Notified Body Review: Unlocking EU Market Access for AI Health

Listen to this article · 8 min listen

The path to market for AI health tools in Europe increasingly hinges on a single, critical gatekeeper: the Notified Body. For companies working through the complex regulatory field of the Medical Device Regulation (MDR), understanding the Notified Body’s role is not merely a formality. It is central to a successful submission and in the end, market access. This entity is the primary reviewing layer, auditing conformity to the MDR framework and determining whether a technical file meets the stringent requirements for CE marking.

The Notified Body’s Centrality in EU MDR

Unlike the FDA’s direct review process, the European system for higher-risk medical devices, including many AI-driven Software as a Medical Device (SaMD) tools, mandates engagement with a Notified Body. These are independent organizations designated by national authorities to assess the conformity of devices before they can be placed on the EU market. Their role is expansive, encompassing everything from quality management system audits to the detailed review of technical documentation. The MDR significantly elevated the scrutiny applied by these bodies, particularly for software and AI, moving many devices that previously self-certified into a category requiring Notified Body oversight. Plus, for AI-driven SaMD, the EU AI Act, which began applying its high-risk obligations from August 2026 for new products, introduces an additional layer of regulatory requirements that Notified Bodies must also assess, creating a dual compliance framework alongside the MDR. This shift means that a company’s readiness for Europe is, at its core, a demonstration of documented compliance that can withstand an external, expert audit.

Structuring the Technical File for Conformity

A successful European submission is fundamentally a document discipline. The technical file, a complete dossier detailing every aspect of a medical device, must systematically address classification, evidence, and post-market duties. Its structure is not arbitrary. It follows a logical progression designed to demonstrate compliance with the General Safety and Performance Requirements (GSPRs) laid out in Annex I of the MDR. The initial sections of the technical file must clearly establish the device’s classification. Under MDR, AI-powered SaMD often falls into higher risk classes (IIa, IIb, or even III) due to its potential impact on patient outcomes. This classification dictates the specific conformity assessment route and, consequently, the depth of Notified Body involvement. Misclassification can lead to significant delays or outright rejection. Following classification, the file transitions to evidence. This section is where the bulk of development and validation work resides. It includes:

  • Risk Management: A thorough analysis of identified risks, their probability, severity, and the measures implemented to mitigate them. For AI, this extends to algorithmic bias, data privacy, and potential for misinterpretation.
  • Clinical Evaluation: A systematic and planned process to continuously generate, collect, analyze, and assess clinical data pertaining to a device to verify its safety and performance. This often involves clinical investigations, literature reviews, and post-market clinical follow-up data. EU MDR guidance on clinical evaluation
  • Verification and Validation: Documentation proving the software functions as intended and meets user needs. This includes testing protocols, results, and traceability to requirements. For AI, this is critical, often involving strong testing against diverse datasets to demonstrate performance across relevant patient populations.
  • Usability Engineering: Evidence that the device is safe and effective for its intended users and use environments, minimizing use errors.

Finally, the technical file must detail the post-market surveillance plan. This demonstrates a commitment to ongoing monitoring of the device’s performance, collection of real-world data, and proactive identification of any safety or performance issues once the device is on the market. This includes plans for Post-Market Clinical Follow-up (PMCF) and vigilance reporting.

