PMCF Plan (R-TF-007-002) — Activities D.1 / D.2 rewrite + legacy-PMS-slice commitment

Scope. Replacement prose for R-TF-007-002-Post-Market-Clinical-Follow-up-PMCF-Plan.mdx covering the four PMCF edits in the task-3b5 plan (C1 declaration lines 54–55; C2 coverage lines 91–92; C3 Activities D.1/D.2 full rewrite at lines 318–345; C4 legacy-PMS-slice commitment). Internal scratch-pad.

Discipline. The §6.3 wording gate: "confirms / strengthens" ✓, "fills / closes" ✗. The model is the existing PMCF §"Residual uncertainties from legacy PMS: confirmation in PMCF" paragraph (PMCF lines 103–113): it already states explicitly that the Plus PMCF programme "confirms and strengthens" the legacy evidence and that the legacy evidence "is not treated as a pre-cert gap to be filled or closed". The D.1/D.2 rewrite mirrors this discipline.

---

C1 — PMCF replacement for Gap 4 and Gap 5 at lines 54–55

Current text (DELETE):

• Gap 4: Autoimmune diseases evidence coverage. The pre-market evidence portfolio contains insufficient direct evidence for autoimmune skin conditions (3% of dermatological presentations). The gap is declared acceptable per MDCG 2020-6 § 6.5(e) and is addressed through Activity D.1.
• Gap 5: Genodermatoses evidence coverage. No genodermatoses cases appear in the pre-market clinical evidence portfolio (1% of presentations). The gap is declared acceptable per MDCG 2020-6 § 6.5(e) and is addressed through Activity D.2.

Replacement text (INSERT in place):

• Low-prevalence sub-indication categories — autoimmune dermatoses and genodermatoses (not declared as §6.5(e) acceptable gaps). Pre-certification evidence for these two low-prevalence sub-indication categories (autoimmune dermatoses ~3 %; genodermatoses ~1 % of real-world dermatological presentations) is triangulated under MDCG 2020-6 §6.3 from (i) the Pillar 1 Valid Clinical Association structured literature review appended to R-TF-015-011 State of the Art §Autoimmune and genodermatoses (22 load-bearing anchors CRIT1–7 ≥ 15/21); and (ii) the Pillar 2 Technical Performance per-epidemiological-group V&V in R-TF-028-006 §Per-Epidemiological-Group Performance (autoimmune AUC 0.948 with 95 % CI 0.941 – 0.954, N = 2,040, 38 classes; genodermatoses AUC 0.905 with 95 % CI 0.886 – 0.924, N = 391, 31 classes; both above the pre-specified ≥ 0.80 acceptance criterion inherited from R-TF-028-002 AI Development Plan). Activity D.1 (autoimmune) and Activity D.2 (genodermatoses) of this PMCF Plan confirm and strengthen this pre-certification base in real-world deployment; they do not generate pre-certification evidence and are not invoked to fill or close a pre-certification gap.

(If a numbering update is required for the surrounding list: Gaps 1–3 remain as numbered; former Gaps 4–5 are replaced by a single unnumbered bullet or renumbered item titled "Low-prevalence sub-indication categories (§6.3 triangulated evidence)".)

---

C2 — PMCF replacement for Coverage-of-Evidence-Gaps block at lines 89–92

Current text (DELETE):

Coverage of Evidence Gaps (MDCG 2020-6 § 6.5(e)):

• Gap 4 (Autoimmune diseases): Addressed by Activity D.1 (prospective surveillance, 50-case target, Top-3 accuracy ≥ 60%).
• Gap 5 (Genodermatoses): Addressed by Activity D.2 (passive surveillance, safety-trigger-based, zero-harm criterion).

Replacement text (INSERT in place):

Coverage of low-prevalence sub-indication categories (MDCG 2020-6 §6.3 triangulated pre-certification evidence + PMCF confirmation):

• Autoimmune dermatoses: pre-certification evidence via Pillar 1 (R-TF-015-011 §Autoimmune and genodermatoses) and Pillar 2 (R-TF-028-006 §Per-Epidemiological-Group Performance); PMCF confirmation via Activity D.1 (prospective surveillance, 50-case target, Top-3 accuracy ≥ 60 %, interim analyses at 12 and 36 months post-certification).
• Genodermatoses: pre-certification evidence via Pillar 1 (R-TF-015-011 §Autoimmune and genodermatoses) and Pillar 2 (R-TF-028-006 §Per-Epidemiological-Group Performance); PMCF confirmation via Activity D.2 (passive surveillance with zero-harm safety trigger and a 30-case cumulative coverage trigger). Active prospective recruitment is not methodologically appropriate for this category; see Activity D.2 §No active recruitment.

