Research and planning

Internal working document

This document is for internal use only. It contains analysis, gap identification, and response strategy for Item 6a of the BSI Clinical Review Round 1. It will not be included in the final response to BSI.

1. What BSI is asking

BSI's clinical reviewer found that the PMCF Plan (R-TF-007-002) describes 9 specific PMCF activities but provides insufficient detail for assessment. Specifically, BSI raises four concerns:

1. Justification/rationale: Why is each activity needed? Several reference "teledermatology" which does not appear to be a device feature.
2. Insufficient detail: Methods, sample sizes with justification, acceptance criteria, and timelines are not fully described for each activity.
3. Device name confusion: Activities reference "Legit.Health" rather than "Legit.Health Plus" — unclear which device is being studied.
4. Missing equivalent device evaluation: No assessment of clinical data from equivalent or similar devices.

BSI requests "a full and comprehensive description of each activity (including any supporting documents such as plans and protocols)."

This is a deficiency finding, not an observation. The regulatory basis is Annex XIV 5, 6.1, 6.2, and MDCG 2020-7.

2. Root cause analysis

The PMCF Plan was written as a high-level summary of planned activities. For each study, it provides a code, type, start date, and primary endpoint — but omits the level of detail that MDCG 2020-7 expects:

MDCG 2020-7 expectation	Present in R-TF-007-002?
Rationale for each activity	No — activities are listed under CER gap headings but individual justification is missing
Detailed methodology	No — only study type is stated (e.g., "Observational and Retrospective")
Sample size with justification	Partial — only A.1 states a sample size (30,000 images); others omit it entirely
Acceptance criteria	Partial — only C.1 states explicit thresholds (AUC >0.8, Top-5 >70%, etc.)
Timeline (start and end)	Partial — only start dates are given; no expected completion dates
Link to specific CER gap	Partial — activities are grouped by CER gap but the link is implicit, not explicit
Device identification	Inconsistent — some say "Legit.Health", some say "Legit.Health Plus", some omit the device name

The root cause is that the PMCF Plan was drafted as a planning summary rather than the detailed methodological document MDCG 2020-7 requires.

3. Activity-by-activity gap analysis

Group A: Triage and Malignancy Prioritization (CER Gap 1)

Activity	Sample size	Acceptance criteria	Method detail	Device name	Study status
A.1: Automated triage in teledermatology	30,000 images	Not defined	Retrospective — no further detail	"Legit.Health"	Planned (Feb 2026)
A.2: Melanoma follow-up prioritisation	Not stated	Not defined	Prospective, single-centre, interventional — no further detail	Not specified	Planned (Jan 2026)
A.3: Pilot triage in teledermatology	Not stated	Not defined	Prospective, multicentre, interventional — no further detail	"Legit.Health"	Planned (Mar 2026)

Teledermatology concern: BSI notes that "several seem to be collecting data on teledermatology, which does not seem to be a feature of this device." This is a terminology issue. The device is an API that processes clinical images and returns diagnostic support results. "Teledermatology" is not a device feature but rather a clinical workflow in which the device is used as a decision support tool. The device's intended purpose includes processing dermatological images regardless of whether the image was taken in-person or remotely. The PMCF Plan should clarify that teledermatology is a use scenario (the clinical context in which the device operates) rather than a device feature, and that these studies evaluate the device's core triage capabilities within that workflow.

Group B: Severity Assessment and Monitoring (CER Gap 2)

Activity	Sample size	Acceptance criteria	Method detail	Device name	Study status
B.1: FFA severity/monitoring	Not stated	Not defined	Observational, prospective — no further detail	Not specified	Planned (Apr/May 2026)
B.2: Acne severity/monitoring	Not stated	ICC agreement	Single-centre, interventional — no further detail	Not specified	Planned (Mar 2026)
B.3: Atopic dermatitis EASI scoring	Not stated	ICC agreement	Retrospective, single-centre — no further detail	Not specified	Planned (Feb 2026)
B.4: Vitiligo AVASI validation	Not stated	ICC agreement	Prospective, multicentre, observational — no further detail	"Legit.Health Plus"	Planned (Mar 2026)

ICC threshold gap: B.2, B.3, and B.4 mention ICC as the agreement measure but do not define what ICC value constitutes acceptable agreement. An ICC >0.75 is generally considered "good" agreement — an explicit threshold should be stated for each activity.

