Transparency: Transparency in AI-assisted screening

A scholarly practice guide on transparency in ai-assisted screening for authors, editors, and GCC research offices—with standards links, checklists, limitations, and Vision 2030 context.

Introduction and aim

Transparency in AI-assisted screening is not a peripheral concern for scholarly publishing—it shapes how manuscripts are screened, how peer reviewers are briefed, and how readers judge trust. Across Saudi Arabia and the wider Gulf Cooperation Council (GCC), research output is expanding under Vision 2030 priorities that reward credible journals, transparent ethics, and discoverable records. Lumora Editorial Office publishes this guide for authors, editors, society officers, and research administrators who need a field-usable map rather than a slogan.

The aim of this article is to explain transparency in AI-assisted screening with enough operational detail that a corresponding author or managing editor can act this week: which standards to open first, which records to keep, which disclosures to draft, and which failure modes to avoid. Guidance is aligned with verified institutional sources such as ClinicalTrials.gov, PROSPERO, and OSF Registries—and with Lumora’s open-access, double-blind peer-review model.

This is an editorial practice paper, not a systematic review and not a substitute for journal-specific author instructions or institutional counsel. Where national law, hospital IRB rules, or funder contracts are stricter than publisher norms, follow the stricter requirement and document the rationale in the submission package.

Problem and goal

The problem is familiar: teams treat transparency in AI-assisted screening as a late-stage polish item. Authorship is renegotiated after acceptance; conflicts are disclosed incompletely; metadata is repaired only when a DOI deposit fails; indexing language is overstated on the website; AI assistance is used silently; and regional clinical context is either ignored or overstated without design justification. Each of these gaps increases desk-return risk, lengthens revision cycles, and can escalate into integrity investigations under PROSPERO or comparable pathways.

The goal is a controlled workflow. Every co-author should be able to answer: Who owns this step? What evidence sits in the shared folder? Which public policy (COPE, ICMJE, DOAJ, Crossref, Helsinki, EQUATOR, or funder rules) governs the decision? For transparency work, that answer must be written—not assumed—before the cover letter is uploaded.

Success looks concrete: fewer administrative returns, cleaner Crossref metadata, reviewer reports that address methods rather than missing forms, and public pages that match what the journal actually does. For GCC institutions, success also means bilingual readiness where needed and records that survive staff turnover.

Core framework for transparency

Start with a framework, not a tool list. Map the manuscript journey: conceptualization → ethics and data stewardship → drafting and referencing → submission metadata → peer review → revision → production → identifiers and archiving → post-publication corrections if needed. Transparency in AI-assisted screening sits inside that journey at more than one node; treating it as a single checkbox is how teams fail audits.

Name standards early. Depending on design, that may include ClinicalTrials.gov, PROSPERO, WHO ICTRP, and reporting checklists from networks such as EQUATOR when methods papers apply. Do not invent citation identifiers. Prefer landing pages and official guidance URLs that resolve today, and keep a dated screenshot or PDF of the version you followed if policy pages change.

Assign roles with the CRediT taxonomy mindset even when the journal’s form is simpler: who designed the study, who acquired data, who drafted, who supervised, who is accountable for integrity. Silent contributors and gift authorship both damage trust. ICMJE authorship criteria remain the baseline many medical journals expect authors to affirm.

Build a evidence pack beside the manuscript: protocol or analysis plan (if any), ethics approvals and consent templates, conflict statements, funding contracts, data availability notes, AI-use disclosure, and a reference library export (BibTeX/RIS) that matches the final PDF. Lumora editors can resolve questions faster when the pack is complete on first submission.

Evidence, standards, and what “good” looks like

Evidence in publishing practice means traceable decisions. If you claim indexing readiness, show the application status and the criteria you meet—not a badge copied from another title. If you claim open access, show the license (for example Creative Commons terms when used) and the fee/waiver policy in plain language. If you claim double-blind review, show how identities are masked in files and invitations.

Read the primary sources linked in the References section—especially ClinicalTrials.gov and PROSPERO—and annotate which clauses apply to your study design or journal role. Strong teams keep a one-page “standards map” in the project folder. Weak teams paste generic ethics paragraphs that contradict the methods.

