HOME
QUICK SEARCH
AUTHORS
KEYWORDS
HELP
BUGSPHUSE EU 2025 Papers - Index
214 papers. Source: PHUSE (Pharmaceutical Users Software Exchange) EU Connect conference proceedings.
| # | Title× | Authors× | Abstract× | Keywords× | Extras | Pages |
|---|---|---|---|---|---|---|
| 1 | %batchrunprograms: SAS Studio Background Submit on Multiple Programs | Luis Andre Soriano | As with traditional local SAS Batch Submit, SAS Studio Background Submit on a program is important in ensuring the program runs successfully from a clean state and contains no errors, warnings, or undesirable notes. Often, multiple programs need to be run sequentially, independently, and frequently (e.g. developing SDTMs/ADaMs/TLFs). There is unfortunately no built-in way in SAS Studio to Background Submit multiple programs. Current solutions involve tinkering with XCMD option (security risk), creating non-SAS tools (separate skill to learn), or using %INCLUDE on multiple programs (independence hard to ensure). In this presentation we'll show that %batchrunprograms is developed by only incorporating tools from SAS/CONNECT and Base SAS (e.g. ODS, PROC PRINTTO) |
sas studio, excel, sas base, proc step, macro definition | source presentation source |
8 |
| 2 | %CallR | Duong Tran | The integration of software applications within workflows is both desirable and often essential, as it enables the combined strengths of these tools to increase productivity and gives better quality results. When it comes to leveraging R from within SAS, the licensed SAS/IML module is the commonly recognised solution. However, there exists a cost-free alternative. This technique was first introduced by Hollands[1] in the paper "SAS to R to SAS" and has since been developed and explored by other contributors[3]. This presentation seeks to expand on this body of work by providing additional insights and demonstrating practical applications through %CallR, a SAS macro designed to facilitate seamless interaction with R. Key words: systask, x, %sysexec, BATCH, R.EXE |
macro definition, output delivery system, sas enterprise guide, macro facility, proc step, data science | source poster source |
14 |
| 3 | A Code Assistant for Clinical Trial Tables | Pietro Mascheroni, Torben Brehme |
Tables, listings, and figures (TLFs) are key components in clinical trial reporting, yet their generation remains a time- consuming task. Programmers often face tight deadlines and must ensure outputs are both accurate and traceable. In this work, we present an agentic workflow for automating table creation using large language models (LLMs). Our code assistant, meRlin, accepts requests for a clinical reporting table from the user, retrieves relevant R code templates, and accesses clinical data and trial documents such as the ADS plan and Clinical Trial Protocol (CTP). Based on these inputs, meRlin adapts the code templates to generate tailored R code. Programmers can then review, execute, and debug the code with the assistant's support. We illustrate the approach with reference table examples and discuss future enhancements | clinical trial, data science | source poster source |
6 |
| 4 | A Collaborative Fast Track to Clinical Data Flow Modernisation and Standards Governance Implementation | Paul Vervuren, Roger Gagali Angoh |
Presentation | source | 14 | |
| 5 | A Git-Driven Automated Workflow for Efficient ADaM and TFL Validation and Quality Control Documentation | Xavier Chenede, Paulin Charliquart |
Traditional quality control (QC) processes for Analysis Data Model (ADaM) and Tables, Listings, & Figures (TLF) programming rely on manually filling EXCEL templates for planning and tracking program validation. We propose an evolution towards an automated end -to-end approach leveraging R, Git -based code development, workflow automation, AI-enhanced code review, and HTML dashboards. Our framework integrates multiple complementary quality control components: programmer -developed unit tests for code validation, AI -powered reviews utilizing Large Language Models (LLM) fine-tuned by internal standards, and a comprehensive HTML tracker that consolidates all quality checks. This integrated approach significantly reduces manual interventions, minimizes human error, and accelerates turnaround time from data availability to delivery, while maintaining the rigorous quality standards essential to our work |
quality control, data cleaning, cdisc | source presentation source |
8 |
| 6 | A New Gateway to Open Source in SAS | Jim Box | PROC GATEWAY is a new procedure in SAS Viya that allows you to run code written in other programming languages directly from within SAS. We will show how GATEWAY can be used to integrate R and Python code in your SAS Programs, allowing you to take advantage of new open-source procedures to solve more complex coding tasks. We'll also show you how to take advantage of Viya's cloud capabilities to submit open-source programs in a way to take advantage of multi-threading, allowing you to really leverage large data or complex analyses |
proc step, proc iml, cas action, sas viya | source presentation source |
7 |
| 7 | A Smarter Way To Track QC: Integrated Workflows in a GxP-Validated Statistical Computing Environment | Jim Coates | Presentation | excel | source | 6 |
| 8 | A Splash in Time Insights into a Data Pooling Time Machine | Lutz Hoffmann, Dana Vinzelberg |
A SAS macro system (iPool) was designed to support integration of clinical study data into data pools in a highly efficient manner. As a pool (metadata and data) changes over time like a "living object" all necessary modifications can be maintained as pool definitions are stored separately from the data. The data itself is stored only once in a pool. In addition, medical coding (e.g. MedDRA and WHODD) is supported for multiple versions. A new pool is initially defined with pool metadata and controlled terminology. Afterwards studies can be added, and pool transformations can be defined with the according timeframe information. Therefore, it is possible to extract the pooled study data as it was at a certain point in time in the pool's lifetime. With this approach it is possible to "walk back in time" to extract a pool from the past | character variable, data integrity | source presentation source |
7 |
| 9 | ADaM Conformance Rules: Bridging the Gap Between Standards and Practice | Rafik Ghazaryan | The ADaM Conformance Rules provide a standardized framework for assessing the compliance of ADaM datasets with CDISC standards, helping to ensure quality, consistency, and transparency. Yet, passing conformance checks does not necessarily confirm full compliance, as true adherence depends on fundamental principles of traceability, clarity, metadata, a nd analysis readiness. This paper examines the purpose, structure, and limitations of ADaM Conformance Rules, drawing on regulatory context and practical experience. It highlights common challenges - such as misalignment with study -specific requirements or specific analysis needs - and identifies gaps between rule-based validation and broader CDISC standards. Finally, it emphasizes best practices for incorporating conformance checks into programming workflows and underscores the need to look beyond rule compliance to ensure datasets are reliable for all their intended uses | data cleaning, adam dataset, data science, data validation rules, cdisc | source presentation source |
6 |
| 10 | Advancing Automated Non-Compartmental Analysis with R Shiny | Jana Spinner, Gerardo Rodriguez |
Non-Compartmental Analysis (NCA) is a cornerstone of pharmacokinetic studies, providing critical insights into drug absorption, distribution, metabolism, and elimination. Despite its importance, current workflows are heavily reliant on proprietary tools and manual processes, limiting efficiency, reproducibility and accessibility, particularly in complex and time-sensitive study designs. To address these challenges, we developed aNCA, an open-source R package with a user-friendly Shiny application, that fully automates NCA calculations and streamlines reporting. Built on the regulatory-aligned PKNCA package, aNCA enables half-life calculations, and exploratory visualizations. A key feature is its ability to generate regulatory-compliant CDISC datasets (PP and ADPP), as well as clinical submission -ready tables, listings, and graphs, dramatically simplifying one of the most time-intensive aspects of NCA workflows. Figure 1 | compartmental analysis, data visualization, cdisc | source presentation source |
9 |
| 11 | AI Chatbot in an SCE | Shivoy Pandita, Marc Jantke |
As AI adoption evolves, it's integration into Statistical Computing Environments (SCEs) presents both opportunities and regulatory challenges in the pharmaceutical industry. entimICE FastTrack introduces a secure, configurable AI assistant, enabling compliant, role-based interactions with a locally deployed large language model (LLM). No clinical or corporate data leaves the environment, ensuring full data privacy and non-training compliance. Users can leverage Retrieval Augmented Generation (RAG) using documents, programs, or datasets, facilitating productivity enhancements such as synthetic data creation or program mock-ups. This AI assistant focuses on secure, domain-specific information retrieval rather than general-purpose chat or coding, which are better handled in IDEs like Posit Workbench or by the more popular Chat GPT and Copilot | data cleaning, mock shell, data science, safety population, proc means | source presentation source |
8 |
| 12 | AI-Native Issue-Risk Intelligence: Connecting Issues to Known Risks and Identifying Emerging Risks | Lawrence Dennison-Hall, Alok Singh |
We introduce an AI-native process designed to enhance Risk-Based Quality Management (RBQM) in clinical trials by autonomously mapping issues to known risks and identifying emerging risks. Using natural language processing (NLP) and machine learning, AI algorithms analyse textual data to identify and categorise issues. By providing insights into the materialisation of issues, study teams can optimise control strategies, address recurring issues and improve overall risk oversight. We aim to demonstrate how this AI solution can transform RBQM processes and provide actionable insights for improved quality and safety in clinical trials. This approach leverages AI capabilities to perform tasks that were previously unattainable which can enable proactive risk management. By integrating this solution, clinical trials could ensure data quality and integrity as well as enhanced patient safety | clinical trial, sample, data cleaning | source presentation source |
14 |
| 13 | An Evaluation of Data Formats for Clinical Reporting in a Multilingual Environment | Craig Gower-Page | The use of non-SAS programming languages for analyzing clinical trials data is becoming increasingly common. This trend raises an important question: How can we format and store clinical data in a way that ensures interoperability without compromising on features or performance? This paper explores the strengths and weaknesses of several potential data formats, ranging from human-readable options like CSV and JSON to more advanced binary formats such as DuckDB and Parquet. Each format is assessed based on key criteria, including the availability of language-specific libraries, performance characteristics, and support for various data types. Particular attention is given to the Apache Parquet format, a highly efficient binary data format that stands out for its robust performance, compatibility with popular analytical programming languages, and seamless integration with many widely used tools and databases |
clinical data, clinical trial | source presentation source |
12 |
| 14 | An Evolving Role of Statistical Programmers: Leveraging New Capabilities in Open-Source Environments | Rob Livermore, Maciej Wojtecki-Glod |
In this paper, we discuss the emerging opportunities for professional statistical programmers to leverage open-source technologies, enhancing contributions to clinical projects as well as the operational frameworks in which we work. Through a case study on the implementation of an R Shiny app designed for clinical project and resource management, we explore how we can drive innovation and improve processes in adjacent areas. The experience gained from the implementation not only cultivates skill transfer back into the clinical project domain but also fosters a culture of continuous learning and development. Additionally, we emphasize the importance of strategically aligning these efforts to maximize organizational benefits while maintaining collaborative relationships with other technical functions within the organization | clinical programming, statistical programmer, excel | source presentation source |
9 |
| 15 | An Integrated J&J Platform To Enable Sustainable Clinical Trial Design | Archana Nayak, Jason LaRoche |
Pharmaceutical companies have partnered with the Sustainable Healthcare Coalition (SHC) to develop standardized methods for measuring greenhouse gas emissions in clinical trials. However, the faster adoption of these methods into everyday trial design remains a challenge. To address this, we have developed our internal J&J application module, which includes a front-end application leveraging industry-standard methodologies to embed sustainability into trial design. The application module enables secure, large-scale data pooling from multiple systems, consolidating inputs such as country site distribution, drug product data, and other relevant information, serving as proxies for baseline protocol data elements. This unified approach allows users to design baseline emissions in a single platform, accessible for ongoing refinement, scenario testing, and reuse | clinical trial, logistic distribution | source presentation source |
7 |
| 16 | Are Study Deliverables Under Your RADAR? | Deepak Pandarivasagan, Piotr Karasiewicz |
The Risk Assessment of Deliverables in Analysis and Reporting (RADAR) tool enhances early risk assessment and mitigation for analysis and reporting deliverables, aligning with submission excellence and next -generation submission strategies. By providing a standardized method for tracking progress, it facilitates rapid decision- making and minimizes manual effort and typographical errors. This is achieved using source data inputs with minimal team intervention, employing a validation tracker (also referred to as QC/Efforts tracker) to capture deliverables, effort hours, and overall study milestone dates, thus streamlining business processes. Specifically designed to support the Biometrics team, the tool simplifies decision-making processes and addresses resource gaps and delivery risks within the Biometrics scope. Developed using SAS and Power BI, it offers a user-friendly front-end for easier navigation and improved project management | data cleaning, clinical research | source poster source |
5 |
| 17 | Aspirations of language agnosticism - experiences expanding the programming stack with python | Kaj Arai Hejlesen, Michael Svingel |
The use of R within Novo Nordisk has expanded rapidly in recent years, with approximately 50% of studies now using R for some programming and analyses tasks, and an increasing number using R exclusively. With the expansion came an aspiration of language agnosticism: the ability to one day choose and use any programming language based on the task and/or competencies of the programmer(s) involved. We present our experiences expanding our R-based GxP programming environment to include python for TFL programming and demonstrate that it is not just about installing the language binaries alone. Special consideration must be given to the processes & responsibilities associated with the environment as a whole such as package development, cross-language package sharing, systems validation and user support. We conclude that mastery over these elements is fundamental to language agnosticism and indeed enables a more robust assessment of whether language agnosticism is a worthy aspiration at all | data cleaning, data science, clinical research, statistical software, clinical programming, missing value | source presentation source |
9 |
| 18 | Automated Mock TFL Using an RAG LLM Setup | Nicolaj Thostrup | Creating Mock TFLs in a study, can be a critical but tedious task. It can be crucial, to enable delegation of efforts but it is also in itself a cumbersome task. In shorter studies it might not yield a great benefit/work ratio. Thus, it is relevant to look into innovative approaches, which can automate some of the work of mock generation, to quickly draft outputs. I will present a proof of concept system, using a LLM and RAG setup, which is designed to improve this process. The system uses a database of historical outputs as input as well as an output title, and then generates or modifies an output according to the requests from the user. The result is an automated approach to mock generation leveraging AI | proof of concept, clinical trial | source presentation source |
7 |
| 19 | Automated QC Framework for Reliable ADNCA Dataset Generation with {admiral} | Valentin Legras, Lucy Aspridis |
The generation of robust analysis datasets for Non-Compartmental Analysis (NCA), such as ADNCA, is a critical and time-intensive task for | adam dataset, sample, compartmental analysis, quality control, sdtm domain, cdisc | source presentation source |
4 |
| 20 | Automating RTF-to-PDF Conversion for BIMO Submissions: A Python-Based End-to-End Solution | Yi Yan, Po-Chun Lin |
Presentation | clinical programming, regulatory submission | source | 15 |
| 21 | Bayesian Inference for Medical Device Reliability: Improving Safety and Effectiveness Assessments | Shrutimita Pokhariyal | Medical devices, from implantable sensors to injectable treatments, require rigorous safety and efficacy evaluation, often with limited failure data and censored observations. Traditional frequentist methods struggle in these cases due to small sample sizes and the need for prior knowledge incorporation. Bayesian inference provides a robust alternative by integrating prior clinical insights with observed data, improving uncertainty quantification and predictive accuracy. This paper explores Bayesian methods for medical device reliability assessment, demonstrating their advantages in modeling rare adverse events, handling incomplete trial data, and adapting to real -world evidence. A case study on injectables used in minimally invasive treatments highlights Bayesian modeling's ability to refine risk assessment and optimize regulatory decision-making | sample, sample size, bayesian statistics, credible interval, posterior probability, adverse events, confidence interval, likelihood function, censoring, missing value, clinical trial | source presentation source |
7 |
| 22 | Becoming an Independent Consultant And Stepping into the Fourth Dimension | Jean-Marc Ferran | privacy, data science | source presentation source |
6 | |
| 23 | Being a Perfect Programmer: Code Smarter, Review Sharper, and Embrace the Future | Akhil Vijayan, Limna Salim |
What does it take to be the "perfect" programmer in today's dynamic world of clinical data programming? It's more than technical proficiency, it's about cultivating smart habits, leveraging automation, and embracing the power of AI. This paper presents practical tips, coding strategies, and real-world examples that help clinical programmers work more efficiently, code more robustly, and review more effectively | data cleaning, independent qc, macro parameter, excel | source presentation source |
11 |
| 24 | Beyond Authority: Navigating Challenges, Earning Trust and Building High-Performance Teams in FSP and Outsourced Delivery Models | Emmanuelle Hoarau, Tim Southwick |
In today's dynamic clinical research landscape, effective leadership is no longer defined by position alone-it demands emotional intelligence, adaptability, and inclusivity. This paper explores the nuanced challenges faced by leaders assuming new responsibilities - navigating the complexities of managing diverse personalities, fostering team cohesion, and driving compliance in both functional service provider (FSP) and outsourcing delivery models. With a focus on real-world lessons learned, it delves into strategies for gaining buy-in for routine processes, earning trust without authority, and managing both neurodivergent and neurotypical team members. This work includes experiences of leading a functionally distinct team, managing difficult personalities in low-visibility settings, and maintaining stability in rapidly evolving environments | service provider, clinical research | source presentation source |
4 |
| 25 | Beyond the Code: Evolving from Clinical Programmer to Strategic Partner in Clinical Trials | Anastasiia Khoba | In today's fast-paced clinical trial landscape, the role of a clinical programmer is evolving far beyond code writing. As data play crucial role for regulatory decisions and trial outcomes, programmers are increasingly expected to contribute not only technical solutions but also strategic insights. This paper will explore how programmers can grow into influential roles within cross-functional teams by developing leadership, communication, and business-awareness skills as well as how to effectively collaborate with non- programmers' roles to become a strategic partner rather than a task executor | clinical research, clinical programming, data cleaning, data science | source presentation source |
4 |
| 26 | Beyond the Code: Leading Innovation from the Heart of Statistical Programming | Bhavesh Patkar | clinical programming, adaptive design | source presentation source |
5 | |
| 27 | Beyond Tokenisation: Considerations for Linking Healthcare Data Sets for Scientific Research | Jennifer Dusendang, Yuval Koren |
Linking data sources is highly valuable when performing scientific studies using real-world data. using real-world data for scientific studies. Although creating privacy preserving record linkage (PPRL) tokens is part of the linkage process, additional methods and considerations are necessary to produce a reliable and usable linked data set. Linkage rates provide an upper bound of linked patient coverage and usability for analyses. However, these are typically basic calculations of the frequency of PPRL token matches between two data sets, without regard to data element availability, study period overlap, or issues with low-specificity tokens. In particular, not identifying an appropriate timeframe in which linked data is applicable for a study can lead to unexpected decreases in sample size | sample size, sample, proc freq | source presentation source |
6 |
| 28 | Blinded & Coordinated: Aligning People and Processes To Prevent Unblinding | Monika Bajek, Dubravka Georgijevic |
Double-blind clinical trials are the gold standard for minimizing bias and ensuring scientific precision, yet they present significant operational and ethical challenges, particularly when interim analyses or safety reviews necessitate unblinded data access. Maintaining the integrity of the blind for core study teams, while enabling timely and compliant unblinded assessments by independent groups such as Data Monitoring Committees (DMCs), requires meticulous planning and execution. This abstract outline our experience in navigating these complexities across multiple global trials. We describe governance frameworks, role segregation strategies, and data flow controls that have proven effective in mitigating these risks. Key documentation practices are discussed, including the use of unblinded charters, protocols, and audit trails to ensure transparency and regulatory compliance | interim analysis, clinical trial, randomization, data science | source presentation source |
6 |
| 29 | Blockchain for Secure and Transparent Clinical Trials | Raj Kumar | Blockchain introduced a paradigm of verifiable trust - immutability, cryptographic integrity, and append-only recordkeeping - that directly addresses enduring challenges in clinical research data management. While decentralized ledgers are not operationally suited for regulated statistical computing, their fundamental principles can be leveraged within enterprise data infrastructures to ensure integrity, reproducibility, and audit transparency. This paper presents a blockchain-inspired Statistical Computing Environment (SCE) that integrates cryptographic hashing, Merkle-tree validation, and immutable lineage capture to achieve end-to-end traceability of datasets, programs, and outputs. Every analytic execution generates verifiable hashes and provenance links, enabling automatic reconstruction of lineage and guaranteeing tamper-evident audit trails | data cleaning, audit trail, raw data | source presentation source |
8 |
| 30 | Blockr: Advancing Data Pipelines with DAG-Based Workflows and LLM Orchestration | Karma Tarap, Nicolas Bennet |
clinical research, data cleaning | source presentation source |
7 | |
| 31 | Bridging Languages: Building Integrated Analytical Tools with SAS, R, and Python for Clinical Data | Margarethe Povey | Addressing the evolving analytical challenges in clinical research, we showcase an innovative methodology for building interoperable programs leveraging SAS-language, R (including "Shiny"), and Python. We explore techniques for creating single, cohesive applications that effectively combine the data manipulation capabilities of the SAS- language, the statistical power and visualization potential of R (including interactive "Shiny" dashboards), and the flexibility of Python for advanced tasks. The presentation will cover key considerations for seamless language integration and robust validation to ensure the integrity of results. Practical examples will showcase the benefits of this multi-language approach in enhancing analytical flexibility, promoting collaboration, and driving innovation in statistical programming workflows | machine learning, clinical research, clinical trial, regulatory submission, data integrity, data cleaning, artificial intelligence, audit trail, pharmaceutical industry, adverse events, data science, proc step, proc mixed | source presentation source |
7 |
| 32 | Bridging the Gap: Translating Intercurrent Events into Analysis-Ready ADaM Datasets | Meghna Diedrich | With the implementation of the ICH E9 (R1) Estimand Framework, handling intercurrent events (ICEs) has become a critical part of clinical trials analysis. Events like treatment discontinuation, use of rescue medication or death can alter the interpretation of treatment effects if not properly accounted for. Statistical programmers are at the core for ensuring estimand-aligned, robust analysis ready-ADaM datasets. This paper presents a practical guide to identifying, deriving and mapping ICEs using a custom ADICE dataset. Key variables such as ESTzzSTR, ADECOD, ICESEQzz are highlighted, along with five strategy types - Treatment Policy, Hypothetical, While-on-Treatment, Composite, and Principal stratum Through mock data, a few possible mock TFLs and a dataset flowchart will be presented to demonstrate how programmers can enable estimand-alignment, transparency, and regulatory compliance | intercurrent event, adam dataset, data science, treatment effect, while-on-treatment, hypothetical strategy, imputation, clinical trial, cdisc | source poster source |
9 |
| 33 | Building Reusable Precision Logic in SAS Using PROC FCMP | Mariam Babayan, Syuzanna Simonyan |
This paper introduces a focused and practical approach to utilizing PROC FCMP in SAS for building reusable, user- defined functions, demonstrating its power and flexibility through clear, real-world programming examples. As analytical programming evolves, there's an increasing need for modular, transparent, and testable logic, especially for tasks that appear simple but grow complex in regulatory, clinical, or data reporting environments. This paper demonstrates the full power and flexibility of PROC FCMP through practical examples with broad relevance to real- world programming challenges. Beyond the technical strategy, the paper emphasizes how reusable logic and structured programming practices can increase adaptability, reduce rework, and strengthen |
proc fcmp, proc step, standard deviation, raw data, proc means, analysis set | source presentation source |
6 |
| 34 | Busy ? Productive: Breaking the Myth of Multitasking | Goele Aerts | In today's rapidly changing world and workplace, speed and availability have become the new currency. In clinical trials, project managers are expected to juggle dozens of tasks, timelines, emails, and instant messages - often all at once. Multitasking has become the norm. Worse, a badge of honor. But here's the catch, multitasking is a myth. It doesn't boost productivity. It kills it. And in high-stakes environments like ours, that's dangerous. In this paper, I will share how hustle culture crept into my daily routines as a project manager - and how it almost derailed my focus. I'll testify how single-tasking helped me regain control, deliver better results, and reclaim my attention. Expect real perspectives, honest reflections, and one big insight: less really is more | clinical trial, clinical research | source presentation source |
8 |
| 35 | But Wait, There's More: When RECIST 1.1 Is Not Enough | Anny Yamileth Olivares Mundo | pd, adam dataset, cdisc | source poster source |
8 | |
| 36 | CDISCDataset.com: An AI-Powered Platform for Generating Synthetic CDISC-Compliant SDTM and ADaM Datasets, CRFs, Protocols, TFL Shells, and Specs | Ara Baghumyan | Access to realistic CDISC-compliant datasets is limited due to privacy and compliance restrictions, creating challenges for training, validation, and development of SDTM, ADaM, and TFL workflows. CDISCdataset.com addresses this gap with an AI-powered, web-based platform that generates synthetic datasets aligned with CDISC standards. Users can customize study parameters while ensuring outputs that mirror the structure of real clinical data without exposing sensitive information. By enabling secure, reproducible, and customizable dataset creation, the platform supports learning, validation, and collaboration across biometrics teams. This innovation provides the PHUSE community with a practical solution for advancing data science in clinical research while reducing dependency on real trial data | data cleaning, clinical research, clinical data, cdisc | source presentation source |
2 |
| 37 | Check Yourself: Edit Checks and Queries Made Easy with Shiny | Kirsty R. Lauderdale, Maya Gans |
