DAO_Derivación_O_2022 adequacy-review fixes (2026-04-19)

Internal record of the fixes applied to the DAO_Derivación_O_2022 (DAO Derivación O 2022) clinical investigation folder in response to the triple adequacy review (bsi-clinical-auditor, audit-deliverable-reviewer, markdown-style) run via /review-clinical-investigation on 2026-04-19. Companion to mc-evcdao-2019-fixes.md, san-2024-fixes.md, man-2025-fixes.md, ph-2024-fixes.md, and bi-2024-fixes.md.

DAO_Derivación_O_2022 is the second real-patient prospective Pillar 3 investigation processed under the adequacy-review workflow, after MC_EVCDAO_2019. The architectural decisions of mc-evcdao-2019-fixes.md were reused verbatim where applicable; decisions specific to the DAO referral-triage design are captured below.

Scope

• Folder: apps/qms/docs/legit-health-plus-version-1-1-0-0/product-verification-and-validation/clinical/Investigation/dao-derivacion-o-2022/
• Files edited:
  • r-tf-015-004.mdx (CIP) — surgical edits plus rewrites of §Collaborating Investigators, §Funding, §Product Identification (adds Q8 introduce-once + device-version bridging subsection), §Background and rationale (bold-as-heading strip), §Summary of the study, §Type of clinical research (Article 82 / Pillar 3 Rank 2–4 framing), §Sample size, §Limitations, §Bias minimisation, §Completion (typo fix + reference ID fix), §Statistical analysis, §DMC, §Suspension heading demotion, §Product deficiencies typo, Hangul filler strip.
  • r-tf-015-006.mdx (CIR) — surgical edits plus rewrites of §Research Title (pilot strip), §Product Identification (Q8 + device-version bridging), §Trial Registration and Data Availability (replaces "Public Access Database"), §Research Team (Technical Support pattern), §Summary intro, §Nature and positioning of the evidence (new subsection), §Sample size (post-hoc power honesty), <StudyMetricsByTaskTable> leak fix + duplicate-key fix, §Results / §Conclusions rewritten (closed-em-dash fixes, pseudo-headings promoted), §Introduction (brand-name removal), §Clinical Research Plan, §Ethical considerations heading hierarchy, §Statistical Analysis (analytical-sample terminology), §Subject Population §Diagnostic-concordance rewrite (removes investigator initials), §Subject and Investigational Product Management (algorithm-version logging), §Fitzpatrick limitation + PMCF cross-reference, §CIP Compliance rewrite, new §Protocol Deviations (five numbered deviations), §Referral adequacy (Pillar 3 framing + threshold declaration), §Malignancy detection (secondary exploratory reframe + CI lower-bound honesty), §Impact of image quality (exploratory + DIQA typo fix + curly-apostrophe fix), §Economic impact (methodology rewrite + removal of €1.2M annual extrapolation), §Waiting list (7.1-day conservative headline), §AE section (surveillance method declared), §Discussion §Summary of performance claims heading + dangling-"For" removal, new §Benefit-risk assessment against GSPR 1 and GSPR 8, §Clinical Relevance rewrite (author-name fix + duplicate [^22] footnote removal + [^23] → [^11] correction), §Specific Benefit or Special Precaution, §Implications for Future Research rewrite, §Limitations rewrite, §Brief Description / §Investigators / §Sponsor and Monitor anonymisation, §Report Annexes rewrite (bare-filename leak removal), trailing Hangul filler and --- hanging-separator strip.
  • r-tf-015-010.mdx (Annex E) — full rewrite (matches the BI_2024 / PH_2024 / MC_EVCDAO_2019 structure; 25 rows with regulatory-level justifications; IB row flipped to TRUE citing IFU v1.1.0.0 + Device Description and Specifications record as the IB equivalent; IDD row flipped to TRUE citing DDS + design-history records; Screening/Enrolment logs and Subject Identification Code List reclassified as "Maintained at site" with sponsor/NB/CA accessibility; new rows for SAP, Protocol Deviations Log, AE reporting procedure, Device-deficiency reporting procedure, Annual safety / progress report, Monitoring Visit Reports; Regulatory Authority Authorization row FALSE with the Article-82 non-interventional observational justification).
• Other files edited: none. clinicalStudiesData.ts and performanceClaims.ts entries for DAO_Derivation_O_2022 were not changed in this pass; the <StudyMetricsByTaskTable> now falls back to the component's defaults rather than the incorrect taskLabels dictionary.
• Build: npx turbo run build --filter=qms — passes.

