SUPPLEMENTARY DATA FILE 1
An Evidence Map of Pupillary Assessment in a Preselected Acute-Care Literature Archive
Public minimized reproducibility release prepared 6 November 2026

PUBLIC FILES
This release is designated to accompany the article. Preparation of this local package does not establish actual journal upload or publication. No separate repository DOI or permanent identifier is claimed.
data/Report_Level_Data_871.json and .csv: final coded report-level decisions, six mapping domains, nonexclusive acquisition component flags, bibliographic identifiers and concise source locators.
S4_Searchable_Report_Ledger.html: offline searchable view of the same minimized 871-record dataset. Open it in a browser; no network connection is needed.
data/Audit_Return_and_Reconciliation_100.json: compact first-pass and final coded judgments, preserving discrepancies rather than overwriting them.
data/Eligibility_Review_239.json: compact eligibility decisions and actual technical source-check depth.
data/Sampling_Frames_and_Draws.json: historical frame IDs, random seed, draw order and coded baseline fields.
Data_Dictionary.json and Provenance_and_Review_Stages.json: definitions, stage distinctions, human responsibilities and release scope.
Reproduce_All_Counts.py and expected/: executable analysis checks and reference outputs. The analysis algorithm and scientific values are unchanged by data minimization.
FILE_MANIFEST.json and SHA256SUMS.txt: payload checksums; the two manifest files exclude themselves to avoid circular hashes.

RUN
Python 3.9 or newer, standard library only. No installation or network access required.
  python Reproduce_All_Counts.py --output reproduction_results.json
The script resolves inputs relative to its own location and may run from another working directory. It verifies 358 file-integrity and calculation checks, including all six domains, saved matrices, component flags, original random samples, audit outcomes, the 38+199 split, and the conditional sensitivity calculation. Passing these checks establishes numerical reproducibility, not independent full-text eligibility verification.

KEY DENOMINATORS
The fixed archive comprises 871 records: 355 core inclusions, 497 eligibility exclusions, three unresolved-source closures, 15 noncore descriptive/linkage records and one duplicate. The operative data field Exclude combines the 497 exclusions and three unresolved-source closures; they remain distinct in the published flow description. Reports do not equal independent studies or patients.
237/355 acquisition labels remain unestablished: 38 sampled/reconciled and 199 not sampled in that acquisition audit. The separate 239-record review concerned eligibility, not new acquisition classification. The 43/50 first-pass age confirmations were confirmed by Jen-Tang Sun before subsequent reconciliation, according to author clarification. Seven questioned inclusions were later retained; twelve of sixteen proposed manual classifications were supported. These stages are not interchangeable.
The conditional (38 + 199q)/355 calculation with q = 38/50 gives 53.3%, with a Wilson-derived scenario range of 45.8-58.7%. It is not a newly observed count, a confidence interval for all 355 reports or a clinical adoption estimate.
The combined 629-record sensitivity frame is 355 core reports plus 274 EX07 records; it does not make EX07 records eligible. Any-automated flags are nonexclusive and total 113/629, including two mixed-other-method reports.

MATERIALS AVAILABLE ON REQUEST
The public release omits complete source-evidence narratives, verbatim excerpts, full completed review forms, lengthy baseline and reconciliation notes, historical search/review exports, original source files and the unpublished comparison manuscript. Detailed retained research records are available on reasonable request from corresponding author Jen-Tang Sun, femh83316@femh.org.tw, subject to availability, confidentiality and source access/copyright restrictions. Copyrighted source articles must be obtained through their original lawful access routes; this package does not grant redistribution rights.
Long notes were omitted rather than converted into new classifications. The concise public locators and codes trace the published counts; they do not replace the detailed evidence required to reassess each eligibility decision. The historical full records remain preserved by the authors.
