A Guided Experiment on Clinical Trial Data Sharing Quality and Data Reuse

AI Supported Data Retrieval and Data Management in the Health Domain

AI-Driven Therapeutic Modelling in Biomedical Informatics

Analysis and Modelling of Infectious Diseases in the Age of AI: Lessons from COVID-19

Benchmarking AI Scientists

Between Data, Practice, and Evidence: (inter)nationally EHR Research

Deep Dive into eHealth with the Project MINT Meets Medicine

Designing Next-Generation Respiratory Virus Trials: Estimands, Core Outcomes, and Adaptive Pandemic-Ready Frameworks

Developing and Executing Privacy-Preserving Federated Analyses with FLAME

Forum Digital Medicine and Artificial Intelligence

From Research Question to Access: Using Nationwide Claims Data Today – and Electronic Health Record Data Tomorrow

From Theory to Practice in the EU HTA JCA: Early Insights and Lessons Learned

Introduction to Individual Participant Data (IPD) Meta-Analysis

Large Language Model Informed Real-World Cohort Data

Mapping Metadata Schemas to the HealthDCAT-AP Standard: Lessons Learned and Hands on Community Examples

On the Ethical and Technical Aspects for Deployment and Evaluation of AI-Driven Clinical Decision Support Systems (CDSS)

Ordinal Outcomes: Friend or Foe?

Population-Adjusted Indirect Comparisons in the Framework of Health Technology Assessment

Regulatory Topics for Statisticians: Implementation of Risk-Based Quality Management from Biostatistical Perspective

Reliable Subgroup Identification and Analysis – Results of GMDS Biostatistics Competition 2026

Sensor Analytics for Rehabilitation

Teaching Statistical Literacy in Times of Political Changes

Uncertainty in Clinical Predictions

Use of Bayesian Methods for Clinical Trial Design and Analysis

What Makes a Trial Complex? Evidence, Experience, and AI

2x 90min

Organizers/Moderators: Ulrich Mansmann

Clinical trial data sharing is increasingly mandated. Meaningful data reuse often fails due to inadequate data preparation, missing documentation, or misalignment between sharing modes and reuse needs. This workshop uses a guided experiment to explore clinical trial data sharing and reuse and their technical implementation into a TRE (Trusted research environment). Besides practicing specific tools, the participants will anticipate challenges, trade-offs, and risks across the data sharing pipeline. Participants will develop a deeper understanding of how data quality and structure determine downstream reuse feasibility, FAIRness, and scientific value.

Format: Interactive, discussion-driven session combined with elements of a hackathon. Practical work complemented by theoretical training.

90min

Organizers/Moderators: Martin Golebiewski, Caroline Bönisch, Harald Kusch, Matthias Löbe

The workshop will introduce and discuss different approaches that develop workflows and tools which use Artifical Intelligence (AI), Machine Learning (ML) and Large Language Models (LLM) for supporting data retrieval, structuring, storing as well as sharing of health-related data and corresponding metadata. The aim is to showcase possiblities to integrate AI, ML and LLM-based methods in health-data management infrastructures and workflows. Attendees will be provided with insights into how such methods could be applied on what kind of data with which kind of questions. Impulse lectures will highlight the creation of an AI-generated digital twin infrastructure for healthcare in Europe and how AI methods can support the enrichment of health metadata or the usage of biosignals for clinical decision support. Also data quality aspects as well as difficulties and pitfalls of these modern technologies in the context of the health domain will be addressed. An interactive panel discussion with all speakers concludes the workshop, allowing the audience to discuss the topics with the panelists.

Format: Short talks and subsequent interactive panel discussion including the audience.

2x 90min

Organizers/Moderators: Benjamin Löhnhardt, Ulrich Sax, Tim Beißbarth

The workshop focuses on the application of artificial intelligence (AI) for predicting and developing new therapeutic approaches in a translational context. With the increasing availability of complex biomedical data, AI-supported methods are opening up new avenues for drug prediction, drug design and the modelling of individual therapeutic responses. The topic will be examined from the perspective of research and industry, as well as from the interdisciplinary perspective of biomedical informatics.

Format: Short talks with moderated discussion.