Also update (adjacent): the paragraph at PMCF line 66 currently reads

Activities D.1 and D.2 address the two evidence coverage gaps declared acceptable in the CER per MDCG 2020-6 § 6.5(e). These activities do not generate pre-market evidence — they are prospective and passive surveillance activities respectively. Their function is to confirm, through real-world deployment monitoring, that the acceptability justifications made in the CER (limited clinical role, no acute mortality risk, physician final decision) hold in practice. If either activity's surveillance trigger is met, the PMCF program initiates corrective action including an unscheduled CER update.

Replacement:

Activities D.1 and D.2 confirm and strengthen the triangulated pre-certification evidence declared sufficient in the CER (MDCG 2020-6 §6.3) for the autoimmune dermatoses and genodermatoses sub-indication categories — Pillar 1 Valid Clinical Association in R-TF-015-011 §Autoimmune and genodermatoses and Pillar 2 Technical Performance in R-TF-028-006 §Per-Epidemiological-Group Performance. These activities do not generate pre-certification evidence and are not invoked to fill or close a pre-certification evidence gap. Their function is to confirm, through real-world deployment monitoring, that the §6.3 sufficient-evidence determination holds in routine clinical practice. If either activity's surveillance trigger is met, the PMCF programme initiates corrective action including an unscheduled CER update and a protocol-driven re-review of the §6.3 determination.

---

C3 — PMCF replacement for Activities D.1 and D.2 (lines 314–345)

Heading replacement at line 314–316

Current heading block:

D. Consolidated CER Gaps 4 & 5: Evidence Coverage for Autoimmune and Genodermatoses Conditions (Clinical Benefit 7GH, sub-criterion a — indication coverage)

Per MDCG 2020-6 § 6.5(e), two epidemiological categories of dermatological disease are declared as acceptable evidence gaps in the Clinical Evaluation Report (Gaps 4 and 5 in R-TF-015-003 § "Need for more clinical evidence"). Both gaps affect clinical benefit 7GH (Diagnostic Accuracy), specifically the breadth of indication coverage under sub-criterion (a). These gaps are declared acceptable based on the arguments documented in the CER: the rarity of these conditions, their diagnostic reliance on serological or genetic testing beyond visual assessment, and the absence of acute mortality risk from an incorrect visual ranking. The activities below are designed to monitor these gaps prospectively and retrospectively, ensuring that real-world deployment data does not reveal unexpected safety concerns or systematic misclassification in these categories.

Replacement heading block:

D. Post-certification confirmation of low-prevalence sub-indication categories (autoimmune dermatoses and genodermatoses) (Clinical Benefit 7GH, sub-criterion a — indication coverage)

For the two low-prevalence sub-indication categories of autoimmune dermatoses (~3 % of real-world dermatological presentations) and genodermatoses (~1 %), pre-certification evidence is triangulated under MDCG 2020-6 §6.3 from Pillar 1 Valid Clinical Association (R-TF-015-011 §Autoimmune and genodermatoses — 22 load-bearing literature anchors CRIT1–7 ≥ 15/21) and Pillar 2 Technical Performance (R-TF-028-006 §Per-Epidemiological-Group Performance — autoimmune AUC 0.948 with 95 % CI 0.941 – 0.954, genodermatoses AUC 0.905 with 95 % CI 0.886 – 0.924; both above the pre-specified ≥ 0.80 acceptance criterion). Both categories remain within the intended use. The device's clinical role for these categories is supporting the healthcare professional's differential-diagnosis workup (image-based probability ranking within the broader ICD-11 output distribution); final diagnosis relies on clinical evaluation and histopathological, serological or genetic testing as per the current standard of care, and is an HCP determination, not a device determination. Activities D.1 and D.2 below confirm and strengthen the triangulated pre-certification base in routine clinical deployment; they are not invoked to fill or close a pre-certification evidence gap.

Activity D.1 full replacement (lines 318–331)

Replacement text:

Activity D.1: Prospective surveillance of autoimmune dermatoses in clinical deployment (§6.3 PMCF confirmation)

