On December 16, 2025, the European Commission published its proposal to revise MDR and IVDR. At more than 200 pages in total, this is not exactly light reading. That’s why I’ve put together a summary and my assessment for you in a new technical article.
Johner Institute

Medical Device Briefings

Issue 01/26

Goals and Predictions for 2026

Hi there,

Has the new year swept you off your feet as well—so much so that it’s already hard to keep track of everything? At least for me, a veritable flood of new laws and insights has come crashing in. But let’s take this one step at a time.


Your goals (and how I can help)

First and foremost, I wish you a 2026 in which you more than fulfil your wishes. That does, of course, require knowing what those wishes actually are. When it comes to your professional goals, I want to support you  — also through the Medical Device Briefings. That’s precisely why they exist.

My goals (and what you can take away from them)

A lot of personal focus areas and predictions emerged from my AI experiments over the turn of the year. I’ll share those in today’s Executive Corner, and in much more detail in a webinar on February 3 at 9:30 a.m. Save the date.

But now, let’s get started with some truly noteworthy Regulatory Updates.

Regulatory Updates

a. MDR and IVDR 2.0

Revision of the Regulations

On December 16, 2025, the European Commission published its proposal to revise MDR and IVDR. At more than 200 pages in total, this is not exactly light reading. That’s why I’ve put together a summary and my assessment for you in a new technical article.

 

Next week I’ll report, how successful the revision of Rule 11 really is.

 

Conclusion: The re-regulation is moving in the right direction. However, for manufacturers who have already declared MDR conformity, the relief remains quite limited. For everyone else (including manufacturers of breakthrough and orphan devices) here is reason for optimism, thanks to new conformity assessment procedures and, for example, more generous requirements regarding the equivalence of comparator products.

 

Regulation (EU) 2026/2457

Regulation (EU) 2025/2457 from November 2025 has an impact on the current MDR. Article 3 amends details in Annex I, section 10.4, specifically regarding endocrine disruptors and phthalates.

 

MDCG Guidelines

The MDCG closed out the year with a noticeable productivity boost. This is reflected in several new guidelines, such as MDCG 2025-9 on breakthrough devices.

 

Less of a “breakthrough,” according to many of my colleagues, is the new MDCG 2025-10 guideline on post-market surveillance. Tables 1 to 3 are useful. However, the clarifications regarding custom-made devices contradict the planned MDR amendments mentioned above.

 

If you’re currently preparing technical documentation and wondering how you can save work in light of these changes, just get in touch.

 

b. Cybersecurity

NIS-2 Implementation Act

Back in December, I had already planned to report that the NIS-2 Implementation Act has entered into force. Whether this law applies to you—and what exactly it requires—is explained in the updated technical article on the NIS-2 Directive.

 

New MDR and IVDR

How seriously the European Commission views cybersecurity in the context of medical devices and IVDs is also evident from the MDR and IVDR revisions described above. In addition to the announced simplifications, these revisions introduce new cybersecurity requirements, for example about reporting obligations.

 

You can’t simply “patch” cybersecurity into your products. It has to be designed in from the very beginning. The best person to talk to about how to do this quickly and safely is my colleague Christian Rosenzweig. You can reach him via our contact channels.

 

c. USA/FDA

The FDA has not been idle, either:

  • It published more than 100 updates to its list of Recognized Consensus Standards.
  • It also revised the guidance document Clinical Decision Support Systems. The changes (here shown in a side-by-side comparison) are manageable, but they

A similar intent is reflected in the guideline General Wellness: Policy for Low-Risk Devices.

 

Executive Corner: Predictions 2026

a. My experiments

Objective of the experiments

With my experiments, I always aim to understand how effectively AI can help to

  • assess and improve the conformity of products and processes,
  • reduce the associated effort and thus increase productivity, and
  • optimize time to market as well as innovation.

Ultimately, this is about how we, as individuals and as organizations, can remain competitive.

 

Methodology

Following my experiments on the automated extraction of regulatory requirements from regulations, laws, standards, and guidelines, my focus “between the years” shifted to the automated review of technical documentation.

 

I wanted to find out how well I could identify errors in technical files without actually looking into them.

.

b. Results of my experiments

The Instagram version would be: I was able to successfully identify countless issues and consolidate them into an automatically generated, clear, and interactive report.

 

I’ll be demonstrating this in the webinar on February 3—feel free to save the date.