Presentation | outlier, data validation rules, data cleaning | source | 10 |
| 38 | Clinical Trials in the Quantum Era | Michael Cooper, Edyta Winciorek |
Quantum computing holds transformative potential for the design, execution, and analysis of clinical trials. By lever- agingquantum algorithms, researchers can address complex challenges such as optimizing patient recruitment, modeling multifactorial biological systems, and enhancing the analysis of large, heterogeneous datasets.Quantum computing'sability to perform high-dimensional data analysis and complex optimization at unprecedented speeds offers the promise of more efficient trial designs, faster identification of therapeutic effects, and more personalized treatment strategies. Early-stage applications includequantum-enhanced machine learningfor biomarker discovery, simulation of molecular interactions for drug development, and optimization of adaptive trial protocols. While significant technical and infrastructural challenges remain, including hardware limitations and the need for quantum-specific algorithm development, ongoing interdisciplinary research is rapidly advancing | clinical trial, clinical research | source poster source |
11 |
| 39 | Close Encounters with Real-World Data | Jeffrey M. Abolafia, Sarah Ferko, Ingeborg Holt |
There is a rapid increase in the use of RWD to support marketing applications. However, submitting RWD may pose numerous challenges for regulatory reviewers in their analysis of the data. In our six previous papers, we outline these challenges and present several high-level solutions. This current paper builds on our previous work by taking a deep dive into CDISC SDTM example data and highlighting updates needed to the SDTM to better accommodate RWD. Specifically, we examine the gaps in the Subject Visits (SV) domain for its use in regulatory review. Considering how other models such as HL7 FHIR, OMOP, and PCORnet handle certain elements, we propose updated domains and data elements to represent patient encounters that are a key component of RWD. These ideas are intended to help sponsors submit RWD in the currently required CDISC format and to continue the conversation with SDOs such as CDISC around the updates needed for more efficient regulatory review | subject visits, unique key, clinical trial, data standard, insurance claims, clinical data, foreign key, key, cdisc | source presentation source |
21 |
| 40 | Closing the Knowledge Gap with a Knowledge Management Strategy Empowered by Dashboards and AI | Jannick Reymann Ravnborg, Vicky Poulsen |
The unprecedent pace of technological advancements, together with constantly evolving standards and regulatory requirements, presents a critical people leadership and management challenge in the pharmaceutical industry. Traditionally knowledge has been accumulated and transferred via on-the-job training - knowledge gap. It is not uncommon to see knowledge centring in a handful of subject matter experts due to the lack of a robust knowledge management strategy that ensures the efficient collection, storage and dissemination of knowledge. In this presentation, we will share our thoughts and illustrate our initiative to close the gap with the establishment of a knowledge base that employs interactive dashboard and AI to facilitate knowledge sharing. Our goal is to foster a collaborative environment where knowledge sharing is key and enable employees to build competencies across the organisation | pharmaceutical industry, clinical trial | source presentation source |
5 |
| 41 | Code Compare A Machine Learning Based Framework for Code Plagiarism Analysis and Detection | Saroj Kumar Jena | This paper introduces an innovative framework for detecting programming code plagiarism by integrating cutting - edge tools and techniques. The system leverages R Shiny as an intuitive user interface and Python -Flask as the backend API to ensure seamless interaction and robust functionality. At the core of the model, the TF -IDF (Term Frequency-Inverse Document Frequency) method is utilized to quantify the significance of terms within the code, while cosine similarity measures the angle-based distance between vectorized code samples for efficient comparison. Together, these algorithms ensure precise and computationally efficient identification of code similarities, empowering developers to identify potential instances of plagiarism or similarity. With its ability to compare coding practices across different environments, such as SAS -to-SAS or R -to-R scripts, the framework showcases its adaptability for real - world programming scenarios | clinical trial, proc freq, data cleaning | source presentation source |
6 |
| 42 | Code, Collaborate, Conquer: The Future-Proof Programmer in an AI World | James Shipley-Burgess | The role of the statistical programmer is at a critical crossroads. Traditional programming tasks are increasingly becoming automated, as AI reshapes the life sciences landscape. This raises an important question: What does the future hold for programmers? This paper showcases how a new model for drug development has programmers at its very core. Clinical trials costs have reached a point of unsustainability, with the current model based on blockbuster drugs. There is a need to reduce costs, and whilst everyone is turning to AI for the answer, part of the solution lies with programmers. Programmers must now serve as strategic collaborators working cross-functionally across departments, guiding AI output, optimising and leveraging data workflows across the business, and unlocking the hidden potential in trial data. Far from being replaced, programmers are becoming more essential, reimagined as technical linchpins in a human-in-the-loop model | statistical programmer, clinical research, clinical trial, data transformation, clinical programming | source presentation source |
4 |
| 43 | Collaborative Industry Open-Source Solutions for Clinical Trial Quality Management | Björn Koneswarakantha, Frederik Collin |
clinical trial, time series analysis, quality assurance, type i error, outlier, clinical data, adverse events, protocol deviations, quality control, bootstrap, likelihood function, data cleaning, raw data | source presentation source |
9 | |
| 44 | Contributions to CAMIS PHUSE Project in 2025: Tobit Regression, Reference-based Multiple Imputation and Tipping Point Analysis | Yannick Vandendijck | imputation, multiple imputation, regression analysis, sample, data science, proc lifereg, statistical software, missing value, clinical trial, pharmaceutical industry | source presentation source |
12 | |
| 45 | Cracking the Curve: Unlocking Clinical Insights with Ridgeline Plots | Mariam Babayan, Rafik Ghazaryan |
Ridgeline plots represent an emerging data visualization technique that enables the comparison of multiple distributions across parameters, timepoints, or treatment arms. Within clinical trials, these plots provide a powerful means of uncovering trends in laboratory data and other repeated measures, revealing shifts and differences that may be hidden in traditional tables. This paper demonstrates how ridgeline plots can be applied to ADaM laboratory (ADLB) data using R and the ggridges package, and discusses practical challenges such as handling skewed values, ensuring interpretability, and maintaining regulatory compliance. The implementation is demonstrated on a simulated ADLB dataset with laboratory categories including Chemistry, Hematology, and Urinalysis panels | treatment arm, clinical trial, laboratory data, box plot, data visualization, regulatory submission, treatment effect, adverse events | source presentation source |
6 |