• Code: PMCF-AutoImmune-Coverage-2026
• Rationale: Confirms and strengthens the triangulated pre-certification evidence for autoimmune dermatoses (Pillar 1 Valid Clinical Association in R-TF-015-011 §Autoimmune and genodermatoses; Pillar 2 Technical Performance in R-TF-028-006 §Per-Epidemiological-Group Performance). This activity is pre-specified under MDCG 2020-6 §6.3 and does not generate pre-certification evidence; it is not invoked to fill or close a pre-certification evidence gap.
• Methodology: Prospective, observational, real-world data collection from clinical sites deploying the device. When a healthcare professional confirms an autoimmune skin-condition diagnosis (using serological testing or biopsy, where clinically indicated), the case is flagged for retrospective analysis of the device's output. The device's probability distribution for that image is reviewed to determine whether the correct autoimmune category was ranked within the Top-1, Top-3, and Top-5 results, and the HCP's use of that ranking in the device-aided differential-diagnosis workup is recorded per protocol. No intervention is required from the HCP beyond standard clinical practice; the review is conducted by the clinical evaluation team.
• Conditions in scope: Bullous pemphigoid, cutaneous lupus erythematosus, dermatomyositis, morphea, pemphigus foliaceus, pemphigus vulgaris (tracked separately from Tier 2 rare-diseases analysis — see note below), mucous membrane pemphigoid, lichen planus (cutaneous and nail; oral lichen planus is specifically within scope as the residual Pillar 1 coverage item), cutaneous vasculitis, and any other ICD-11 autoimmune skin condition confirmed in clinical practice. Note: pemphigus vulgaris data that has already been counted as Tier 2 rare-diseases evidence is not double-counted here; the activity reports the union of the autoimmune-specific cohort and the double-count-free Tier 2 overlap separately.
• Diagnosis confirmation standard: Diagnoses must be confirmed by a dermatologist. For bullous diseases, serological confirmation (anti-desmoglein antibodies for pemphigus; anti-BP180/BP230 for bullous pemphigoid) is required where clinically performed. For lupus and dermatomyositis, ANA panel results are required where available. For lichen planus, clinicopathological correlation is recorded where available.
• Sample size and enrolment target: 50 confirmed autoimmune cases across all in-scope conditions within 36 months of certification (first interim at 12 months post-certification or at 15 cases, whichever comes first). The 50-case target is not a statistical-power specification; it is the minimum evidence threshold at which systematic misclassification (autoimmune categories consistently ranked outside Top-5) can be distinguished from random variation across ≥ 5 autoimmune sub-conditions. Rationale: 50 cases distributed across ≥ 5 autoimmune sub-conditions yields a minimum per-sub-condition cell count of approximately 10, sufficient to detect a systematic ≥ 30 percentage-point drop from the V&V-demonstrated Top-5 performance (0.891) at p < 0.05 under a simple one-sample proportion comparison. At the observed ~3 % prevalence and the expected deployed image volume across participating sites, the target is achievable within the specified window without active recruitment.
• Acceptance criteria and triggers:
  • Primary safety-floor acceptance criterion: Top-3 accuracy for the confirmed autoimmune category ≥ 60 % on the 50-case dataset. Rationale: 60 % is the safety-floor Top-3 threshold below which the device's supporting role in the HCP's differential-diagnosis workup for these sub-categories (where the final diagnosis relies on serological or histopathological testing) would be methodologically compromised. It is deliberately set below — and is not a target for — the pre-certification V&V-demonstrated Top-3 on the autoimmune sub-analysis (0.820, 95 % CI 0.803 – 0.836, per R-TF-028-006 §Per-Epidemiological-Group Performance); the target performance remains the V&V-demonstrated value.
  • Non-inferiority secondary acceptance criterion: Top-3 accuracy post-certification must not fall more than 15 percentage points below the V&V-demonstrated Top-3 of 0.820 (i.e., must remain ≥ 0.67 on the 50-case cohort). This criterion is the substantive post-market performance-maintenance check; breach triggers an unscheduled CER update and a protocol-driven re-review of the §6.3 sufficient-evidence determination on the same footing as the primary safety-floor criterion.
  • User-concordance supporting criterion: in ≥ 70 % of the confirmed autoimmune cases where the correct category was within the device's Top-5 ranking, the HCP's recorded differential-diagnosis workup reflects consultation of the device's ranking (per-protocol measurement at case review, supporting criterion only — does not trigger unscheduled CER update on its own).
  • Safety criterion: zero confirmed autoimmune conditions where the device ranked all autoimmune categories below Top-10 AND the physician subsequently reported that the device output contributed to a clinically significant delay in diagnosis.
  • Surveillance trigger (unscheduled CER update): if at any annual interim review more than 20 % of confirmed autoimmune cases have the correct category ranked below Top-5, OR the non-inferiority secondary criterion is breached, OR the primary safety-floor is breached, an unscheduled CER update and a protocol-driven re-review of the §6.3 sufficient-evidence determination must be initiated.
• Timeline: Intended start: upon CE marking. Data collection: continuous. First interim analysis: 12 months post-CE marking (or at 15 cases, whichever comes first). Target completion: 50 confirmed cases accumulated or 36 months post-CE marking, whichever comes first.
• Legacy post-market report corpus slice (pre-specified concurrent confirmation). The autoimmune-dermatoses slice of the ≈ 250,000 legacy-predecessor post-market report corpus will be analysed and reported as part of the first PMS Update Report R-TF-007-003, consistent with MDCG 2020-6 §6.3 under which PMCF confirms and strengthens an adequately-evidenced pre-certification base and is not invoked to fill or close pre-certification evidence gaps. The slice is queried for autoimmune ICD-11 codes across the multi-year deployment window, deduplicated, case-categorised and summarised with complaint and incident rates (rule-of-three upper bounds where zero events); the results feed the R-TF-007-003 first update and, via that update, back into the CER.
• Reporting: Results integrated into the annual PMCF Evaluation Report and the corresponding CER update.