For quantitative or clinical work, pair publisher ethics with reporting standards. Incomplete outcome reporting, undocumented protocol deviations, and selective citation are scientific problems as much as editorial ones. Peer reviewers should be asked to examine methods and inference, not to invent missing disclosures for the authors.

Qualitative signal ≠ metric theatre. Avoid fabricated impact claims, invented DOIs, or AI-generated reference lists. If a citation cannot be verified, remove it. If a DOI does not resolve, fix or replace it before production. Integrity findings often begin with a single mismatched reference or an image that cannot be traced to raw data.

Finally, treat peer review as a scientific conversation. When reviewers ask for sensitivity analyses, clearer population definitions, or tempered conclusions, respond with revised text and—when appropriate—supplementary material. Editors notice when authors improve the record versus when they argue for prestige. That distinction is part of transparency in AI-assisted screening as much as any policy PDF.

Practice guidance: authors

Before writing, agree authorship order and corresponding-author duties in writing. Discuss transparency in AI-assisted screening explicitly: what will be disclosed, who will respond to editors, and which institutional contacts approve data or media release. Early clarity prevents mid-revision authorship disputes that force editorial holds.

During drafting, keep claims proportional to design. Distinguish association from causation; label exploratory analyses; report limitations in the manuscript, not only in a cover letter. Use reference managers that preserve Crossref/PubMed metadata, and verify each citation supports the sentence it accompanies.

At submission, complete every form as if it will be audited. Upload ethics documents when required; declare funding and conflicts; state AI assistance honestly; confirm that the work is not under consideration elsewhere; and ensure English and Arabic titles/abstracts (when used) are substantively aligned, not machine-calqued.

After decision, treat revision letters as scholarly documents: quote each reviewer point, answer with evidence, and mark manuscript changes. If you disagree, argue with methods—not with tone. Timely, precise responses are among the strongest predictors of a clean path from revise to accept in Lumora journals.

For multi-author GCC teams, appoint one integrity lead who owns the evidence pack and one language lead who owns bilingual abstract parity. Rotate neither role mid-revision without written handover. That simple governance reduces the most common failure mode behind transparency in AI-assisted screening: everyone assumed someone else checked.

Practice guidance: editors, societies, and research offices

Editors should operationalize transparency in AI-assisted screening in triage checklists: scope fit, ethics completeness, reporting standards, plagiarism/image flags, and conflict screening before reviewer invitations. Desk returns that name the missing item save months of avoidable peer-review labor.

Societies that own or partner on journals must separate prestige from capacity. Publish governance documents, editor independence norms, and fee transparency early. Member value should include training for reviewers and authors—not only a PDF archive. Succession plans for editor-in-chief roles prevent operational collapse when volunteers rotate.

Research offices and libraries can raise institutional quality faster than any single workshop by standardizing ORCID collection, providing reference-manager training, hosting template ethics language that matches local IRBs, and reviewing publisher contracts for licensing and APC clarity. Vision 2030-aligned research systems need infrastructure, not only output counts.

Production teams should treat metadata as part of scholarship: accurate titles, abstracts, contributor names, affiliations, funder identifiers, license URLs, and references enable Crossref registration and downstream discovery. A beautiful PDF with broken metadata is a discoverability failure.

When commissioning special issues or society supplements, write the guest-editor brief as a miniature journal policy: scope boundaries, conflict rules, expected turnaround, and a prohibition on inventing indexing claims. Special issues fail publicly when transparency in AI-assisted screening is treated as optional marketing copy rather than editorial law.

Saudi and GCC regional perspective

Regional scholarship is growing in volume and ambition. That growth raises the cost of weak transparency practice: international readers and indexers apply the same integrity expectations to Riyadh, Jeddah, Doha, Abu Dhabi, and Manama that they apply elsewhere. Transparency in AI-assisted screening is therefore both a local operational issue and a global trust issue.

Bilingual communication is a strength when done rigorously. Arabic and English abstracts should convey the same study design, population, and conclusions. Poor translation that changes clinical meaning is an integrity risk. Where possible, subject-fluent editors review both languages before DOI deposit.

Health-system and education research in the GCC often involves multi-site approvals, student researchers, and sensitive data. Document power dynamics in authorship, keep consent records retrievable, and avoid public claims that identify participants. Follow Helsinki principles and local IRB rules; publisher templates do not override hospital law.

Lumora’s regional stance is practical: meet international baselines (ClinicalTrials.gov, PROSPERO, OSF Registries), teach early-career researchers good habits, and refuse vanity metrics. Journals that publish slowly but honestly outperform titles that publish quickly with theatrical indexing claims.

Finally, connect institutional KPIs to quality infrastructure: ORCID coverage, timely corrections, transparent APC/waiver pages, and reviewer training hours—not only paper counts. Vision 2030 research ambition is credible only when transparency practice can be audited by an external reader using WHO ICTRP and peer sources.

Worked checklist for this week

1) Open ClinicalTrials.gov and PROSPERO; list three requirements that apply to your current manuscript or journal role. 2) Create or update a shared evidence pack (ethics, conflicts, funding, AI disclosure, reference export). 3) Hold a 20-minute co-author huddle focused only on transparency in AI-assisted screening. 4) Run a citation integrity pass: every reference resolves and supports its claim.

5) Compare your draft against Lumora author guidelines and the target journal’s scope statement. 6) If you are an editor, add one triage question about transparency to the desk checklist. 7) If you are a society officer, confirm that public policy pages match internal practice. 8) Schedule a post-acceptance metadata review before DOI registration.

9) Record decisions in writing—email or lab notebook—so future corrections are possible. 10) When unsure, ask the editorial office early; silence followed by a surprise integrity issue is more expensive than a clarifying question.

11) Spot-check two competitor or peer journals’ public policies and note one practice worth adopting honestly. 12) Re-read your own website claims about transparency as if you were a skeptical indexer—then fix any mismatch before the next submission cycle.

Limitations of this guide

This article cannot replace journal-specific instructions, institutional legal advice, or clinical judgment. Field norms differ (for example, computer-science proceedings versus clinical trials). Timelines and tooling examples are qualitative and will age; always verify the current text of ClinicalTrials.gov and companion sources.

We do not invent DOIs, impact factors, or acceptance guarantees. Indexing outcomes depend on sustained practice, not on a single checklist. Illustrations in this series are conceptual emphasis charts—not measured effect sizes. Where your institution’s policy conflicts with a general recommendation here, follow the institution and document why.

Peer review of this guide itself treats it as practice literature: useful when it reduces avoidable errors; incomplete when your specialty reporting standard (CONSORT, PRISMA, STROBE, ARRIVE, and others) imposes stricter itemization than the publisher ethics baseline.

Conclusion and actionable takeaways

Transparency in AI-assisted screening becomes manageable when treated as a designed workflow with standards, roles, evidence, and honest public claims. Authors should disclose early and cite carefully; editors should triage with checklists; societies and research offices should fund training and metadata infrastructure; production teams should treat identifiers as scholarship.

Takeaways: (1) read primary sources before submission theatre; (2) keep an evidence pack; (3) never invent citations or badges; (4) align bilingual abstracts; (5) answer revision letters with methods; (6) match website claims to real practice; (7) ask Lumora Editorial Office when edge cases appear. Done consistently, these habits raise trust in GCC research and in the journals that publish it.

If you implement only three actions after reading: open ClinicalTrials.gov, complete the evidence pack for your active manuscript, and rewrite one public claim so it matches operational reality. That triad—standards, evidence, honesty—is the scientific core of transparency in AI-assisted screening on the Lumora platform.

References

1. National Library of Medicine. ClinicalTrials.gov . Accessed 26 Dec 2026.
2. University of York CRD. PROSPERO . Accessed 26 Dec 2026.
3. Center for Open Science. OSF Registries . Accessed 26 Dec 2026.
4. WHO. WHO ICTRP . Accessed 26 Dec 2026.
5. EQUATOR Network. EQUATOR Network . Accessed 26 Dec 2026.