| 46 | Debug Your "Work Language" | Jess Anderson | source presentation source |
5 | ||
| 47 | Decode Your Future: A Professional Development Growth Map for Programmers and Statisticians by Role Type | Diana Stuart | Job titles often suggest seniority or experience but rarely reflect the diversity of responsibilities and development needs within a role. From technical experts to project leads, programmers and statisticians span a wide spectrum of strengths and priorities. This poster and accompanying paper introduce a practical guide that breaks down common role types within these job families, mapping them to key skills, development focus areas, and growth pathways. Whether you are an early-career professional, a technical specialist, or transitioning into leadership, this guide helps identify where you are and what to focus on next. By offering a clear structure for professional development aligned to role type, it empowers individuals to take ownership of their learning journey and supports managers in tailoring development conversations. A must-have resource for anyone seeking clarity, direction, or inspiration in navigating their career in statistical and programming roles | data science, clinical trial, data cleaning, data standard | source poster source |
8 |
| 48 | Detecting Logical Errors in Clinical Programming by Applying ChatGPT | Illia Skliar | The paper presents an approach for the semantic SAS-code review in the innovative field of clinical programming with the help of NLP ChatGPT model. The primary goal is to detect logical inconsistencies or even mistakes between the implemented SAS code and rules, which are outlined in the unstructured but critical clinical trials documents such as Statistical Analysis Plan (SAP) or Protocol. Unlike standard tools such as PROC COMPARE, which only focus on comparing output results, the integration of a GPT model through an API directly into the SAS environment is proposed. This approach specifically targets the validation of the logic implemented in the code fragments, for example, the assessment of Best Overall Response |
data cleaning, data science, proc step, clinical programming, proc compare, audit trail, clinical trial, macro variable, quality control, independent qc, adverse events | source presentation source |
9 |
| 49 | Developing a Compliant and Scalable Statistical Computing Environment Based on PHUSE SCE Guidelines | Casey Higgins, Mark Sellors |
The Phuse SCE White Paper has become a critical industry guide for building compliant, scalable, and auditable Statistical Computing Environments. In alignment with these standards, Atorus developed Ageirein, a deployment framework that operationalizes the paper's recommendations into a modular, cloud-enabled solution. Ageirein embeds GxP compliance, validation, reproducibility, traceability, and role-based access control directly into its architecture, enabling rapid deployment without sacrificing regulatory rigor. By incorporating compliance controls from the start, Ageirein reduces the overall validation burden, simplifies ongoing maintenance, and streamlines audit readiness. Designed to support both exploratory and production statistical workflows, it provides a standardized, scalable foundation that helps life sciences organizations modernize data science environments while maintaining alignment with Phuse best practices | data cleaning, audit trail, life sciences, access control, analytics platform, clinical trial | source | 5 |
| 50 | Digital SAP Model Using Biomedical Concepts | Tom Ratford | The SAP is a pivotal biometrics document which is the source of truth for the detailed definitions used in clinical trial analytics and reporting. However, traditionally the SAP is created as a Word document meaning the semantic information is not easily accessible and cannot be used to drive downstream automation - resulting in inefficient manual processes existing throughout the programming lifecycle. We need access to the semantic information contained in a SAP in a machine-readable format, this enables automation of statistical analysis and reporting processes, resulting in faster and higher quality analysis projects | adam dataset, adverse events, data science, cdisc | source presentation source |
5 |
| 51 | Disruption in Clinical Trials: How Conventional Roles Are Set to Change | Amit Javkhedkar, Kala Shivali |
clinical trial, database lock, raw data, data governance, data cleaning | source presentation source |
7 | |
| 52 | Diversity, Inclusion and Equality Its Just Good Science! | Roxanne McGuire, Chris Ward |
Diversity, inclusion, and equality in clinical trials are critical to advancing equitable healthcare and ensuring scientific integrity. Historically, many clinical studies have underrepresented populations based on race, ethnicity, gender, age, and socioeconomic background, leading to gaps in knowledge and disparities in healthcare outcomes. The National Institutes of Health (NIH) Revitalisation Act 1993 was the first legal requirement pushing for the inclusion of women in federally funded medical research and a turning point of clinical trials. In this paper I will discuss that incorporating diverse populations in clinical research allows for a more accurate understanding of how treatments affect various groups, enhances the generalisability of results, and promotes equitable access to innovative therapies. As regulatory bodies increasingly emphasise the need for representative data, prioritising diversity and equality in clinical trials is both a scientific and moral imperative | clinical trial, clinical research, data mining, data science | source presentation source |
19 |
| 53 | Efficient Conversion of SDTM/ADaM Specifications from Excel to Define-XML v2.1 Using Pinnacle 21 Community | Hiromi Okada, Seiko Yamazaki |
Define.xml is a critical component of electronic data submissions, providing detailed metadata descriptions for clinical trial datasets, particularly those conforming to the SDTM and ADaM standards. While many professionals develop their dataset specifications in EXCEL during the early stages of implementation, generating the Define.xml file can be challenging due to its technical complexity and strict formatting requirements. This is especially true for users with limited experience in XML structure and regulatory standards. To streamline this process, we utilized Pinnacle 21 Community, a widely adopted tool in clinical data standardization, to convert EXCEL-based specifications into a compliant Define.xml v2.1 file. We also explored the use of Pinnacle 21 Enterprise, the commercial version of the tool |
excel, adam dataset, data cleaning, bds, adverse events, format statement, sdtm domain, length statement, data validation rules, data science, cdisc | source presentation source |
15 |
| 54 | Efficient SAS programming using CHOOSEC/CHOOSEN and WHICHC/WHICHN over traditional IF-THEN-ELSE clause | Nimit Agrawal | While writing a SAS program for a conditional statement we always tend to use IF-THEN-ELSE clause as it is a straightforward and common approach but do we ever ponder over it whether it is always an efficient approach to use IF-THEN-ELSE clause in each scenario. CHOOSEC/CHOOSEN and WHICHC/WHICHN are set of two efficient SAS functions that come into picture here and can be used effectively in many scenarios over traditional IF-THEN-ELSE clause e.g. assignment of numeric value to different Overall Response values (CR, PR, etc.) for the derivation of BOR. assignment of CRF text to a numeric toxicity grading scale This approach will help reduce the long time we spend for writing repetitive IF-THEN-ELSE clauses and as a result, will make our program more efficient, readable, precise and robust. In this paper we will discuss in detail the syntax, usage and application of CHOOSEC/CHOOSEN and WHICHC/WHICHN set of functions in different programming scenarios | whichc, if-then-else, whichn | source presentation source |