Group C: Core Diagnostic Performance (CER Gap 3)

Activity	Sample size	Acceptance criteria	Method detail	Device name	Study status
C.1: Non-interventional performance analysis	Not stated	AUC >0.8; Top-5 >70%; Top-3 >55%; Top-1 >40%	Reader study — no further detail	Not specified	Planned (Nov 2026)
C.2: FDA pivotal real-world performance	30 readers	Top-1 and Top-3 accuracy vs unassisted	Multireader multicase — no further detail	"Legit.Health Plus"	Planned (Jun 2026)

New detail for C.2 from FDA Strategy (March 2026):

• Sample size: 30 board-certified PCPs (readers) evaluating a subset of 100 cases (50 malignant, 50 benign) from the standalone study.
• Methodology: Prospective, blinded, sequential-read Multi-Reader Multi-Case (MRMC) study.
• Acceptance criteria: Superior diagnostic sensitivity of PCPs when aided by the device compared to their unaided clinical assessment (the adjunctive benefit). Measuring Corrective Direction and Clinical Utility.
• Timeline: Execution Nov 2026 - Feb 2027. CSR in March 2027.

New Activity C.3 Proposal (to address BSI camera concern):

• Title: System Reproducibility and Robustness Study.
• Rationale: To prove that diagnostic sensitivity is not altered by the image capture device (various smartphones and dermatoscopes), addressing BSI's concern about dermatoscopic-only study data.
• Sample size: At least 50 prospectively captured lesions.
• Methodology: Prospective, observational, cross-sectional method-comparison study utilizing a "Repeated Measures" design.
• Acceptance criteria: Overall Percent Agreement (OPA) point estimate of >= 95% between different validated acquisition systems.

4. CEP discrepancy

The Clinical Evaluation Plan (R-TF-015-001) states:

"At this point of the clinical evaluation, the manufacturer does not plan to conduct a new PMCF study."

This directly contradicts the PMCF Plan which describes 9 specific activities. The CEP was written before the PMCF Plan was finalised and needs to be updated to reference the approved 9-activity (now 10-activity) PMCF plan.

5. Cross-NC connections

Clinical Review Item 5 — PMS Plan

Item 5 research identified that the PMCF Plan is a sub-component of the PMS system. The PMS Plan (R-TF-007-001) has been updated with heading-level references to the PMCF Plan. The PMCF Plan's general methods section explicitly defers to R-TF-007-001 for detail on routine data collection — this cross-reference should be maintained.

Clinical Review Item 3a — Clinical data analysis

Item 3a research identified that the CER does not integrate legacy PMS data despite 4+ years of market experience. The PMCF Plan's activities will generate additional clinical evidence that feeds back into the CER. The PMCF Plan should explicitly state how PMCF results will be integrated into future CER updates, closing the loop on the "continuous process" requirement of Annex XIV 5.

Clinical Review Item 3b — Data sufficiency justification

Item 3b research maps to the PMCF Plan's CER gaps. The three gaps identified in the PMCF Plan (triage, severity assessment, core performance) correspond to areas where 3b identified insufficient evidence. Strengthening the PMCF Plan's activity descriptions directly supports the data sufficiency argument.

Clinical Review Item 2 — Device description & intended purpose

BSI's concern about "teledermatology" in PMCF activities links to Item 2's question about intended purpose. The response to 6a's teledermatology concern should be consistent with how Item 2 describes the device's intended use scenarios.

Clinical Review Item 1 — CER update frequency

The PMCF Plan feeds into CER updates. Item 1 addresses CER update frequency. The PMCF Plan's estimated evaluation report completion (January 2027) should align with the CER update schedule.

6. Response strategy

The response should:

1. Provide expanded descriptions for each activity — for each of the 9 activities, include: rationale, detailed methodology, sample size with justification, acceptance criteria, expected completion date, and explicit mapping to the CER gap it addresses
2. Clarify the teledermatology issue — explain that teledermatology is a clinical workflow/use scenario, not a device feature, and that the device's core image analysis capabilities are what's being evaluated within that context
3. Fix device name references — confirm that all activities evaluate the MDR device (Legit.Health Plus) and correct any inconsistent naming
4. Add equivalent/similar device evaluation — include a section assessing clinical data from comparable AI dermatology devices (this may be brief since the device is novel, but the section must exist per Annex XIV 6.2)
5. Reference supporting documents — for activities where separate protocols exist, reference them

Fixes required (in R-TF-007-002)

Fix 1: Expand each activity description

For each of the 9 activities (A.1–A.3, B.1–B.4, C.1–C.2), add:

• Rationale: Why this activity is needed, linked to the specific CER gap
• Detailed methodology: Study design, data collection procedures, analysis methods
• Sample size with justification: Number of subjects/images, statistical power rationale
• Acceptance criteria: What constitutes a successful outcome (e.g., ICC >0.75, AUC >0.8)
• Expected completion date: Not just start date
• Device identification: Confirm "Legit.Health Plus (version 1.1.0.0)" for all activities

Fix 2: Add teledermatology clarification

Add a paragraph explaining that "teledermatology" in activities A.1–A.3 refers to the clinical workflow context in which the device operates, not a device feature. The device processes clinical images for diagnostic support; in the teledermatology use scenario, these images are captured and transmitted remotely.