90min

Organizers/Moderators: Martin Wolkewitz

This workshop brings together leading experts to address the analysis and modelling of infectious diseases in the age of artificial intelligence, using the COVID-19 pandemic as a motivating example. The objective is to integrate perspectives from epidemiology, epidemic dynamic modelling, survival analysis, machine learning, and related methodological disciplines. Rather than focusing on technical implementation or software details, the workshop emphasizes conceptual understanding, methodological challenges, interpretation, and data infrastructure requirements. Through concrete examples from the COVID-19 pandemic, the speakers will reflect on key analytical challenges, lessons learned, and future directions for infectious disease research in increasingly data-rich and AI-supported environments.

Format: Short talks with moderated discussion.

90min

Organizers/Moderators: Georg Fuellen, Klaus Jung

The ”Benchmarking AI scientists” workshop shall give an overview on the use of AI Agents for Science, with a focus on benchmarking their performance, and a focus on the application area of biodata analysis / bioinformatics.

Format: Short talks with moderated discussion.

90min

Organizers/Moderators: Veronika Strotbaum, Saskia Kröner, Robin Grashof

We invite participants to a collaborative workshop on national electronic health record (EHR) research. On one hand, we aim to assess the current status of (inter)national EHR Research across medical, technical, legal, governance, communicative and public health dimensions. In this context, we will identify research potential from identified gaps and needs. On the other hand, we aim to strengthen networking within German national EHR research by connecting the CHI working group with other GMDS-focus groups, researches outside the GMDS community and practitioners.

Format: Short impulse presentation, followed by an interactive on-site selection of discussion topics using an online live voting system. Discussions in moderated small groups. In a subsequent plenary session, groups will present their results.

90min

Organizers/Moderators: Michael Marschollek, Mattea Müller

This workshop presents results from the MINT Meets Medicine Deep Dive Challenges, an interdisciplinary training initiative at the interface of STEM disciplines and medicine hosted at the Peter L. Reichertz Institute for Medical Informatics, Hannover. Within the program, six interdisciplinary groups of early-career researchers collaborated over several months on three clinically motivated eHealth data challenges. The first challenge analyzed continuous glucose monitoring data to investigate individual glucose responses in daily life. The second challenge explored early warning systems for chronically ill patients using large-scale longitudinal synthetic electronic health record data to identify risk patterns and support proactive care. The third challenge integrated multiomics data from kidney disease patients, including transcriptomics, to study molecular mechanisms and clinically relevant cell-type–specific signatures.

Format: Station-based exchange format. Participants engage directly with project teams in small-group settings.

90min

Organizers/Moderators: Inge Christoffer Olsen, Alain Amstutz, Matthias Briel, Dominique Costagliola

This workshop explores methodological, operational and regulatory challenges in designing clinical studies for respiratory tract viral infections, including pandemic settings. Participants will examine how estimand selection, core outcome set development, and adaptive protocol frameworks influence trial interpretability, comparability, and decision-making. Special attention will be given to the emerging role of AI when designing a trial, e.g., using AI in the work of identifying core outcome sets. Furthermore, we will elaborate on how to deal with operational, legal and logistical hurdles in case of health emergency situations and how to potentially overcome them, e.g., implementing a federated trial approach to start as quickly as possible in different indications. Finally, we will look at choice of development phase and how to select candidate interventions. The workshop aims to provide practical guidance for selecting meaningful outcomes, designing flexible trial infrastructures (umbrella, master, or pathogen-specific protocols), and aligning designs with regulatory expectations. Using real case studies from large European pandemic research initiatives, participants will learn how to translate methodological principles into operational study design.

Format: Case studies and interactive discussions. Participants will discuss real design scenarios. Ending with panel discussion

90min

Organizers/Moderators: Marius de Arruda Botelho, Peter Placzek, Bruce Schultz, David Hieber, Alexander Röhl, Mehrshad Jaberansary, Philipp Brassel, M Schaible Hammam Abu Attieh, Mehmed Halilovic, Matthias Meyer, Toralf Kirsten, Fabian Prasser, Oliver Kohlbacher