Activity D.2 full replacement (lines 333–345)

Replacement text:

Activity D.2: Passive surveillance of genodermatoses in post-market deployment (§6.3 PMCF confirmation)

• Code: PMCF-Genodermatoses-Surveillance-2026
• Rationale: Confirms and strengthens the triangulated pre-certification evidence for genodermatoses (Pillar 1 Valid Clinical Association in R-TF-015-011 §Autoimmune and genodermatoses; Pillar 2 Technical Performance in R-TF-028-006 §Per-Epidemiological-Group Performance). Pre-certification evidence is judged sufficient under MDCG 2020-6 §6.3; this activity is post-certification confirmation and is not invoked to fill or close a pre-certification evidence gap. Active prospective recruitment is not methodologically appropriate for this category at the observed ~1 % prevalence and given that genodermatosis diagnosis is based on genetic testing and clinical history rather than image-based assessment; the device's clinical role for these conditions is supportive only (probability ranking within the broader ICD-11 output distribution).
• Methodology: Passive surveillance. Any case reported through the PMS system (complaint, vigilance report, user feedback, or clinical site communication) where the patient's confirmed diagnosis is a genodermatosis (epidermolysis bullosa, ichthyosis vulgaris, lamellar ichthyosis, Darier disease, Hailey-Hailey disease, neurofibromatosis type 1 and type 2 cutaneous manifestations, tuberous sclerosis complex cutaneous manifestations, or similar) is captured and retrospectively reviewed. The review examines: (a) what the device output was for that image; (b) whether the correct genodermatosis category was present in the Top-5 of the probability distribution; (c) whether the HCP reported any clinical harm attributable to the device output.
• No active recruitment: Active recruitment for genodermatoses is explicitly not conducted, for the methodological reasons recorded in the Rationale above. Passive surveillance is the methodologically appropriate and proportionate choice.
• Sample size: No pre-specified enrolment target. The activity is governed by safety and coverage triggers (see below).
• Acceptance criteria and triggers:
  • Safety criterion (primary): zero confirmed genodermatosis cases where the device output was identified by the treating HCP as contributing to patient harm. Any such case invalidates the §6.3 sufficient-evidence determination for this category and requires immediate CER update and CAPA.
  • Per-case Top-5 concordance performance reporting: for each confirmed genodermatosis case identified through PMS, the per-case Top-5 ranking of the correct genodermatosis category is recorded and reported. The aggregate per-case Top-5 concordance distribution is published in each annual PMCF Evaluation Report, including year-over-year trend analysis and a reviewer-adjudicated narrative concordance score per case. This is a positive performance-confirmation element; it complements rather than replaces the early Pillar 3-equivalent performance readout on the legacy post-market report corpus (see below).
  • User-concordance supporting criterion: for each identified case, the HCP's recorded differential-diagnosis workup is examined for consultation of the device's Top-5 ranking; a narrative user-concordance note is included in the annual PMCF Evaluation Report (supporting only — does not trigger unscheduled CER update on its own).
  • Surveillance trigger (unscheduled clinical-evaluation review): if more than 3 confirmed genodermatosis cases are identified through PMS in any 12-month period AND the device ranked all genodermatosis categories below Top-5 for those cases, an unscheduled clinical-evaluation review must be initiated to reassess whether the §6.3 sufficient-evidence determination holds.
  • Coverage trigger (formal diagnostic-accuracy analysis + §6.3 reconfirmation): if at any annual review the cumulative number of genodermatosis cases in real-world deployment reaches 30, a formal diagnostic-accuracy analysis is conducted and the results are fed back into the §6.3 sufficient-evidence determination in the CER (either reconfirming the determination with new evidence, or initiating active recruitment as a consequent corrective action).
• Timeline: Intended start: upon CE marking. Surveillance: continuous throughout device lifetime. Annual review: incorporated into the PMCF Evaluation Report and PSUR.
• Early Pillar 3-equivalent performance readout (legacy post-market report corpus slice). To avoid deferring all positive Pillar 3-equivalent performance confirmation for genodermatoses to the 30-case cumulative coverage trigger, the genodermatoses slice of the ≈ 250,000 legacy-predecessor post-market report corpus is pre-specified as an early confirmation readout and is analysed and reported in the first PMS Update Report R-TF-007-003, targeting a publication window of approximately six months post-CE marking. The readout is not a complaint-and-incident summary alone: it reports the per-case Top-5 ranking of confirmed genodermatosis cases in the legacy corpus, a case-categorised diagnostic-accuracy summary (sensitivity / specificity / Top-N distribution where the ground-truth category can be established from the legacy clinical record), rule-of-three upper bounds for zero-event safety categories, and a narrative concordance note. The results concurrently anchor the positive performance-confirmation element of this Activity D.2, the Test 4 confirmation of the §6.3 sufficient-evidence determination in R-TF-015-003 §Representativeness of the Study Populations, and — via R-TF-007-003 — the PMS / PMCF closure of the CER. The readout is consistent with MDCG 2020-6 §6.3 under which PMCF confirms and strengthens an adequately-evidenced pre-certification base and is not invoked to fill or close pre-certification evidence gaps.
• Reporting: Any genodermatosis case identified through PMS is reported in the next annual PMCF Evaluation Report. If the safety trigger is met, an unscheduled PMCF Evaluation Report is issued within 30 days.