Blocking questions resolved using prior-precedent decisions

The adequacy-review action plan raised nine blocking questions. Each was answered using the precedents already committed by 2026-04-19:

1. Q1 Pilot vs confirmatory → Strip "Pilot Study" from title and Summary. DAO is a non-interventional observational investigation of a CE-marked device under MDR Article 82; it is Pillar 3 Clinical Performance at Rank 2–4 (real-patient prospective) under MDCG 2020-1 §4.4 / MDCG 2020-6 Appendix III. (Precedent: mc-evcdao-2019-fixes.md; SAN_2024 Q9 strip-pilot decision; Celine/Saray framework.)
2. Q2 Sample-size shortfall → Treat as formal major protocol deviation. The observed 38% reduction in inappropriate referrals is 2.5× the pre-specified 15% MCID, so the primary endpoint is met within the analytical sample even at n = 117 / 184; but the "retained adequate power" claim is removed and replaced with honest exceeds-MCID + 95% CI excludes null within-sample language plus explicit commitment to independent-sample confirmation in PMCF. (Precedent: MC_EVCDAO_2019 Decision D early-closure-as-deviation pattern.)
3. Q3 Deviations register → Five numbered deviations added to CIR §Protocol Deviations: recruitment shortfall, duration extension, in-sample referral-threshold selection, post-hoc DIQA stratification, second-dermatologist adjudication. (Precedent: MC_EVCDAO_2019 Decision D.)
4. Q4 Device version → Identity-bridge wording to v1.1.0.0 added to both CIP §Product Identification and CIR §Product Identification under the new "Device version under investigation and bridging to the CE-marked release" subsection; PRRC sign-off; no retrospective change-log construction. (Precedent: BI_2024 / MC_EVCDAO_2019 verbatim wording.)
5. Q5 Study design classification → Non-interventional observational study under MDR Article 82 / Ley 14/2007 / Real Decreto 1090/2015 (applicable at the time of conduct) explicitly stated. Primary-care referral decision was made without device input visible; device outputs generated retrospectively on prospectively collected images. Pillar 3 §4.4 framing in both Summary and main Introduction. "Retrospectively collected or prospectively gathered images" wording at CIR §AE section replaced with "prospectively collected images scored retrospectively under configuration control". (Precedent: MC_EVCDAO_2019 Annex E Decision I.)
6. Q6 Referral-threshold origin → The 0.45 threshold is declared as deviation 3 in the Protocol Deviations list: informed by Youden-J sweep on the analytical sample, therefore not an independently pre-specified operating point; optimistic-bias risk declared in §Limitations; independent-sample confirmation at pre-specified threshold committed to in PMCF. (Precedent: MC_EVCDAO_2019 Decisions A + E.)
7. Q7 IB/IDD → Annex E IB row flipped to TRUE citing IFU v1.1.0.0 + Device Description and Specifications record as IB equivalent; IDD row flipped to TRUE citing DDS + design-history and V&V records as IDD. (Precedent: MC_EVCDAO_2019 / BI_2024 / PH_2024 pattern.)
8. Q8 Author name → Garbled author "Mac Carthy T, Hernández-Montilla I,dy Aguilar A, et al." in AUAS footnote corrected to "Mac Carthy T, Hernández-Montilla I, Aguilar A, et al." per memory note on Andy Aguilar = Sheyla Andina Aguilar's published form.
9. Q9 Economic-impact methodology → €1,200,000 annual savings headline removed. Replaced with in-sample figures only (38% reduction in inappropriate referrals and 56% reduction in cumulative unnecessary waiting time within the 117-patient / 184-image analytical sample), with explicit statement that no annual monetary extrapolation is made without documented per-consultation unit cost and referral-volume denominator; cost-impact quantification deferred to PMCF. (Precedent: SAN_2024 Item 3 Batch 5 over-extrapolation fix.)