7 |
| 55 | Elevating Data Advocacy: Transforming Dashboards with React.JS and AI-Powered Innovation | Richard Becker, Markus Niederstrasser |
Dashboards play a critical role in delivering real-time data insights and fostering data advocacy within organizations. At Merck, we leverage dashboards to empower decision-making and drive data-driven strategies. In this presentation, we will showcase practical examples of how dashboards are utilized across various functions at Merck, highlighting key lessons learned in their development and adoption. A focus will be placed on the use of React.JS components leveraging Merck's branded Liquid Design System to enhance user experience, streamline functionality, and ensure a consistent look and feel across applications. Additionally, we will investigate the integration of Large Language Models (LLMs) into dashboards to assist users with output generation, thereby improving usability and optimizing standard user interactions | data cleaning, bootstrap | source presentation source |
8 |
| 56 | Embracing the Fear of Change in an Ever-Changing Industry | Wilri Mackay | The pharmaceutical industry is undergoing continuous transformations, driven by scientific innovation, regulatory changes, and evolving patient expectations. While essential for progress, change often triggers fear: fear of failure, the unknown, and disrupting the status quo. This presentation reframes fear not as a barrier, but as a potential catalyst for innovation and resilience. Drawing on personal experience and examples, we'll explore the roots of fear in professional settings and the organizational patterns that reinforce it. We'll examine strategies for transforming uncertainty into a driving force for progress. Encouraging adaptive mindsets, experimentation, and building trust amid ambiguity. Together with attendees, we'll identify fear-based barriers within teams and discuss how to embed pro- change practices across functions. Whether you're facing AI integration, restructuring, or evolving clinical trials, this session offers a roadmap to turn discomfort into opportunity | clinical trial, pharmaceutical industry | source presentation source |
4 |
| 57 | Enabling Standardised Data Collection Through Governance Processes and Automation | Clara Giménez Arteaga, Christine Mathiesen |
Presentation | edc | source | 17 |
| 58 | End-to-End R: Where To Start? | Luke Thompson | data cleaning, adam dataset, pharmaceutical industry, data science, regulatory submission, statistical programmer, clinical trial, cdisc | source presentation source |
6 | |
| 59 | Enhancing Clinical Data Analysis: The PROC GREPLAY Advantage | Vaisint Singh Muthudurai, Kalyane Subramanian |
In clinical trials, summarizing extensive data for each subject often results in dense, hard -to-review tables. Graphical representations offer clearer insights by displaying each subject's response separately, but individual graphs can lead to lengthy reports. PROC GREPLAY in SAS addresses this by customizing graph layouts, allowing multiple graphs on a single page. This customization enhances the accessibility of findings, making them more effective for scientific publications, conference presentations, and stakeholder communications. This method conserves space and improves the ability to compare responses between subjects or track changes within the same subject over time. By leveraging PROC GREPLAY, we can enhance data visualization and facilitate more efficient and effective analysis of clinical trial results. In this paper, we will discuss the application of PROC GREPLAY in SAS in detail, providing practical insights, challenges and best practices |
proc greplay, proc step, clinical trial | source presentation source |
9 |
| 60 | Enhancing Clinical Trial Management with Automated Data Integration and Real-Time Analytics | Katrijn Op, Sopra Steria |
Argenx is a biotech company managing multiple compounds across different phases of development, working with various contract research organizations (CROs). One key challenge has been achieving unified, timely oversight of operational performance and trial progress across portfolios. To address this, we developed a data mapping framework in SAS Life Science Analytics Framework (LSAF). Vendor data ingestion is set up in collaboration with each vendor (CRO) and the operational data is standardized internally. Another argenx team generates automated SDTM datasets. With both the clinical and operational data, we can generate important study metrics which are then used to create visualizations designed to meet the needs of our internal stakeholders. This fully automated integration enables daily-refreshed, cross-study and cross-vendor analytics, driving better visibility into study progress, vendor performance, and protocol compliance | clinical trial, excel, life sciences, cdisc | source presentation source |
5 |
| 61 | Enhancing Healthcare Intelligence: Advancing Clinical Information Extraction through Knowledge Graph Construction | Maria Puri, George Panagiotakis |
normalization, json | source poster source |
11 | |
| 62 | Enhancing Stats Programming Efficiency in Pharma: Harnessing Large Language Models (LLMs) such as ChatGPT for Accelerated Development and Validation | Andy Ratcliffe | This paper demonstrates the significant advantages of integrating cutting-edge natural language processing models such as ChatGPT into statistical programming workflows within the pharmaceutical industry. By strategically leveraging a model's capabilities, this approach offers substantial gains in development speed and resource optimisation, leading to faster project timelines and reduced operational costs. Beyond mere automation, this paper showcases practical applications and innovative techniques that empower statistical programmers with intelligent assistance, streamlining complex tasks, improving code quality, and ultimately increasing their productivity and impact. This exploration highlights how embracing AI tools like ChatGPT can augment human expertise, transforming challenges into opportunities for enhanced efficiency and innovation in pharmaceutical development | statistical programmer, clinical programming, data cleaning, proc step | source presentation source |
3 |
| 63 | Enhancing the Validation and Simulation of Real-World Multi-Site Data with R | Nikolay Trankov, Elena Chaparova, Alexandrina Lambova, Stavros Oikonomou |
Real-world data (RWD) encompasses a broad range of sources, including databases, registries, and electronic health records (EHR). Data from EHR is frequently sourced across multiple hospitals situated in different countries. Due to differences between practices and regions, data often comes in a variety of formats. This data is collected based on a common data specification (CDS). We have developed a standardised format of the CDS, containing definitions of the variables, logic checks and data constraint rules. We have additionally created an R package, which uses the CDS to validate and simulate data. The package reads the CDS and produces outputs of patients and their data fields which fail the logic checks and rules. Furthermore, data can be simulated for the purpose of writing programming code in parallel with the data curation and validation process, which leads to shorter timelines in Real- world evidence studies | data cleaning, sample, date variable, key, data validation rules | source presentation source |