---

C4 — Legacy PMS slice commitment (stand-alone commitment paragraph)

The legacy-PMS-slice commitment is pre-specified as a bullet inside each of Activities D.1 and D.2 (see C3 above — the "Legacy post-market report corpus slice (pre-specified concurrent confirmation)" bullet). This is the intended location per MDCG 2020-6 §6.3 (the confirmation activity is tied to its target category).

Optional additional paragraph for PMCF Plan §"Residual uncertainties from legacy PMS: confirmation in PMCF" (i.e., slot between lines 113 and 115, just before §"Evidence-quality substantiation: continuity from legacy PMS"):

The autoimmune-dermatoses and genodermatoses slices of the legacy-predecessor post-market report corpus (≈ 250,000 diagnostic reports over ≥ 4 years of continuous deployment of the legacy device) are additionally committed for analysis and reporting in the first PMS Update Report R-TF-007-003 under MDCG 2020-6 §6.3. Each slice is filtered by autoimmune or genodermatosis ICD-11 codes, deduplicated, case-categorised, and summarised with complaint and incident rates including rule-of-three upper bounds for zero-event categories. The results concurrently anchor the confirmation component of Activities D.1 and D.2 and the §6.3 sufficient-evidence determination in the CER; they do not generate pre-certification evidence.

---

Review notes for this file

• Audit-deliverable-reviewer concern: check wording discipline — "confirms / strengthens" ✓, "fills / closes" ✗. Check no engineering-layer leaks (no task-3b5, no bsi-non-conformities, no .ts / .json / React component names, no npm / git references). Check that every regulatory cross-reference is to the MDR / MDCG / ISO / IEC / MEDDEV standard or to an R-TF record identifier, not to an internal file path.
• BSI clinical-auditor concern: check that §6.3 is invoked with the correct wording (sufficient clinical evidence; PMCF confirms and strengthens) and that §6.5(e) is not invoked for these two categories. Check that Activities D.1/D.2 retain statistically defensible thresholds (primary Top-3 ≥ 60 % for D.1, safety zero-harm and >3-case 12-month / 30-case cumulative triggers for D.2) and that the user-concordance supporting criterion is clearly flagged as supporting-only (not a stand-alone unscheduled-CER trigger).
• Celine clinical-consultant concern: check pillar mapping — Pillar 1 (VCA literature) and Pillar 2 (analytical performance) are the anchors; PMCF confirmation is Pillar 3 real-world evidence. Check Rank-vs-Pillar orthogonality (Pillar 1 anchors score at Rank 6; Pillar 2 analytical performance sits with the AI-model V&V evidence at Rank 12; PMCF Activity D.1 prospective real-patient evidence populates Rank 3 post-certification — prospective non-randomised clinical investigations — or, if the activity is scoped as proactive post-market surveillance rather than as a formal clinical investigation, Rank 8). Check that the D.1/D.2 rewrite does not accidentally suggest that these activities substitute for pre-certification evidence (they do not — Pillar 1 + Pillar 2 carry the §6.3 sufficient-evidence determination, PMCF confirms).