 

There are, however, a few “buts”:

  • With each run, the AI identifies different issues. In other words, the results are not fully reproducible. Just like with us humans.
  • When working with huge data sets (e.g. an extensive risk table), you need to do a bit more than “just” prompting. In these cases, it helps to segment the data according to certain criteria (which I’ll also present in the webinar) and process it in batches.
  • Without at least some technical understanding, certain tasks cannot be solved with sufficient quality. For example, structured data such as JSON schema files are extremely helpful. In some cases, orchestrating multiple AI systems is required.

c. Predictions, Conclusions, and recommendations

From these findings, the following predictions emerge:

  • By the end of this year, even auditors without special technical expertise will be able to review technical documentation with the help of AI at a sufficiently high level of quality.
  • In doing so, Notified Bodies will have an asymmetric advantage compared to manufacturers—a point also illustrated by the four-quadrant matrix below.
  • The direction of travel is clearly toward structured, machine-readable data. This is not just an assumption; it is already explicitly envisaged in the upcoming MDR/IVDR revisions.
  • Over the course of this year, we will get the automated extraction of regulatory requirements (from laws, standards, and guidelines) under sufficiently good control.
nb-confusion-matrix-small

My recommendations for you are as follows:

  • Start the transformation from documents to structured data now. I’ll also show how to do this in practice during the webinar.
  • Make sure that your colleagues—and of course you yourself—have sufficient AI skills and actively use AI in their day-to-day work. I teach these competencies in the hands-on AI workshop on January 28 and through the Johner Academy.
  • Identify tasks you want to automate because they are repetitive, time-consuming, critical, or otherwise simply annoying, and start automating them right away.

Conclusion: When it comes to AI, apart from some initial quick wins, very little comes for free. But it does help. And without using AI, you will no longer be competitive. 

Practical Tips

a. Achieving better AI results through neurostimulation

Let’s stay with AI. This time, a hands-on tip for the very advanced users:

If you want to generate optimal outputs with LLMs, so far you could

  • optimize the inputs (context engineering), or
  • adapt the model itself (fine-tuning or switching models).

A remarkable blog post now shows an additional approach: With open-source models, the model’s behavior can be influenced through “neurostimulation” (using so-called steering vectors) — without changing either the input or the model weights. This avoids the enormous resource requirements of fine-tuning.

 

If you’d like to dive deeper, you’ll find explanations from the author of the blog post in this video.

 

And one more practical tip: Before you consider neurostimulation or fine-tuning, optimize your inputs. In my projects so far, that has always been sufficient. 

neurostimulation-small

b. Understanding and meeting EUDAMED deadlines

I already wrote before Christmas that additional EUDAMED modules are now available and that transitional periods therefore apply. However, these EU requirements are anything but easy to understand. (Usability clearly played no role in drafting these texts.)

 

That’s why, in the latest podcast episode, I asked medical law expert Boris Handorn to shed some light on the situation. As this podcast is in German, just let me know if you need a summary.

 

With that, we’ve reached the end of the first Journal issue of 2026. What a start to the new year! You’ll remember:

  • MDR and IVDR are being revised.
  • On February 3 at 9:30 a.m., I’ll explain how you can also automate the review of technical documentation.

Best regards,

Christian

Founder Johner Institute

New and updated articles

On December 16, 2025, the EU Commission published its proposal to revise the MDR and IVDR—already being referred to as MDR 2.0 and IVDR 2.0. But do manufacturers really benefit from these changes? The new article provides the answers.

 

Read the new article 

 

We update our collection of articles every week. Particularly relevant this week:

  • Regulatory requirements for machine learning: Repeal of the Executive Order and additional notes on NIST requirements, new chapters “Requirements of individual countries” and “Further countries”
  • Computerized Systems Validation: Now including link on the validation of ChatGPT

Support from the Johner Institute

The Johner Institute's Product & Service Portfolio

 

The Johner Institute supports the medical device ecosystem (manufacturers, authorities, notified bodies, operators) in all regulatory challenges.

  • Strategic Consulting
    e.g., regulatory and clinical strategies for global device approval, quality systems, technical documentation, artificial intelligence
  • Business Process Outsourcing
    e.g., regulatory monitoring, post-market surveillance, clinical evaluation
  • Software (Enterprise SaaS)
    e.g., for regulatory monitoring, post-market surveillance
  • Product Testing
    e.g., clinical investigations, usability tests, pentesting, biocompatibility tests
  • Training
    e.g., seminars and e-learning
  • Taking on Regulatory Roles
    e.g., QM representative, legal manufacturer, in-country representation

Prof. Dr. Christian Johner

 

Johner Institut GmbH

Reichenaustr. 39a

78467 Konstanz

 

 

+49 (0) 7531 94500 20 
info@johner-institute.com
www.johner-institute.com

Privacy Policy   |   Subscription Settings

Johner Institut GmbH, Reichenaustr. 39a, Konstanz, Baden-Württemberg 78467, Deutschland

Manage preferences