Pharma’s great data mess: shouldn’t IDMP have solved this by now?

The International Organization for Standardization’s (ISO) Identification of Medicinal Products (IDMP) standards for medicinal product identification (1, 2) have been on the agenda for life sciences – in the EU at least – for about 13 years now. But repeated postponements and shifting timelines have led to a loss of momentum. As a result, companies have failed to capitalise on the inherent benefits of having better, richer and more reliable data in a much more re-usable format. Instead of investing early in modern data foundations, many organisations adopted a “wait-and-see” stance—doing the minimum to maintain compliance, rather than building toward a broader data strategy.

So is it too late to turn things around, especially given the industry’s appetite to harness artificial intelligence technologies to transform the way they work – and the work they do?

Peter Brandstetter of Accenture, believes it could be – unless something significant changes. “We should have started 10 or 15 years ago,” he said, of the industry as a whole, referencing the largely missed opportunity to modernise the way product data flows across manufacturing, supply chain, regulatory, quality and safety.

Caught up in the details of compliance, rather than the bigger picture, companies have seen projects exceed budgets, systems failing to align and internal teams still at odds over sources of product “truth”. Implement’s Frits Stulp noted that in some cases entire careers have been derailed by hasty decisions, immature requirements and runaway implementations—those that might that solved yesterday’s problems but have created new ones in the process.

Without good data, AI ambitions will falter

Now, one of the emerging issues is that ambitions for AI cannot be readily realised. As Remco Munnik of Arcana Life Sciences Consulting noted: “Without structure – without governance that provides meaning – AI struggles to make sense of information.”

Brandstetter agreed, noting that the current AI hype cycle is causing some companies to jump straight into experimentation without ensuring they have the foundations to support trustworthy output. This, he warned, “will lead to wrong results.” All of which could undermine trust in AI.

The more effort companies put into improving their data and what can be done with them, the greater gains they can expect from their use of AI. There are no real short cuts.

And yet AI pilots are becoming increasingly commonplace, applied to improve safety signal detection, submission generation, labelling harmonisation and more. And yes, there have already been some promising results. But proper progress (e.g., beyond a single use case, or function) depends on what lies underneath.

Companies that have put off or skimped on IDMP, continuing to see it primarily as just another regulatory compliance burden; or those whose product data remains unstructured and locked in documents, remain fundamentally ill-equipped for the AI-enabled future they can now picture.

Should Regulatory Affairs be leading more forcefully?

Regulatory Affairs departments could be doing more to lead the way, the panel suggested. As Stulp noted, this is a function that holds some of the most valuable, regulator-validated product information in a life sciences company. If structured more optimally, that data could serve as a strategic engine. AI could then do more than generate templates or speed up submissions; it could help to answer portfolio-level questions, reveal trends, support patent strategy and reshape the way that organisations anticipate changes in global markets.

In so doing AI won’t replace regulatory specialists, but rather elevate them, Munnik said. He referred to one prototype scenario where structured product data allowed automated propagation of approved company core safety information (CCSI) changes downstream, through English and local labelling, patient leaflets and multiple translations. Rather than making people redundant, this led to greater efficiency, consistency and the elimination of expensive manual translation cycles.
How do IDMP and AI impact the industry’s future?

Ultimately, IDMP is – or should be – a critical enabler of automation, interoperability and intelligence. The panel noted that the European Medicines Agency’s Product Management Service (PMS) is the “linchpin” of the shift toward structured regulatory data.

Stulp pointed to its increasing maturity and its use in shortage management, electronic application forms and future replacement of XEVMPD (the current Extended EudraVigilance Medicinal Product Dictionary). EMA, Munnik said, has “done its homework”; in other words, the burden now sits with Marketing Authorisation Holders (MAHs) to enrich, validate and align their data.

Another perspective is that IDMP creates a single language for product data – not just for EMA submissions, but also across internal functions and global markets. For AI, this consistency is essential. It is transformative for regulators too, while for patients it is the key to faster access to better-quality information.

Where companies embrace IDMP as a foundational data strategy, they will increase their opportunities to innovate. But what of those organisations that have fallen behind?

Course correction: understanding where companies have left the right path

The panel pinpointed some of the things companies have done wrong, and which they now need to redress to recover lost ground:

• Fragmented leadership
  Successful organisations have cross-functional leadership: not regulatory alone, or IT alone, but rather enterprise-level alignment around data as an asset.
• Projects that have run away from their purpose
  Programs often start well but eventually veer off, losing sight of their original goals until someone is brave enough to stop the clock, Crone noted.
• Companies chasing tools rather than outcomes
  Front-runners view tools as experiments to pilot, test, adopt, or discard quickly; they don’t invest millions before proving value, Munnik said.
• Teams clinging to “waterfall planning” in an agile world
  Incremental wins matter; so does transparency. Regulatory data journeys can’t be executed as monolithic, multi-year programs with no visible progress.
• Minimum compliance mindsets, which then backfire
  Doing “just enough” in time for each respective IDMP deadline has left many organisations with an incomplete, inconsistent, or contradictory data estate. Now, attempts to introduce AI are exposing the cracks.
• Vendors are too often viewed as “black boxes”, rather than partners
  As EMA’s interfaces go live (3) companies should be working closely with their suppliers, Stulp said, to maximise readiness, transparency and alignment on roadmap and capability.

The good news is that MAHs can now harness everything that EMA has done to ease their respective transitions. In parallel, companies should consult their preferred vendors to ensure they will be able to exchange data with EMA PMS and support the required transparency; develop a long-term data vision that goes beyond Regulatory; and embrace small, value-driven steps that demonstrate visible progress. If companies begin this work now, they could still catch up, Brandstetter suggested.

There is much to play for, if companies do the work now. Stulp pointed to the potential to harness emerging “trusted regulatory spaces”—shared cloud environments where regulators and industry are able to work collaboratively on data, review processes and documents. This could help accelerate approvals, reduce back-and-forth cycles, and improve the quality of information patients receive.

References and notes

The panel discussion, which took place in late 2025, was the latest episode of the Life Science AI Exchange Podcast, sparked by ArisGlobal. AI-Powered Life Sciences Software Platform | ArisGlobal LifeSphere

1. ISO. ISO 11615:2017 Health Informatics—Identification of Medicinal Products—Data Elements and Structures for the Unique Identification and Exchange of Regulated Medicinal Product Information (ISO, October 2017) https://www.iso.org/standard/70150.html
2. FDA. Identification of Medicinal Products (IDMP). FDA.gov, https://www.fda.gov/industry/fda-data-standards-advisory-board/identification-medicinal-products-idmp (accessed Dec. 15, 2025)
3. EMA’s PMS application programming interface (API) allows registered industry and network users to view and edit medicinal product data directly through their database systems