floor) on public healthcare incident reports across four genres, with one designated model across all arms. The deposit contains the byte-frozen preregistration document (SHA-256 6c8f22130017d92974f54b1ea8780176fd253c27e903be80d3582c5ad4e6e55e), its hash manifest, the signed freeze note, and a README. The preregistration file retains its DRAFT filename and internal markings; the freeze mechanism is external; the included signed freeze note of 2026-06-12 freezes the exact byte object by SHA-256. Auxiliary artifacts (arm instruments, power-table generator, amendment changelog) are committed by SHA-256 in the deposited hash manifest and will be released with the study publication. No real-data run precedes this deposit.
No takes yet. Share an insight, caveat, or question.
John Mackay Søfteland (2026) studied this question.
Synapse has enriched 5 closely related papers on similar clinical questions. Consider them for comparative context: