huang-sir1 Research Ethics, Consent & Privacy
Use this skill to write the Ethics, Informed Consent, and privacy/governance parts of
an imaging (or imaging+omics) manuscript or grant so they are compliant and honestly bounded.
This is the governance counterpart to radiology-data (which handles the mechanics of
de-identification, repositories, and availability statements). This skill frames the human-
subjects side: approvals, consent, re-identification risk, and what may lawfully be shared.
Core stance
- State the approval, don't invent it. Ethics-committee/IRB name, approval number, and date, plus the consent status, come from the author. Use placeholders; never fabricate them.
- Consent must match the design. Retrospective studies often have a documented waiver of consent; prospective studies need consent (and a consent process described). Say which, and on what basis the waiver was granted.
- Re-identification is a spectrum. DICOM metadata, dates, rare-disease small cohorts, facial reconstruction from head imaging, and genomic data each raise risk; address them specifically.
- Sharing is governed, not automatic. Imaging may be openly shareable after de-identification;
genomic/clinical data are often controlled access. The Ethics/consent must permit whatever
the Data Availability statement promises (keep them consistent with
radiology-data). - Jurisdiction matters. HIPAA (US), GDPR (EU), PIPL (China) and local rules differ; name the framework that applies and defer specifics to the institution.
- Integrity & scope. This is drafting and risk-flagging support, not legal advice; approval numbers, consent terms, and DUAs must be confirmed with the institution/IRB.
When to use
- "Write the Ethics / Informed Consent / privacy statement / 帮我写伦理与知情同意表述。"
- "Retrospective multi-center study — how do I word the consent waiver?"
- "We have genomic + imaging data — what's the re-identification risk and how do I share it?"
- "Which approval/consent text does a grant or Radiology submission need?"
- "Is 'available on reasonable request' acceptable, and what governance must back it?"
When to open extra files
| File | Open when |
|---|---|
| references/approval-consent.md | Study-type framing, IRB/ethics approval, consent vs waiver, multi-center approvals, registration |
| references/reidentification-risk.md | Assessing/мitigating re-identification: DICOM metadata, dates, small cohorts, defacing, genomic data |
| references/governance-sharing.md | HIPAA/GDPR/PIPL framing, data-use agreements, controlled access, aligning consent ↔ Data Availability |
Workflow
- Classify the study — retrospective / prospective / registered / multi-center; secondary use of existing data; involvement of imaging, clinical records, pathology, genomics.
- Establish approvals & consent (approval-consent.md) — committee, number, date (placeholders if unknown), consent obtained vs documented waiver and its basis; per-center approvals.
- Assess re-identification risk (reidentification-risk.md) — what in the data could
re-identify a participant, and the mitigations (de-id, defacing, aggregation, controlled
access). Coordinate de-id mechanics with
radiology-data. - Set governance for sharing (governance-sharing.md) — what may be public vs controlled vs restricted, the applicable framework, DUA needs; ensure it matches the Data Availability statement.
- Draft statements — Ethics, Informed Consent, and (if needed) a privacy/governance note, in submission-ready English, with a 待确认 list of author-only facts.
Output contract
Study/ethics classification— design, data types, jurisdiction/framework.Ethics statement— committee, approval number, date (placeholders where unknown).Consent statement— consent obtained or documented waiver + basis; prospective process if applicable.Re-identification risk read— risks + mitigations; defacing/controlled-access decisions.Governance/sharing note— what can be shared and how, consistent withradiology-data.待确认（中文）— approval numbers, dates, consent terms, DUA details to confirm with the IRB.
Quality bar
A good ethics section is specific, consistent with the data-sharing plan, and never claims an approval, consent, or shareability the study does not actually have. It reads like an investigator who knows the governance — and flags exactly what only the IRB can confirm.
Handoffs
- De-identification mechanics, repositories, Data/Code Availability statements →
radiology-data. - Controlled-cohort design (dbGaP/EGA), matched imaging+omics governance →
radiology-radiogenomics. - Ethics/open-science checklist items (CLAIM/TRIPOD+AI) →
radiology-reporting. - Grant ethics/feasibility framing →
radiology-grant. - Not legal advice — confirm all approvals, consent, and DUAs with your institution/IRB.