Conformity in Practice: Viz.ai, Butterfly Network, and Tempus AI

The recorded conformity material for companies like Viz.ai, Butterfly Network, and Tempus AI provides instructive examples of how this framework translates into practice. While specific filings are proprietary, the public signals from their regulatory engagements with EU Notified Bodies and adherence to MDR principles illuminate common threads. Viz.ai, for instance, has navigated the European regulatory field with its AI-powered stroke detection and care coordination platform. Their success in obtaining CE marks for various modules indicates a strong technical file that effectively addressed the classification of their SaMD, provided compelling clinical evidence, and outlined clear post-market surveillance strategies. The nature of their AI, which analyzes medical images for critical findings, places a high burden of proof on its performance and reliability, directly impacting patient management. Similarly, Butterfly Network, with its portable ultrasound device and integrated AI capabilities, has demonstrated conformity. The AI components, which assist in image acquisition and interpretation, would have required detailed validation within their technical file, showing how the AI integrates safely and effectively into the overall device functionality. Their approach to regulatory compliance would have hinged on demonstrating the accuracy and safety of these AI features, particularly given the point-of-care nature of their technology. Tempus AI, a company focused on precision medicine through AI-driven data analysis, also operates within this conformity framework for its diagnostic and prognostic tools. While their offerings are diverse, any AI-driven software intended for medical purposes in Europe would necessitate a rigorous technical file review by a Notified Body. This would involve demonstrating the analytical and clinical validity of their AI algorithms, particularly concerning the generation of actionable insights from complex genomic and clinical data. The evidence for such tools would need to be carefully documented, covering data provenance, algorithm development, and validation against clinical endpoints. European Commission guidance on AI in healthcare Across these vendors, the common denominator is the careful assembly of a technical file that systematically addresses each MDR requirement. The Notified Body’s assessment is not a cursory glance. It is a deep dive into the documentation, often involving questions, requests for clarification, and potentially on-site audits of the quality management system (QMS) which is foundational to the entire process. ISO 13485 standard for medical device quality management systems

Judging Readiness Before Assessment

For quality and regulatory teams preparing European submissions, the ability to pre-assess readiness is invaluable. A clear indicator of whether a file is prepared for Notified Body scrutiny lies in its ability to answer fundamental questions, in order. First, is the device classification unequivocally justified and supported by the intended purpose and risk profile? Any ambiguity here will be a critical point of contention. Second, does the evidence section provide a complete and traceable narrative from requirements to validation, particularly for the AI components? This includes strong data management, algorithm transparency, and clinical performance data that directly supports the claims made for the device. And third, is there a credible and complete plan for ongoing post-market activities, demonstrating a commitment to safety and performance throughout the device’s lifecycle? If these questions can be answered with clarity, precision, and strong documentation, the technical file stands a strong chance of working through the Notified Body review successfully. The exercise is less about innovation and more about diligence. The gatekeeper is looking for documented proof of compliance, not just promising technology.

Frequently Asked Questions

What is the primary role of a Notified Body in the EU market access process for AI health tools?

The Notified Body serves as the primary reviewing layer, auditing conformity to the Medical Device Regulation (MDR) framework. They determine whether a technical file meets the stringent requirements for CE marking, which is essential for placing devices on the EU market. Their role is central to a successful submission and ultimately, market access for higher-risk medical devices, including many AI-driven Software as a Medical Device (SaMD) tools.

How has the MDR changed the regulatory landscape for AI-driven SaMD regarding Notified Body involvement?

The MDR significantly elevated the scrutiny applied by Notified Bodies, particularly for software and AI, moving many devices that previously self-certified into a category requiring Notified Body oversight. For AI-driven SaMD, the EU AI Act also introduces additional regulatory requirements that Notified Bodies must assess, creating a dual compliance framework alongside the MDR. This means a company’s readiness for Europe requires demonstrating documented compliance that can withstand an external, expert audit.

What key elements must a technical file include to demonstrate conformity for AI health tools under the MDR?

A technical file must systematically address classification, evidence, and post-market duties, following a logical progression to demonstrate compliance with the General Safety and Performance Requirements (GSPRs). Key evidence sections include risk management (extending to algorithmic bias and data privacy for AI), clinical evaluation, verification and validation (including robust testing against diverse datasets for AI), and usability engineering. It must also detail a post-market surveillance plan, including PMCF and vigilance reporting.

Why is device classification critical for AI-powered SaMD in the EU submission process?

Under MDR, AI-powered SaMD often falls into higher risk classes (IIa, IIb, or even III) due to its potential impact on patient outcomes. This classification dictates the specific conformity assessment route and the depth of Notified Body involvement. Misclassification can lead to significant delays or outright rejection of the submission.

Editorial Team

The Editorial Team at Regulated AI Health is a group of writers and editors who cover health, workplace wellness, and everyday care. They research each story and publish under a shared desk byline — experts in their fields, working as one team.