Fix 3: Correct device name references throughout

Replace all instances of "Legit.Health" (without "Plus") with the correct MDR device name in activity titles and descriptions, or add a statement clarifying that "Legit.Health" in study codes/titles refers to the Legit.Health Plus device.

Fix 4: Add equivalent/similar device data section

Add a section evaluating clinical data from equivalent or similar AI-based dermatology devices, per Annex XIV 6.2.

Fix 5: Remove backtick formatting in references section

Lines 179–186 of R-TF-007-002 use backtick code formatting for document references. Replace with plain text (same fix applied to R-TF-007-001 in Item 5).

Fix 6: Update CEP

In R-TF-015-001, update the PMCF section to reflect the approved 9-activity PMCF plan, removing the "does not plan to conduct a new PMCF study" statement.

7. Risk assessment

Risk	Impact	Mitigation
BSI may find the expanded activity descriptions still insufficient	Medium — MDCG 2020-7 is prescriptive about PMCF plan content	Include all elements from MDCG 2020-7 §7.2 for each activity
BSI may question why PMCF activities haven't started yet (all planned for 2026)	Low — the device is seeking initial CE mark; PMCF is inherently post-market	Note that timelines begin from CE marking
BSI may probe the teledermatology issue further in Item 2	Medium — if teledermatology isn't in the intended purpose, BSI may question its presence in PMCF	Coordinate response with Item 2 to ensure consistency
CEP discrepancy ("no new PMCF study") may raise credibility concerns	Medium — BSI may see this as contradictory documentation	Fix CEP proactively and note the update in the response

8. Open items

#	Item	Owner	Status
1	Obtain sample size calculations or justifications for activities A.2, A.3, B.1–B.4, C.1, C.2	Jordi	Complete
2	Confirm whether separate protocols exist for any of the 9 activities (BSI asks for "supporting documents such as plans and protocols")	Jordi	Complete
3	Define ICC acceptance thresholds for B.1–B.4 activities	Jordi	Complete
4	Define acceptance criteria for A.1–A.3 activities	Jordi	Complete
5	Confirm expected completion dates for all 9 activities	Jordi	Complete
6	Confirm device version for each activity	Jordi	Complete
7	Identify equivalent/similar AI dermatology devices and their published clinical data	Jordi	Complete
8	Resolve teledermatology terminology with Item 2 response	Taig/Jordi	Complete
9	Create investigation folders and add references to PMCF Plan	Gemini	Complete

---

Regulatory framework: what the BSI meeting revealed

Severity warning

Nick stated during the BSI meeting (2026-03-25) that refusal is extremely likely unless all PMCF gaps are closed. The PMCF Plan is assessed under CEAR Section F, which is a blocking section. Nick's specific condition for CE mark approval was that the PMCF plan must be "robust" — meaning BSI must be able to assess whether each activity will credibly confirm the pre-market evidence, not just that activities are planned. Item 6a directly determines whether BSI can make that assessment.

The four applicable guidance documents

Document	Role for Item 6a
MDCG 2020-7	The primary PMCF guidance document. Section 7.2 lists the required content for each PMCF activity: objective, design and method, population with inclusion/exclusion criteria, sample size with statistical justification, acceptance criteria, timeline (start AND end), and link to specific CER gap or risk. The current PMCF Plan (R-TF-007-002) provides a study type and start date — it omits every other required element. BSI's 4 concerns all trace to MDCG 2020-7 § 7.2 requirements. The fix for each activity must specifically address each § 7.2 element, not just add narrative detail.
MDCG 2020-6, § 6.4 and § 6.5(e)	Two obligations apply. First, § 6.4: PMCF activities cannot be framed as filling pre-market clinical gaps. They must be framed as CONFIRMING conclusions already supported by sufficient pre-market evidence. This means the rationale for each activity (Fix 1) must not say "this activity fills Gap X" — it must say "this activity confirms the clinical performance already demonstrated in [study], extending monitoring to [specific dimension]." Framing activities as gap-fillers tells BSI that the pre-market evidence is insufficient, which triggers a separate finding under MEDDEV A7.4. Second, § 6.5(e): where gaps have been explicitly declared (Gaps A and B — autoimmune and genodermatoses conditions), specific PMCF activities must be named. Activities D.1 and D.2 (documented in issue-6-pmcf.mdx) are the mandatory PMCF activities for these declared gaps. They are not optional additions — they are the regulatory basis for the declaration that Gaps A and B are acceptable.
MDCG 2020-1, continuous monitoring pillar	For MDSW, MDCG 2020-1 requires ongoing monitoring across all three pillars: Valid Clinical Association, Technical Performance, and Clinical Performance. PMCF activities must collectively address all three pillars — not just Clinical Performance. Activity C.1 (non-interventional performance analysis) and C.2 (MRMC study for FDA) address Clinical Performance. The PMCF Plan must also show how VCA and Technical Performance are monitored continuously. This is relevant to how the teledermatology activities (A.1–A.3) are framed: they do not just study teledermatology workflows — they monitor whether the Valid Clinical Association holds in real-world deployment conditions (varied smartphone cameras, varied lighting, remote image capture). This framing resolves BSI's teledermatology concern while simultaneously satisfying MDCG 2020-1's continuous monitoring requirement.
MDCG 2020-13, Section F	BSI's CEAR Section F explicitly checks: "PMCF Plan reviewed; PMCF adequacy or absence justification; PMCF activities described with sufficient detail for assessment." Section F is the blocking gate. A PMCF Plan that lists activities without the elements required by MDCG 2020-7 § 7.2 will fail Section F. Each activity expansion in Fix 1 must produce a description that satisfies Section F independently — meaning a BSI reviewer reading only the PMCF Plan can assess whether the activity is methodologically credible without needing to consult supporting documents.

