Skip to content

AI Compass / News

DIGICAP: medical device dossiers are meant to become machine-readable

RA & medtech

An EU project led by notified bodies wants dossiers exchanged as data instead of PDFs, and RA teams can apply for its panel until 30 October.

Source: DIGICAP – Digitalizing the Conformity Assessment Process (Team-NB),

Note: I provide engineering services in regulatory affairs for medical devices.

Technical drawing: on the left a stack of loose files with a PDF on top, an arrow leads to a structured dossier made of field rows with check marks, from there three lines branch out to boxes labelled NB, EUDAMED and REGTECH.

What happened: On 20 October 2026, Team-NB, the European association of medical device notified bodies, featured the EU-funded project DIGICAP on its website; according to the launch announcement, the project started on 1 September 2026. Conformity assessment, the check that a device meets the requirements of the MDR and IVDR, is to run on structured, machine-readable dossiers instead of PDFs and email chains. Data captured once is meant to be reused by notified body tools, the EU database EUDAMED and RegTech software (software for regulatory tasks). According to the project page, the Horizon Europe project runs until August 2029 with 19 partners in 13 countries, 11 of them notified bodies. Team-NB is the coordinator.

What this means: DIGICAP is to develop a shared digital framework with open data schemas, meaning shared field definitions for dossiers, plus a roadmap, and test the framework in at least four pilots with manufacturers and notified bodies. The targets, such as 60% of applications complete on first submission and 25% fewer requests for missing information, are the consortium’s goals, not results. There is nothing manufacturers can use yet; results are to be published via the website and GitHub.

What you can do:

  1. Apply for the stakeholder panel by 30 October: DIGICAP is looking for 30 to 35 external experts, including from manufacturers, to shape the framework through several feedback rounds (a Delphi process). Registering expresses interest; the page does not say how much time the panel takes.
  2. Collect follow-up requests: Note what your notified body last asked you to supply. That makes your application concrete and shows the gaps in your file.
  3. Keep your file modular: Maintain core data such as intended purpose, classification and variants in one place and reference it instead of copying it into every document. That makes a later move to a data format easier.

Context: In my view, structured dossiers are also a better basis for AI-assisted review tools: checking defined fields rather than PDF pages makes gaps easier to find. The consortium also lists AI tools under human control (human in the loop) as one of five pillars in its launch announcement: they are to speed up documentation preparation and run automated completeness checks before submission; DIGICAP is exploring recommendations and safeguards for them.

What does this mean for your team? Let’s talk it through. Request an intro call