Event
PHUSE EU Connect 2025
17.11.2025
About the event
From November 16 to 19, PHUSE EU Connect 2025 brings together leading experts from life sciences, clinical research, and data analytics in Hamburg. The international conference provides a platform for exchanging ideas on regulatory topics, emerging technologies, and data-driven innovation in drug development.
The event is all about open knowledge sharing, practical insights, and engaging discussions. That is why HMS is once again taking part – contributing our own talks, sharing insights from our project work, and looking forward to new ideas and connections.
Our contribution to PHUSE
Talk
How to Identify and Operationalize GenAI Use Cases Which really help You
Christoph Bergen, Luis Wirth
Generative AI (GenAI) is everywhere right now – but particularly within the complex workflows and regulatory structures of the pharmaceutical industry, it can be difficult to determine whether it is actually the right solution for a specific challenge.
In this beginner-friendly session, we provide practical guidance to help make that decision. We start by clearly distinguishing between traditional (or “discriminative”) AI and generative AI and explore the types of problems each approach is particularly well suited to address. Using real-world examples from the pharmaceutical industry, we explain how GenAI models can be applied in practice – and which additional components are needed to successfully implement a specific use case. We also present different strategies for operationalizing GenAI, ranging from self-service platforms to distributed systems that provide GenAI tools via MCP servers.
This session is aimed at both GenAI beginners and decision-makers looking to integrate generative AI into real-world business processes in a targeted way and with clear benefits.
Download presentation nowDownload paper now
Talk
Poseidon: streamlined regulatory screening of HA websites with AI/ML extracting updates reducing manual efforts
Maria Giovanna Zill (Merck KGaA), Fabian Wahren
The “Poseidon” project aims to streamline the regulatory screening of websites operated by international Health Authorities (HAs) at Merck Healthcare. Currently, subject matter experts (SMEs) invest significant manual effort in screening relevant content and subsequently prioritizing it.
By developing an AI/machine learning system that automatically evaluates updates on HA websites, Poseidon aims to significantly reduce this manual workload and free up resources for more strategic tasks. The project not only increases efficiency and reduces the risk of human error, but also creates a centralized knowledge base that supports informed assessments of regulatory impact and improves compliance.
Scalability: The system is designed to be transferable to other areas within healthcare and life sciences – for example, regulatory topics related to clinical development, medical devices, manufacturing, or pharmacovigilance.
Time savings: The solution is expected to measurably reduce the time subject matter experts spend on regulatory screening.
Download presentation nowDownload paper now
Talk
SAMBA: Bridging the Gap – How GenAI Transforms Analysis Plans into ADaM Specs
Thiago Figueiredo, Richard Becker (Mreck KGaA), Fabian Wahren
Programming specifications, particularly ADaM specifications, combine study-specific information with established industry standards to ensure that analysis results are comparable across studies. Creating analysis packages for clinical trials requires a thorough understanding of the study protocol and Statistical Analysis Plan (SAP) in order to implement statistical analyses based on the variables and datasets described in these documents.
In practice, however, creating these specifications is time-consuming and involves numerous manual steps, making it difficult to standardize metadata across studies.
We present the Smart Assistant for Metadata and Biostatistics Analysis (SAMBA), a tool that helps statistical programmers create ADaM specifications more efficiently based on existing study documents. SAMBA uses Large Language Models (LLMs) to partially automate the creation and updating of specifications, significantly reducing the effort involved.
We provide insights into the system architecture, the development process, user feedback, and potential directions for future development.
Talk
Leveraging AI for Seamless SAS to R Migration: Challenges and Opportunities
Corinna Stöckinger (Boehringer Ingelheim), Milena Kraus, Dr. Robert Bauer
While SAS has traditionally been the de facto standard for the statistical analysis of clinical trials, the industry has increasingly turned its attention to open-source alternatives in recent years. One of the most prominent is the R programming language. At Boehringer Ingelheim, a ClinPharm task force was established to support the transition from SAS to R, backed by a team of HMS experts.
In this talk, we present our joint project to migrate a collection of SAS macros and templates used in PK/PD analyses to R. The aim was to make the migration process more efficient with the help of Generative AI (GenAI). We explore several approaches tested during the project – ranging from direct code translation to the use of existing packages from the pharmaverse – and discuss their respective advantages and disadvantages.
Finally, we examine both the opportunities and limitations of GenAI in this context, as well as the need for a structured framework to harness its potential effectively and successfully.
Download presentation nowDownload paper now
Poster
How can agile R Package validation workflows be achieved?
Dr. Sophia Stahl-Toyota
Moving from feature requests or bug reports to the next validated release of an R package within an agile workflow is a vision shared by many product owners, developers, validation specialists, and users. Add stakeholders from quality assurance, deployment pipeline teams, or management, and it becomes clear that success depends not only on the right tools, but above all on effectively coordinating people and processes.
We show how the tool landscape for software development and validation can be set up using the technical capabilities of platforms such as GitLab, Jira, Azure DevOps, and GitHub. This includes the effective use of workflows, status fields, labels, scoped labels, tags, components, and time-based structures such as iterations, sprints, and milestones.
However, efficient tracking of development and validation activities – which typically involve numerous reviews and coordination steps – ultimately depends on configuring and using these options effectively. And that is where the real challenge lies.