Key takeaways from CDISC EU Interchange and PHUSE-CSS

As clinical data science evolves rapidly, the CDISC EU Interchange and PHUSE-CSS conferences offer a glimpse into the future of regulatory submissions, standardization, and the rise of open-source tools and AI in drug development. In May, I had the privilege of attending both the events, in Geneva and in Utrecht. I’d like to share here some highlights from both conferences.

Data submission in Europe: EMA delays

As anticipated in my previous blog, we were waiting for further announcements from the EMA regarding the outcome of their pilot raw data submission project, for which an interim report was published last year.

Those, like me, who were expecting a final announcement were likely disappointed. The requirement for data submission to the EMA in support of drug approval has been postponed to 2028. The EMA, which was well represented at PHUSE-CSS, needs to further evaluate factors such as tools and technological impacts more broadly. At PHUSE-CSS, they showed particular interest in topics such as Dataset-JSON, BIMO, the use of tools such as R-Shiny and the {teal} framework, as well more advanced topics still under development such as the “Analysis Concept.” The pilot continues, and the EMA is seeking more volunteers. It was guaranteed that submitting data will not negatively affect or delay your drug approval!

It was clear, while discussing with EMA representatives, that a number of stakeholders within the agency still need to be convinced of the benefits of receiving datasets in addition to PDF documents and reports. Some appeared concerned about the additional time and effort required to assess submitted datasets. As we all know, updating regulations, as well as releasing new standards, require a great deal of “diplomacy” and the consensus among multiple stakeholders.

Open source: {teal} and R-Shiny adoption

The “open source” revolution continues to gain its momentum in our industry. At PHUSE-CSS, I attended the “{teal} Success Stories” workshop, where various sponsors, including J&J, Sanofi, Novartis, Boehringer, and Roche, shared their experiences.

I was fascinated by the solutions those sponsors have already implemented using {teal}, and how straightforward it seems to develop R-Shiny applications using the framework provided by this R package, which was originally developed by Roche.

For a deeper insight into the capabilities of this package, I recommend reading the paper presented by Roche at PHUSE US 2024.

Dataset-JSON pilot update

Another interesting workshop I attended at PHUSE-CSS was on Dataset-JSON, where we reviewed and contributed to a consolidated set of comments and feedback in response to the “FDA Requests for Public Comment on CDISC Dataset-JSON Standard,” which closes next week on June 9, 2025.

While the benefits of such a standard were widely acknowledged, particularly in accelerating drug approval and improving overall interoperability, the discussions also highlighted potential risks and implementation challenges. These included concerns about numeric precision when importing Dataset-JSON to and from SAS, as well as handling special characters.

We therefore emphasized the need for the FDA to provide additional guidance to support future adoption; there was also interest in possible future extensions of Dataset-JSON, such as the inclusion of more metadata and the potential to embed define.xml.

BIMO

BIMO was the focus of another PHUSE-CSS workshop. Among all the various topics discussed, such as the presentation of the BIMO PHUSE template reviewer guide, it was particularly interesting to learn that PHUSE will soon be developing a dedicated FAQ to support sponsors and CROs on the gray area of the BIMO requirement such as the definition of “major” studies, currently the object of BIMO requirements.

CDISC 360i reboot: Toward an end-to-end digital pipeline

The CDISC 360 initiative is back, and stronger than before, with a major shift toward a fully digitalized and standard-driven clinical development lifecycle. The goal is to break down silos through the application of standards such as the USDM and Biomedical concept. The mission is ambitious, but unlike when CDISC 360 was first launched, we now have more mature standards and technology to support it.

Use of AI to support clinical standards

AI remains a hot topic, and as in 2024, a full session was dedicated to it at the CDISC EU Interchange. The common theme across most presentations was the use of generative AI to support the implementation of data standards, such as AI acting as a subject matter expert (SME) for study teams. Although many of the showcased solutions from Argenx, SGS, and AstraZeneca are still in beta, they clearly demonstrate how proper model training can enhance search and navigation within complex data standards libraries, or help manage complex, multidimensional data (e.g., omics, wearable biosensors). Other AI use cases were also featured in several posters at PHUSE-CSS; for example, the application of AI to generate synthetic data or automate local lab ranges.

Other topics

For topics such as Digital Data Flow and USDM, I’ll refer you to the LinkedIn newsletter “View From The Coffee Shop,” curated by my friend Dave Iberson-Hurst. In it, he regularly shares insightful thoughts and updates on the ongoing digitalization efforts within our industry. He also summarized some key takeaways from both the CDISC EU Interchange and PHUSE-CSS.

I also had the opportunity to see good use cases of Analysis Results Standards (ARS) at CDISC EU Interchange, showing this relatively new standard have been well perceived by sponsors as well as vendors.

On the regulatory side, aside from the news from the EMA, I found the presentation from Sanofi and GSK particularly interesting. It covered a cross-industry initiative aimed at harmonizing vaccine regulatory submission to FDA-CBER, by sharing experience with this unique division, which often has its own set of sometime unexpected requirements (see also my previous blog on submission experience with FDA-CBER).

For other topics, see also here official CDISC posts for other conference sessions content:

• Digital Data Flow
• Real World Evidence
• CDISC Open Rules

Ongoing innovation

Overall, both conferences continue to showcase ongoing innovation in our Industry. It’s clear that change is happening at a pace I have never seen before in my 30-year career, and that’s good for patients, as well as an exciting time for those of us working in biometrics.

Interested in learning more?

Download Angelo’s new ebook, The Good Data Submission Doctor on Data Submission and Data Integration to the FDA, a collection of Angelo’s most critical insights on clinical data standards submission to the FDA, including key updates from the new FDA Study Data Technical Conformance Guide: