An assessment organisation can expand internationally without creating a single worldwide approval for medical AI. Founders still need to establish the applicable product classification, responsible entity, evidence requirements and market-access route in each jurisdiction. A familiar assessor or reusable technical file can help, but neither removes those decisions.
Scarlet's 12 October 2026 announcement includes US and Asia-Pacific expansion plans. This article examines the planning implications, not an assumption that those plans confer new authority or that every market will accept the same assessment. Regulatory sources were checked on 13 October 2026.
Start with the market and intended function
A fictional company develops software to support clinicians reviewing wound photographs. Its founders plan to offer it in a European hospital, a British community service and a US outpatient network. The interface may look similar in all three settings, but the proposed uses need to be specified separately.
Who supplies the images? Which patients are included? Does the software organise information, suggest an interpretation or influence an immediate treatment decision? Is the clinician expected to examine the patient before acting? Those questions should be settled before a team assumes one evidence package will support all commercial descriptions.
The MHRA's intended-purpose guidance offers a useful planning discipline for the British proposal. It is not a substitute for applying the relevant European or US rules. Shared clarity about the product is reusable; a jurisdiction-specific regulatory conclusion is not automatically portable.
European conformity and Great Britain market access are related, not identical
The European Commission's notified-body guidance describes designated assessment within the European medical-device framework. Great Britain has its own medical-device framework and UK approved-body arrangements, with specified routes for recognising CE-marked devices under the current conditions described by the MHRA.
A founder should check those conditions for the particular device and route rather than assume that either all CE-marked products are permanently accepted or that every European assessment must be repeated from the beginning. The position is category- and route-dependent.
Northern Ireland also needs separate consideration. The MHRA's UK guidance explains the different applicable arrangements; Great Britain should not be used as a synonym for the whole UK in a market-access plan. Precision here avoids a commercial launch promise that the underlying documentation does not support.
The FDA's third-party programme is not European certification with a US label
The FDA's 510(k) Third Party Review Program provides a voluntary alternative review process for eligible devices. An accredited organisation reviews the submission and sends its review and recommendation to the FDA, which retains the final determination.
That means a third-party review should not be described as the reviewer independently issuing unrestricted FDA approval. It also does not establish that every AI device qualifies for this route. Eligibility and the review organisation's scope must be checked for the specific product.
The FDA's submission instructions make those checks operational: establish device eligibility, identify an appropriate review organisation and agree the review engagement. A company's global brand or expansion announcement is not evidence that it is authorised to review every device type in every market.
Build a reuse matrix instead of a universal file
The following planning matrix is an editorial tool for the fictional wound-support product. It does not determine the regulatory pathway or claim that all listed evidence will be sufficient.
| Workstream | Material that may be reusable | Additional market question |
|---|---|---|
| Architecture and development | Versioned system description and engineering evidence | Does the market version use the same components and permissions? |
| Clinical evaluation | Studies relevant to the intended task | Do population, setting and clinical practice match the proposed claim? |
| User interaction | Evidence about review and error recognition | Are language, training and professional roles comparable? |
| Risk management | Identified failure mechanisms and controls | Do local pathways create new consequences or dependencies? |
| Product information | Core explanation of limits and operation | What labelling, language and market-specific statements are required? |
| Monitoring | Processes for detecting and investigating concerns | Which reporting duties and local contacts apply? |
Reusing evidence should be an explicit judgement. The file should say why a finding applies, what differences remain and what work addresses them. Copying a report into another folder is not itself an applicability assessment.
The clinical service can change even when the model does not
Suppose the fictional British deployment expects a community clinician to interpret images alongside an examination. The US proposal instead allows remote review from images supplied by patients. That changes the available information and potentially the suitability of inputs.
The team should not treat this as merely a commercial localisation. It needs to assess whether the existing evidence supports the revised workflow, whether additional restrictions are needed and whether the intended-use statement has changed.
The good machine learning practice principles provide a cross-jurisdictional technical foundation for considering representative data and relevant use conditions. They encourage good development practice, but do not turn a shared principle into identical legal requirements or a common market-access decision.
Separate product permission from local implementation
Even after establishing a market route, the organisation delivering care needs a workable deployment. In an English NHS service, DCB0129 and DCB0160 distinguish manufacturer and adopter clinical-risk responsibilities.
For the fictional community team, questions could include who checks an unusable image, what happens when the reviewer is unavailable and how a patient reaches face-to-face assessment. Those are implementation questions that remain important even where product evidence is relevant and market-access documentation is in order.
Another country may organise clinical delivery differently. The purpose of international planning is not to impose the English NHS process everywhere, but to identify the local decisions rather than leave them invisible.
A sensible expansion sequence
Our proposed sequence begins with a specific market-use statement, then a route assessment, evidence-gap analysis, dependency review and local implementation plan. Commercial commitments should follow what that work supports.
A founder with closely matching uses may be able to reuse substantial evidence. A founder proposing a different patient population, autonomy level or service model should expect a more substantive reassessment. Neither outcome can be predicted solely from the fact that the underlying model is unchanged.
As described on 13 October 2026, Insights by iatroX supports clinical safety and regulatory readiness. Such support can help organise the questions and evidence, but it should not be mistaken for worldwide legal advice, a guarantee of authorisation or a partnership with Scarlet. The useful ambition is a coherent international product story with explicit local decisions, not a single badge stretched across every market.
Frequently asked questions
Does a European medical-device assessment automatically authorise US sales?
No. The manufacturer must establish and satisfy the applicable US pathway; European evidence may be relevant without replacing that decision.
Does a third-party reviewer make the FDA's final 510(k) decision?
No. Under the FDA programme described here, the reviewer submits its assessment and recommendation, while the FDA retains the final determination.
What can founders reuse when expanding internationally?
Relevant engineering, clinical and risk-management evidence may be reusable where the product and conditions match. The team should document that applicability and identify additional market-specific requirements.