The § 6.4 reframing requirement — the most critical regulatory implication

The current PMCF activities are grouped under "CER Gap 1," "CER Gap 2," and "CER Gap 3." This framing tells BSI that these gaps remain unfilled by pre-market evidence — directly triggering MEDDEV A7.4 ("if clinical data lacking, conformity is not fulfilled"). Nick's meeting statement confirms BSI interprets gap-labelled PMCF activities as evidence of pre-market insufficiency.

The fix is not cosmetic. Every activity must be reframed:

Current framing	Correct § 6.4 framing
"A.1: Address CER Gap 1 — triage performance needs further data"	"A.1: Confirm triage performance demonstrated in DAO_O 2022 and DAO_PH 2022 under real-world deployment conditions, extending to 30,000 image-scale monitoring"
"B.1: Address CER Gap 2 — severity assessment evidence is limited"	"B.1: Confirm severity assessment demonstrated in AIHS4 2025 for FFA, extending to a statistically powered prospective cohort"
"C.1: Address CER Gap 3 — core performance requires confirmatory data"	"C.1: Continuously monitor core diagnostic performance as the device is deployed at scale, confirming AUC >0.8 established in pre-market studies"

This reframing is not evasion — it is the regulatory requirement. Activities that confirm existing evidence are legitimate PMCF. Activities framed as filling gaps are evidence of pre-market insufficiency.

Activities D.1 and D.2: mandatory obligations, not additions

Issue-6-pmcf.mdx documents the two new PMCF activities required for the declared acceptable gaps:

• D.1: Autoimmune skin condition monitoring (Gap A — 3% prevalence category). Methodology, sample size, acceptance criteria, and timeline are fully specified in issue-6-pmcf.mdx. This activity must be added to R-TF-007-002 as a new Activity D.1.
• D.2: Genodermatosis natural history monitoring (Gap B — 1% prevalence category). Full specification also in issue-6-pmcf.mdx.

These are not optional enhancements. Without D.1 and D.2, the § 6.5(e) declared gap strategy fails and Gaps A and B are no longer "declared acceptable" — they are simply unsupported claims. BSI will detect the absence of specific activities for declared gaps.

Nick's C.2 confirmation and MRMC reframing

Nick confirmed during the BSI meeting (2026-03-25) that Activity C.2 (MRMC study for FDA) is an appropriate PMCF activity and is consistent with BSI expectations. However, C.2 must be reframed per MDCG 2020-6 Appendix III Rank 11: MRMC studies are NOT clinical data. C.2 is valid as PMCF evidence for Technical Performance and Clinical Performance under specific conditions — but it cannot serve as the primary clinical evidence for any performance claim. The activity description must make this distinction explicit: C.2 is a confirmatory PMCF activity for the adjunctive benefit claim, using MRMC methodology for the FDA regulatory pathway in parallel with BSI's post-market monitoring.

The teledermatology terminology: the MDCG 2020-1 framing resolves BSI's concern

BSI's concern about "teledermatology" activities (A.1–A.3) is that teledermatology does not appear to be a device feature. The resolution per MDCG 2020-1 is:

Teledermatology is a deployment context in which the Valid Clinical Association (VCA) must be confirmed to hold. MDCG 2020-1 requires that the VCA is validated for the intended-use context — which includes the full range of image capture conditions under which the device will be used. When images are captured remotely (varying smartphones, varying lighting, no clinician at point of capture), the VCA between the image input and the diagnostic output may differ from laboratory conditions. Activities A.1–A.3 monitor this VCA stability across real-world deployment conditions.

This framing must appear verbatim in the activity rationale for A.1–A.3. It resolves BSI's concern without modifying the intended purpose statement and satisfies MDCG 2020-1's continuous monitoring pillar simultaneously.