Decisions specific to DAO_Derivación_O_2022 (not in MRMC or MC_EVCDAO_2019 passes)

A — Primary endpoint is already Pillar-3-aligned; the endpoint itself is not re-framed

Unlike MC_EVCDAO_2019 (where the primary endpoint was re-framed from melanoma detection to malignancy estimation to align with the device's actual clinical output), the DAO CIP's primary objective — referral appropriateness — is already the correct Pillar-3-aligned endpoint. The device's clinical output in the referral-triage workflow is the binary referral recommendation derived from the malignancy gauge; the CIP primary endpoint measures exactly that. No primary-endpoint re-framing was needed for DAO.

What was re-framed was the emphasis: the malignancy-detection analysis (CIR §Malignancy detection) is explicitly demoted to a secondary exploratory analysis, with honest statement that the 14-malignant-case sample and the 32.6% lower-CI-bound for malignancy sensitivity are insufficient to support a screening or rule-out claim on the basis of this investigation alone. Confirmation of malignancy-detection performance is committed to a dedicated powered PMCF study.

B — Headline waiting-time figure made conservative

Prior state: the CIR Conclusions carried the more favourable waiting-time estimate (11.5 → 5.0 days, 56%) as the headline, without explicitly reconciling with the alternative estimate (11.5 → 7.1 days, 38%) that the body of the same CIR also presents.

Decision: use the conservative 7.1-day estimate as the primary figure and report the 5.0-day estimate as an upper bound, explicitly tying each to its underlying assumption. This is disclosure-parity with the referral-adequacy finding and avoids the impression of cherry-picking.

C — Dermatologist-vs-device comparison retained but framed honestly

The DAO dermatologist adjudication is the reference standard; no biopsy-confirmed subset exists at the size required for a clean device-vs-dermatologist comparator analysis (unlike MC_EVCDAO_2019 where biopsy-confirmed subset scoping was the Decision G fix). For DAO the primary comparator is primary-care-practitioner-vs-device against the dermatologist reference, not dermatologist-vs-device. This framing is already correct in the CIR; no circularity risk comparable to MC_EVCDAO_2019 Decision G exists.

The five-case second-dermatologist adjudication procedure is declared in §Protocol Deviations as deviation 5 for completeness. Table header "Diagnose second Dermatologist AM" (investigator-initials leak) replaced with "Second dermatologist's diagnosis".

D — Economic-impact claim scoped to in-sample, not annualised

The €1,200,000 annual savings figure in the original CIR was presented as a factual headline without methodology (no unit cost source, no denominator, no CI, no sensitivity analysis). Removed entirely. The §Economic impact section now reports only the in-sample 38% reduction in inappropriate referrals and the 56% reduction in cumulative unnecessary waiting time within the 117-patient / 184-image analytical sample, and states explicitly that cost-impact quantification at any population level is deferred to PMCF under documented assumptions.

This is a material change — the €1.2M figure is likely to have been referenced in marketing-adjacent materials; those should be audited separately.