Data protection regulations such as GDPR1 require that the provision of data in healthcare balances strict protection of patient data with the societal benefits of medical research. At the same time, institutional data silos hinder cross-site analytics and collaborative model development. Federated learning (FL) enables distributed analyses without transferring sensitive data, but practical adoption depends on usable tooling, transparent governance, and reproducible execution of workflows. This workshop introduces FLAME, an open-source federated analytics platform developed within the national PrivateAIM2 project, and focuses on the practical lifecycle of developing, submitting, and executing analyses using a public demonstration instance and FLAME software development kit (SDK). FLAME follows a hub-and node architecture in which analyses are executed locally at data-providing institutions while coordination and aggregation are managed centrally. The SDK abstracts communication, data access, and result handling, enabling developers to focus on methodological implementation. Multiple federated analysis patterns and execution modes are supported and reproducibly deployable across sites. A wide range of analytical use cases, including imaging-based workflows, privacy-preserving genome-wide association studies, and newly developed federated methods, has already been successfully executed on FLAME. Documented example workflows demonstrate reproducible end-to-end execution across heterogeneous data modalities. The platform is currently transitioning toward productive operation across institutions, accompanied by-weekly community meetings for knowledge transfer and feature dissemination. The workshop is designed as a hands-on interactive session in which participants experience the full federated analysis workflow from local development to distributed execution and result inspection. This practical format addresses current barriers such as the complexity of tooling and uncertainty in cross-institutional collaboration. The session targets researchers, clinicians, data scientists, and infrastructure developers interested in privacy-preserving medical analytics and aims to foster interdisciplinary understanding and community adoption.

Format: Short talks followed by guided hands-on exercises using a public demo environment.

3x 90min

Organizers/Moderators: Hans-Ulrich Prokosch, Christian Hinske, Rainer Röhrig

The workshop is part of the Forum Digital Medicine and Artificial Intelligence, which aims to bring together Clinformaticians (Clinical Informaticians) from the clinical societies of the AWMF with the methodological disciplines of the GMDS in order to discuss interdisciplinary methodological, regulatory, and political challenges in digital medicine as well as for the development and application of Artificial Intelligence Tools. Based on stimulating expert presentations the workshop participants will discuss the current challenges and hurdles and interactively develop solutions for future developments and activities. One part of the workshop will be a Pro&Contra Presentation and Discussion on the opportunities, but also the challenges and hurdles for agentic AI systems.

Format: Based on stimulating expert presentations the workshop participants will discuss the current challenges and hurdles and interactively develop solutions for future developments and activities. One part of the workshop will be a Pro & Contra Presentation and discussion on the opportunities, but also the challenges and hurdles for agentic AI systems.

90min

Organizers/Moderators: Rebecca Alvarado

The Forschungsdatenzentrum Gesundheit (FDZ Gesundheit / Health Data Lab at the German Federal Institute for Drugs and Medical Devices, BfArM) provides controlled access to nationwide statutory health insurance (SHI) claims data covering approximately 74 million insured persons in Germany. These longitudinal populationlevel data enable large-scale analyses in epidemiology, biostatistics, health services research, and pharmacoepidemiology. The FDZ data infrastructure is currently evolving in two important ways. First, a new Data Model 4 will soon be introduced, expanding the scope and structure of available claims information and thereby enhancing analytical possibilities. Second, the FDZ is expanding its portfolio to include electronic health record (EHR) data. Starting at the end of 2026, the first EHR data type—the electronic medication list from the German electronic health record—will be made available for scientific use. In the future, researchers will be able to combine claims data and EHR-based information, opening new perspectives for real-world evidence generation and medication-related research. This interactive workshop will guide participants through the pathway from research idea to data access at the FDZ, with a focus on both current analytical opportunities and the evolving data landscape. The objectives are to:

• Introduce the FDZ data infrastructure, including the current claims data, the upcoming Data Model 4, and the integration of EHR medication data, outlining their respective strengths and limitations.
• Demonstrate which research questions can be addressed today, how analytical potential expands with Data Model 4, and how study concepts may evolve with the availability of EHR data.
• Provide practical guidance on submitting a high-quality data use application, including best practices, common pitfalls, and strategies to prepare projects in anticipation of future data extensions.

Format: Using publicly available descriptions of existing FDZ data projects, participants will work through guided examples. Ends with moderated discussion on strategic project planning and drafting applications.

90min

Organizers/Moderators: Kirsten Herrmann, Saskia Trescher, Peter Schlattmann

This workshop examines how the EU’s new Joint Clinical Assessment (JCA) process is transitioning from theory into practice under the recent EU HTA regulation, highlighting early insights and lessons learned with a special focus on the German experience. It will provide a comprehensive overview of the challenges, solutions, and real-world experiences encountered during the first wave of JCA implementations, particularly looking at how joint assessments interface with national procedures like Germany’s AMNOG and how methodological differences between EU-level and domestic HTA requirements are being managed.

Format: Short talks and a moderated panel discussion.

90min

Organizers/Moderators: Tim Mori, Alexandra David, Christian Röver, Tim Mathes

Individual participant data (IPD) meta-analysis has become an essential tool in modern evidence synthesis, particularly in clinical research, regulatory decision-making, and health technology assessment. Compared with conventional aggregate data meta-analysis, IPD approaches enable more flexible modelling of intervention effects, improved handling of rare outcomes, and robust investigation of subgroup effects. The primary objective of this workshop is to provide participants with a clear conceptual and methodological understanding of IPD meta-analysis, its advantages, limitations, and appropriate use cases, thereby supporting rigorous and transparent evidence generation.

Format: Interactive, methodologically focused session combining conceptual explanations with applied guidance.

90min

Organizers/Moderators: Michelle Pfaffenlehner, Nadine Binder

Using large language models (LLMs) in the context of real-world cohort data modeling or analysis offers new opportunities and a wide range of applications. This workshop provides insights into the integration of LLMs with real-world data (RWD) through three illustrative use cases. First, we present the use of LLMs to generate synthetic data, introducing a structured framework that defines key principles and steps for designing such datasets. Second, we discuss the interaction of LLMs with synthetic data, focusing on data enrichment and assessing how LLMs interpret and respond to real-world data structures. Third, we examine the application of LLMs for the analysis of RWD, highlighting potential applications, limitations and methodological considerations.

Format: Short talks followed by a Fishbowl panel discussion.

2x 90min

Organizers/Moderators: Judith Wodke, Deepak Unni , Vasundra Touré,  Hanneke Leegwater, Carina Nina Vorisek, Sylvia Thun, Dagmar Waltemath, Sabine Österle

According to the principles of Findable, Accessible, Interoperable, and Reusable (FAIR) data, research data should be made accessible accompanied by rich metadata that allows the assessment of their utility and reusability for a specific purpose beyond the original intent. With the advancement of the European Health Data Space (EHDS), European countries are required to harmonise their health data collections towards defined interoperability standards. The specified standard for metadata within the EHDS is the Health Data Catalog Vocabulary Application Profile (HealthDCAT-AP). While data formats and metadata schema differ between European countries and even between national initiatives, the challenge to harmonise metadata across countries is shared. The Swiss Personalized Health Network (SPHN), the Dutch Health-RI Initiative, the German National Research Data Infrastructure for Health (NFDI4Health) and Network University Medicine (NUM) / Medical Informatics Initiative (MII), and the FAIR Data Point specification used in several European funded projects are conducting a joint initiative to map their national metadata schema to the HealthDCAT-AP standard. In this workshop, we aim to present different approaches and share lessons learned. We encourage the communities of ISCB and GMDS to bring their own data examples and to work in a hands-on-hackathon on EHDS-compatible metadata enrichment for those data or our provided examples. We will engage in a lively discussion, sharing cross-border experiences and gathering feedback for further improvements of the provided mappings. We expect to align across biomedical disciplines the efforts towards interoperable metadata within the EU.

Format: tba

90min

Organizers/Moderators: Zully Ritter, Stefan Rühlicke, Sebastian Fudickar, A. Koop

Although AI techniques such as Machine Learning, NLP, and Computer Vision are widely used in healthcare for diagnosis, disease management, and risk prediction, far less attention has been paid to ethical metrics in their technical development. This workshop presents methods for measuring and evaluating data ethics, focusing on defining, quantifying, and distinguishing ethical biases and noise from biases and noise in data and algorithms.

Format: Short talks, interactive component (Slido word cloud or live poll), and a panel discussion.

90min

Organizers/Moderators: Katherine Lee

Ordinal outcomes are becoming increasingly popular in the medical and public health literature, particularly in platform trials where it is desirable to have a common primary outcome that is relevant across heterogeneous groups of participants and captures a wide range of clinical outcomes. This workshop will involve a series of presentations followed by a facilitated discussion on the use and analysis of ordinal outcomes, which we hope will spark a discussion of whether ordinal outcomes are our friend or foe.

Format: Short talks followed by a facilitated discussion.

90min

Organizers/Moderators: Ralf Bender, Friedhelm Leverkus, Anika Großhennig

Standard methods for indirect comparisons and network meta-analysis are based upon aggregated data and require the validity of the main assumptions of sufficient similarity, homogeneity, and consistency. The assumption of similarity is also called “constancy of relative effects”, because it requires that all effect modifiers are balanced between the two trial populations. Frequently, typical questions in the framework of health technology assessment (HTA) lead to indirect comparisons where the assumption of sufficient similarity is questionable or obviously invalid. For the case that individual patient data are available in one subset of trials, methods have been developed to relax the assumption of similarity by performing population-adjustments for effect modifiers, which are not balanced in the included trials. The validity of the corresponding results depends on the new assumption that all relevant effect modifiers are known and correctly included in the statistical model. This assumption is called “conditional constancy of relative effects”. Methods for population-adjusted indirect comparisons have been generalized to allow even comparisons in disconnected networks and the inclusion of single-arm trials. These methods, however, make the much stronger assumption of “conditional constancy of absolute effects”, which means that the absolute treatment effect is constant at any level of the included variables. Thus, beside all effect modifiers, in addition all relevant prognostic variables have to be known and correctly included in the model. Not only is this strong assumption unlikely to be valid in practice, but it is furthermore extremely challenging to provide sufficient evidence that this assumption is fulfilled in a given indirect comparison. In this workshop, methods for population-adjusted indirect comparisons are introduced and the advantages and disadvantages of the most important approaches are discussed. A special focus is given on multilevel network meta-regression, because this method has several advantages compared to matching-adjusted indirect comparisons and simulated treatment comparisons. The importance, suitability and limitations of population- adjusted indirect comparisons in the HTA framework are discussed, along with recent methodological developments and first experiences of the new joint clinical assessments at EU level from industry, academic and HTA perspectives.

Format: Short talks and a moderated panel discussion.

90min

Organizers/Moderators: Antonia Zapf, Anja Sander, Anika Großhennig

To ensure data quality, patient safety, and operational efficiency, the implementation of Risk-Based Quality Management (RBQM) has become essential in clinical trials. As emphasized in the ICH E6(R3) guideline, RBQM is a critical framework for safeguarding the integrity of clinical trial data. By systematically identifying, assessing, and prioritizing risks based on their potential impact, RBQM enables targeting monitoring strategies that enhance regulatory compliance, accelerates study execution, and strengthen stakeholders’ confidence in trial outcomes. In this workshop, we will explore practical approaches to implementing RBQM in clinical research and highlight the role that study biostatisticians can or should play in this context. In the workshop, classical strategies, such as central monitoring of missing data, detection of adverse event under- or over-reporting, and trends in protocol deviations, will be discussed. In addition, the workshop will place special emphasis on advanced data analytics including comprehensive database-wide analyses for early issue and misbehaviour detection. Participants will gain insights into leveraging statistical and data-driven tools to proactively identify risks, improve quality of relevant data, and optimize monitoring efforts across multi-site trials.

Format: Short talks and a moderated panel discussion.

90min

Organizers/Moderators: Anika Großhennig, Max Westphal

The GMDS Biostatistics Competition 2026, organized by the Biometrics section of the GMDS in collaboration with Merck Healthcare KGaA, addresses the critical challenge of reliable subgroup identification in clinical trial data. With 30 simulated datasets that closely mirror real-world phase II/III randomized trials, the competition aims to evaluate methododological approaches for detecting treatment-effect heterogeneity, identifying predictive and prognostic biomarkers, assigning patients to subgroups, estimating subgroup proportions, and quantifying treatment effects within identified groups. Participants are challenged to developing transparent, reproducible, and robust analytical workflows that balance statistical power with interpretability, moving beyond black-box machine learning toward clinically meaningful and interpretable approaches. The data-generating mechanism and true underlying subgroup assignment remain undisclosed, ensuring a rigorous and unbiased evaluation. Performance is assessed using task-specific metrics, including agreement rates, variable selection accuracy, subgroup proportion estimation, and treatment effect estimation via RMSE, compared against the true data-generating model. In this workshop, we present the overarching results of the competition, highlight the most effective and innovative approaches, and showcase exemplary solutions from selected teams that combine methodological rigor, reliability and practical relevance. The session will be streamed live, enabling global participation and fostering open dialogue on the future of subgroup analysis in evidence-based medicine.

Format: Short talks and discussion.