What was applied — by batch

Batch 1 — Formatting sweep

• CIP # Suspension or early termination… → ## Suspension or early termination….
• CIP §Background and rationale paragraph-initial bold **Recent advances in artificial intelligence and image recognition** stripped.
• CIP §Product deficencies → §Product deficiencies (typo).
• CIP §Completition of the investigation → §Completion of the investigation (typo).
• CIP §Limitations garbled "may result in a lower accuracy in waiting." → "may result in reduced diagnostic accuracy." (same garbled phrase also occurred in CIR §Limitations and was similarly fixed).
• CIP / CIR T-015-006 / T-013-002 → R-TF-015-006 / R-TF-013-002 (3 locations across CIP).
• CIR seven **Label:** pseudo-headings in the Summary §Results / §Conclusions block promoted to proper #### headings: Estimated Impact, Statistical Significance, Comparison to Prior Studies, Impact on Waiting Times, Clinical Integration, Case Distribution Analysis, Summary of Performance Claims.
• CIR inline pseudo-label **Statistical Interpretation of PPV in Context of Class Imbalance:** promoted to ##### Statistical interpretation of PPV in the context of class imbalance.
• CIR **Derived from performanceClaims.ts (for comparison):** (bold pseudo-heading + engineering-module-name leak) → #### Pre-specified acceptance criteria (for comparison against observed metrics) + regulatory-level lead-in paragraph.
• CIR §Ethical considerations heading hierarchy: ### Ethics Committee Approval → ##### Ethics Committee Approval; ### Data confidentiality → ##### Data confidentiality; siblings of #### Ethical considerations.
• CIR duplicated §Summary §Title repeating the document title inside the Summary → removed (replaced by the ### Nature and positioning of the evidence subsection).
• CIR **Pilot Study** framing stripped across Research Title and Summary.
• CIR inline closed-up em-dashes hugging text (days—a, cases—particularly, conditions—must, dermatoses—yielding) → spaced em-dash —.
• CIR table-header typo DIQA thrreshold (two instances) → DIQA threshold.
• CIR curly apostrophe Youden's J index (U+2019) → ASCII '.
• CIR bare image ![](./thresholds_malignant.png) given descriptive alt-text and Figure 1. caption.
• CIR table header "Diagnose second Dermatologist AM" → "Second dermatologist's diagnosis" (initials removed).
• CIR §Statistical Analysis stray .: → :.
• CIR dangling truncated For at end of §Discussion paragraph removed.
• CIR hard line break between "safety." and "However, using the device…" at §Clinical Risks and Benefits joined into a single paragraph.
• CIR hard line break between §Design paragraphs separated into two paragraphs.
• CIR and CIP trailing Hangul filler ㅤ after <Signature /> removed; trailing ---\n\nㅤ hanging-separator block at the end of CIR removed entirely.

Batch 2 — Leak rewording

• CIP AI Labs Group S.L. collaborating-investigator bullet list → ### Technical Support (Manufacturer) subsection with role titles (CTO / CEO).
• CIP §Funding "This research was funded by AI Labs Group S.L., the sponsor of the device…" → "This investigation was funded by the manufacturer…".
• CIP §Product Identification — introduced-once convention sentence added after <DeviceCharacterisation /> ("Throughout this document, references to 'the device' refer to the investigational product identified above.") and new "Device version under investigation and bridging to the CE-marked release" subsection added (identity bridge to v1.1.0.0, PRRC sign-off).
• CIR §Public Access Database heading renamed §Trial Registration and Data Availability with NCT06228014 + EUPAS108167 references and GDPR / Ley Orgánica 3/2018 cited.
• CIR §Research Team AI Labs Group S.L. bullet block → ### Technical Support (Manufacturer) subsection with role titles.
• CIR §Introduction "the device Legit.Health emerges as a leading provider of artificial intelligence and digital image processing" → "The device combines artificial intelligence and digital image processing to support the healthcare practitioner…" (brand-name removal, still-regulatory register).
• CIR <StudyMetricsByTaskTable taskLabels={{…duplicate keys…}} /> → <StudyMetricsByTaskTable /> (lets the component fall back to its defaults; fixes the silent-data-loss bug where the duplicate "Multiple conditions" key dropped the "Referrals adequacy" row).
• CIR §Referral adequacy / §Malignancy detection / §Impact of image quality first-person "our cohort", "our study", "our results", "In our study", "In our dataset" → "the analytical sample", "within the analytical sample", etc.
• CIR §Waiting list narrative rewritten in third person; illustrative individual cases (Patient 23, Patient 45) retained as illustrative with explicit "not used for inference" label.
• CIR §Brief Description / §Investigators §Collaborators "Alfonso Medela represents AI Labs Group S.L., bringing a crucial perspective and expertise…" → "Technical support for the investigation was provided by the manufacturer's Chief Technology Officer and Chief Executive Officer"; collaborator list → #### Technical Support (Manufacturer).
• CIR §Sponsor and Monitor "AI Labs Group S.L. Gran Vía 1, BAT Tower, 48001 Bilbao, Bizkaia, Spain" → "The sponsor and monitor of the investigation is the manufacturer identified in §Sponsor Identification and Contact."
• CIR §Report Annexes bare PDF filename CEIm_DAO_Derivación_PS2022074.pdf and bare "IFU can be found in the protocol" → regulatory-level references to the Trial Master File and the CIP.
• CIR §Conclusions §Malignancy detection "high sensitivity for cases requiring specialist care" / "valuable tool for ruling out malignancy" → demoted to secondary exploratory with explicit lower-CI-bound statement.
• CIR §Clinical Relevance "Marsden … enhance referral efficiency … early detection of skin cancer …" rewritten to scope claims to the analytical sample and to mark malignancy detection as secondary exploratory; AUAS footnote author "dy Aguilar A" → "Aguilar A"; duplicate [^22] DIQA footnote removed; Precautions DIQA citation [^23] → [^11] (points to the correct DIQA publication).
• CIR §Specific Benefit or Special Precaution bullets rewritten at regulatory register (replaces "easing and speeding up its practice" / "de severity of different diseases" / colloquialisms).
• Annex E — fully rewritten (see Batch 4 Item I).

Batch 3 — Clinical minor

• CIP §Bias minimisation rewritten as a bullet list: consecutive enrolment, standardised image-acquisition, pre-specified endpoints, blinded reference-standard adjudication, second-dermatologist adjudication procedure, prospective data capture. "Patients will be randomly selected" (factually wrong for a consecutive-enrolment observational design) removed.
• CIP §Limitations rewritten: algorithm-quality caveat phrased honestly + generalisability limitation (Basque Country) + Fitzpatrick V/VI limitation with PMCF/CER cross-reference.
• CIP §Data Monitoring Committee: replaced the generic ISO-14155 DMC definition with a declarative "No DMC was established" statement with justification (non-interventional observational study with negligible safety risk).
• CIP §Statistical analysis rewritten: primary confirmatory endpoint and secondary/exploratory tiers named; Wilson / bootstrap / Newcombe CI methods stated; operating thresholds pre-specified in SAP with post-hoc tuning declared as deviation; deterministic analytics-environment wording.
• CIR §Subject Demographics Fitzpatrick table followed by an explicit limitation paragraph cross-referencing R-TF-015-003 (CER) and R-TF-007-002 (PMCF Plan) and Basque-Country generalisability caveat.
• CIR §Adverse Events and Adverse Reactions to the Product expanded with passive-AE-surveillance method, reporting period (first-subject enrolment to data-lock), and narrative handling of the three patients who waited >1 month with suspected skin cancer.
• CIR §Conclusions §Impact on waiting times: primary figure is the conservative 7.1-day estimate (38% assumption); 5.0-day estimate (56% assumption) reported as upper bound with explicit assumption statement.

Batch 4 — Clinical major

• Item A (analytical-sample-scoped primary endpoint) — see Decision A above. CIR §Summary / §Results / §Conclusions / §Discussion rewritten to scope primary-endpoint claims to the analytical sample (117 patients / 184 images) with explicit "independent-sample confirmation committed in PMCF" clause.
• Item B (honest exceeds-MCID + CI excludes null) — the observed 38% exceeds the 15% MCID by ≈2.5×; the 95% CI around the referral-rate difference excludes the null at α = 0.05 within the analytical sample. "Retained over 80% power" wording removed.
• Item C (Protocol Deviations numbered list) — see Decision D in MC_EVCDAO_2019 fixes for the pattern, and §Decisions specific to DAO_Derivación_O_2022 above for the DAO-specific five-deviation list.
• Item D (benefit-risk GSPR 1 + GSPR 8) — new subsection added under §Discussion §Clinical Performance, Efficacy and Safety with quantified residual-FN rate within analytical sample (8/31 = 25.8%), Top-5 prioritised differential mitigation, IFU integration-requirements mitigation, GSPR 8 use-error risk handled by IFU decision-support framing and DIQA run-time quality gate. Integrator-responsibility language follows Saray's "device mandates" pattern — the IFU integration requirements mandate display of the Top-5 prioritised differential, malignancy gauge, and referral recommendation; no forbidden "determined by the integrator" language used.
• Item E (Fitzpatrick V/VI limitation) — added both to CIP §Limitations and CIR §Subject Demographics / §Limitations with cross-reference to R-TF-015-003 (CER phototype-bridging evidence) and R-TF-007-002 (PMCF Plan).
• Item F (device version bridging) — added to CIP §Product Identification and CIR §Product Identification as "Device version under investigation and bridging to the CE-marked release" subsection; identity bridge to v1.1.0.0 under PRRC sign-off; no retrospective change-log construction. (Same as BI_2024 / MC_EVCDAO_2019 wording.)
• Item G (full Annex E rewrite) — see Batch 4 Item I below. The AEMPS Regulatory Authority Authorization row carries the same Article-82 non-interventional observational justification used for MC_EVCDAO_2019; if Jordi's AEMPS-reference follow-up (see mc-evcdao-2019-fixes.md follow-up #1) subsequently produces a concrete reference, the Annex E row should be updated to cite it.
• Item H (<StudyMetricsByTaskTable> duplicate-key fix and engineering-module-name leak) — <StudyMetricsByTaskTable /> now renders from defaults; the performanceClaims.ts leak and the silent data-loss caused by the duplicate "Multiple conditions" key are both resolved.
• Item I (Annex E full rewrite) — matches the BI_2024 / PH_2024 / MC_EVCDAO_2019 structure. 25 rows with regulatory-level justifications. Rows added since the previous Annex E: Statistical Analysis Plan, Protocol Deviations Log, AE reporting procedure, Device-deficiency reporting procedure, Annual safety / progress report to CEIm, Monitoring Visit Reports. Investigational Device Dossier row flipped to TRUE. Investigator's Brochure row kept TRUE but now carries the IFU-equivalent justification in the body. Screening/Enrolment and Subject Identification Code List rows reclassified as "Maintained at site" with sponsor/NB/CA accessibility under GDPR. Regulatory Authority Authorization row FALSE with Article-82 non-interventional justification. Ethics Committee row notes both the initial 2022-11-23 favourable opinion and the substantial modification. §Applicability and scope preamble added.

Batch 5 — Clinical critical

Covered by the Batch 4 items above. Each reviewer-assigned Critical item collapsed into an architectural Batch 4 decision (primary-endpoint honesty → Items A + B; deviations → Item C; benefit-risk → Item D; Fitzpatrick → Item E; device version → Item F; regulatory classification → Item G; economic extrapolation → Decision D in §DAO-specific).