90min

Organizers/Moderators: Behrus Hinrichs-Puladi, Dagmar Krefting

Rehabilitation increasingly benefits from multimodal approaches combining imaging/video, biosignals, and other sensors, enabling objective assessment of movement and function, interactive therapeutic feedback, and safe automation (e.g., robotics and assistive systems). Objectives are: (1) to present practical and research-driven examples of image/video-based rehabilitation technologies, including pose estimation and patient localization, interactive exercise guidance, and quantitative imaging biomarkers; (2) to explore multimodal fusion scenarios that combine image/video information with biosignals and other sensors for improved assessment, feedback, and personalization; and (3) to discuss key requirements for translation and real-world adoption, including validation strategies and clinically meaningful endpoints, robustness across settings, privacy considerations, and integration into clinical workflows.

Format: Interactive workshop with short impulse talks followed by a moderated panel discussion and structured audience interaction (polls + Q&A).

90min

Organizers/Moderators: Ursula Berger, Carolin Herrmann

In times, where artificial intelligence (AI) and data-driven research reshape our views, we see political dynamics that threaten the availability and credibility of statistical data sources. The value of statistical integrity and accuracy is getting increasingly important and with it the need for teaching statistics literacy to a broad audience. This session addresses current challenges and innovations in teaching statistical literacy in times of rapid AI development and increasing political challenges, with an emphasis on life sciences. The focus will be on civic data literacy, enabling life science students and wider audiences to critically engage with AI tools, statistical information, and the role of data in democracy.

Format: Short talks and a moderated panel discussion.

90min

Organizers/Moderators: Ewout Steyerberg

Clinical prediction models for a health condition are commonly evaluated regarding performance or utility for a population, although decisions are made for individuals. In this workshop, we first discuss developments in quantification of finite-sample uncertainty for individual patients in terms of ‘effective sample size’: the number of similar patients that a prediction is effectively based on. This number of ‘patients-like-you’ can be estimated for predictions based on a wide range of models, including generalized linear models and various machine learning methods. Novel developments include the estimation of effective sample size for treatment benefit at the individual level. We discuss implications for model development and transparent risk communication. Uncertainty in risk estimates for individuals also relates to model uncertainty (variability in modeling choices made by different modelers) and applicability uncertainty (variability in measurement procedures and between populations). We will present case studies and simulations to illustrate the huge role of model uncertainty and applicability uncertainty.

Format: Case study presentations with interactive discussions encouraged by having polls on provocative statements and ample room for clarifying questions and debate.

2x 90min

Organizers/Moderators: Annette Kopp-Schneider

The FDA recently published a draft guidance titled “Use of Bayesian Methodology in Clinical Trials of Drug and Biological Products.” This marks a major step toward broader acceptance of Bayesian methods in regulatory submissions, as the guidance is shifting towards a full acceptance of the Bayesian paradigm in the context of drug development and provides a roadmap for implementing Bayesian approaches in clinical trials. The draft guidance has attracted considerable attention and sparked discussions. Almost simultaneously, EMA issued a concept paper on the use of Bayesian methods in clinical development, indicating that attention to the topic is also shared by European regulators and is planned to develop into a future reflection paper. With the traditional paradigm potentially shifting, a discussion on the use of Bayesian methods for clinical trial design and analysis is timely.

Format: First part: panel discussion. Second part: talks and a discussant.

90min

Organizers/Moderators: Franz König, Fabian Eibensteiner, Frank Pétavy, Tim Friede, Sonja Drescher

This workshop explores clinical trial complexity as a dynamic and evolving concept rather than a fixed classification. It examines multiple dimensions of complexity, including ethical, regulatory, statistical, methodological, and operational aspects. These are informed by findings from (a) a structured literature review, (b) a multi-stakeholder Delphi process, and (c) a review of relevant regulatory guidance. A central objective is to critically discuss whether criteria exist that make one trial inherently more complex than another, or whether “complexity” often acts as a surrogate for limited experience, emerging methodology, or infrequent use of specific designs or technologies. Participants will also learn how AI methods, particularly Large Language Models (LLMs), can support the automated identification and characterization of potentially complex trials registered in EU via the Clinical Trials Information System (CTIS, https://euclinicaltrials.eu/), while understanding the limitations and governance challenges of such approaches.

Format: The workshop is designed as an interactive, multi-format session combining short presentations with discussion of literature and regulatory guidance findings, demonstrations of AI-supported trial screening, case-study discussions, and real-time audience interaction via online surveys. A moderated panel discussion will conclude the session.