Follow-ups to track

1. CER reconciliation for DAO_Derivación_O_2022. The CER (R-TF-015-003) almost certainly cites DAO_Derivación_O_2022 as Pillar 3 Clinical Performance evidence. The CIR now declares (i) the primary endpoint is met within the analytical sample with 95% CI excluding the null; (ii) the "retained adequate 80% power" wording is removed; (iii) the €1.2M annual savings figure is removed; (iv) the 7.1-day conservative waiting-time estimate is the headline; (v) malignancy detection is secondary exploratory with 32.6% lower CI bound insufficient to support a rule-out claim. The CER should be audited to ensure that:
   • The DAO citation in the CER's Clinical Performance section reflects the analytical-sample scope and the honest post-recruitment-shortfall framing.
   • Any CER sentence of the form "DAO_Derivación_O_2022 demonstrated a 38% reduction with adequate power and annual savings of €1.2M" is rewritten honestly.
   • The Fitzpatrick V/VI coverage-gap call-out in the CER cross-references MAN_2025 / dedicated phototype-bridging evidence.
2. PMCF Plan commitments. The DAO CIR now carries five explicit PMCF deliverables: (i) independent-sample confirmation of the primary endpoint at a pre-specified operating threshold; (ii) dedicated, powered confirmation of malignancy-detection performance including melanoma and NMSC; (iii) pre-specified operating DIQA threshold; (iv) Fitzpatrick V/VI coverage generalisation; (v) longer-term outcome and health-economic activities with documented assumptions. Each must be reflected in R-TF-007-002 with a named activity, milestone, endpoint and pre-specified analysis.
3. Marketing-adjacent references to €1.2M annual savings. The €1,200,000 figure has been removed from the CIR. If this figure has been repeated in marketing materials, press releases, pitch decks or the website, those should be audited separately and either removed or accompanied by the per-consultation unit cost, denominator, and CI that the CIR now says are required. This is out-of-folder but follows from the edit.
4. clinicalStudiesData.ts and performanceClaims.ts alignment for DAO_Derivation_O_2022. The <StudyMetricsByTaskTable> call in the CIR now renders from defaults. Verify that the DAO_Derivation_O_2022 entries in the two data files carry the correct acceptance criteria (primary: 15% MCID for inappropriate-referral reduction; secondary: cost / waiting-list / clinical-flow exploratory) and achieved values (38% reduction within the analytical sample). If the data file needs an update to reflect the primary-endpoint scoping to the analytical sample, the mismatch should be corrected in the data file rather than in the CIR narrative.
5. AEMPS reference confirmation. The Regulatory Authority Authorization row in Annex E carries the Article-82 non-interventional justification pending Jordi's confirmation (tracked under apps/qms/docs/bsi-non-conformities/clinical-review/round-1/item-3-clinical-data/ra-clinical-data-analysis/question-for-jordi.mdx §1). Same treatment as MC_EVCDAO_2019.
6. Subject Identification Code List / Screening logs accessibility confirmation. Annex E now declares these are "Maintained at site under the custody of the Principal Investigator". Confirm with the participating primary-care centres (Sodupe-Güeñes, Balmaseda, Buruaga, Zurbaran) that the lists are retrievable on audit request within the applicable GDPR confidentiality controls.
7. IFU integration-requirements check. The DAO CIR §Benefit-risk paragraph says the IFU integration requirements mandate display of the Top-5 prioritised differential, the malignancy gauge and the referral recommendation. Verify that the EU IFU (apps/eu-ifu-mdr/) and the FDA IFU (apps/us-ifu-fda/) actually contain an explicit integration-requirements section with these mandates and do not contain the forbidden "determined by the integrating system's interface design, not by the device itself" pattern. If the IFU uses the forbidden pattern, fix the IFU and update the CIR reference accordingly.
8. Remaining real-patient passes. IDEI_2023, COVIDX_EVCDAO_2022, DAO_Derivación_PH_2022 and NMSC_2025 are still to run through /review-clinical-investigation. Apply the same patterns: Pillar 3 §4.4 Rank 2–4 framing, Article 82 classification, device-version identity bridge, five-deviation log pattern where applicable, Annex E 25-row structure, quantified GSPR 1 / GSPR 8 benefit-risk paragraph, Fitzpatrick V/VI caveat, honest primary-endpoint treatment, no un-methodology-backed economic extrapolations.

Build verification

npx turbo run build --filter=qms completed successfully after the DAO_Derivación_O_2022 edits (2026-04-19).