11 |
| 64 | Essential Conflict Management Tools for Emerging Leaders: From Tension to Transformation | Satish Shetty, Nalin Gupta |
Effective leadership starts with self-awareness, especially when it comes to managing conflict. As new leaders navigate the challenges of team dynamics, understanding their instinctive conflict style can mean the difference between tension and transformation. This paper introduces emerging people leaders to Thomas-Kilmann Conflict Mode Instrument (TKI) and the five conflict styles: Avoiding, Accommodating, Competing, Compromising and Collaborating. Through interactive self-assessment, practical frameworks and real-world examples, participants will learn how their default style influences communication, team trust and decision-making. We will explore the strengths and pitfalls of each style and how to flex between them based on context, urgency and relationsh ip dynamics. Attendees will leave equipped with personalized conflict style compass, tools for adapting under pressure and strategies to guide their teams through conflict with empathy and clarity | clinical programming, life sciences | source presentation source |
4 |
| 65 | Evaluating the Use of Real-World Data (RWD) as a Synthetic Control Group: A Statistical Approach | Vasiliki Economou, Dimitrios Kounelis |
synthetic control, treatment effect, regression analysis, logistic regression, predictor variable, sensitivity analysis, sample, logistic distribution, clinical trial, odds ratio, propensity score, average treatment effect, propensity score matching, relative risk, proc phreg, clinical research, data cleaning | source poster source |
17 | |
| 66 | Everyday AI for Statistical Programmers: Global Implementation of a Clinical Analysis Assistant | Mathieu Cayssol, Christoph Centner |
The increasing complexity of clinical trial design and statistical analyses has significantly raised the expectations placed on statistical programmers. With the growing number of therapeutic areas, study designs, and regulatory requirements, programmers are tasked with producing high-quality, validated code under tight timelines. In many organizations, code bases are distributed across multiple projects and repositories, often resulting in fragmented knowledge and duplicated effort, particularly when creating critical Tables, Listings, and Graphs (TLGs). Re-using code remains a time-consuming, manual task. The lack of a centralized, intelligent mechanism to retrieve and reuse previously validated code fragments exacerbates inefficiencies and increases the risk of inconsistencies across studies. Enhancing code reusability is therefore a critical priority | clinical programming, statistical programmer, clinical trial, proc freq | source presentation source |
17 |
| 67 | FAIRchive Cloud-Based and GxP-Compliant Long-Term Archiving for Clinical Trial Data | Frank Stoinski, Alexander Lüders |
Archiving clinical trial data is the final step in the clinical trial process. The data and documentation generated during the course of a clinical trial must be archived and made accessible in a secure and validated environment for the prescribed retention period. This can be particularly difficult and costly as the commonly used data formats and software products may change over time. In addition, archives reveal a particular pattern of use where data is dormant for long periods of time until it is retrieved and reviewed again. This is where the cost factor comes into play, as ideally archiving solutions should only generate low costs for operation and maintenance while the data is at rest. Today's cloud providers support serverless billing models, which could provide a particular advantage here | amazon s3, relational database, clinical trial, proc sql, data cleaning, audit trail, clinical data, data science | source presentation source |
10 |
| 68 | FHIR for RWD Submissions: Why a US Regulation Signals a Global Shift And What You Need To Do Now | Sanyogeeta Andhale, Nachiket Thakkar |
In June 2025, the U.S. FDA closed public consultation on a landmark proposal: to accept HL7 FHIR as a standard for the submission of clinical data derived from real -world data (RWD) sources [1]. While this may appear to be a U.S. - specific regulatory develo pment, its implications are global. FHIR is already widely adopted across European health systems for EHR exchange, and its potential role in regulatory submissions will directly affect EU sponsors, CROs, and vendors engaged in multi-region trials [2]. This session explains why the FDA's move toward FHIR -based submissions represents a critical inflection point -not only for U.S. compliance, but for any organization leveraging EHR, claims, or registry data for global trials. Attendees will gain clarity on how this intersects with USCDI, TEFCA, and broader interoperability initiatives, and why now is the time to assess internal readiness, contribute to global alignment, and build a future-proof RWD strategy | clinical trial, clinical data, regulatory submission, data cleaning, clinical research, insurance claims | source presentation source |
11 |
| 69 | Fit-For-Purpose Approach of RBQM in Small Clinical Trials | Alicja Mark, Agnes Verhoeven |
Presentation | clinical trial, adverse event severity, clinical research, sample size, sample, bayesian statistics, data visualization | source | 15 |
| 70 | Floating Points Are as Easy as 01, 10, 11! | Chloe Floodgate | Why can't our computer do basic maths? Sooner or later as SAS programmer we will come across the dreaded floating-point error and what we see is not always what we get. Sometimes our SAS programs can produce some very peculiar results, and the intent of this paper is to give a basic overview as to why this could occur. We aim to understand what numeric representation and precision mean, or simply how SAS sees numbers, and describe what a floating-point error is | proc compare, proc step, decimal, where statement, numeric functions | source presentation source |
5 |
| 71 | Framework for Integrating Omics Data with Other Clinical and NonClinical Data Sources | Mehdi Harek, Jonathon Keeney |
The integration of omics data into clinical research is increasingly critical for advancing precision medicine, yet current industry standards, such as CDISC , do not fully address the complexity of omics datasets. This presentation introduces a Data Integration Framework designed to merge omics data with clinical and non-clinical sources, offering a comprehensive view of patient health and treatment outcomes. Key challenges will be discussed, including the heterogeneity of omics data formats, ensuring data quality and reproducibility across platforms, managing large -scale data storage, and the complexity of downstream bioinformatics analysis. The proposed frame work emphasizes standardization, interoperability, scalable data management, and reproducible analytics to support reliable integration and interpretation | clinical data, clinical research, clinical trial, big data, regulatory submission, data cleaning, data standardization, cdisc | source presentation source |
12 |
| 72 | From Breaking Bad to Breaking Data: Unlocking TFL PDFs | Andrés Käsa, Darius Tetsa |
Pharmaceutical companies hold vast unstructured data, such as PDF TFL outputs, which are difficult for machines to interpret due to complex layouts with headers, bodies, and footers across multiple hierarchies. While text extrac- tion from PDFs is straightforward, preserving structural organization and semantic meaning remains a challenge. Our work explores layout understanding and content segmentation of standard TFL outputs using techniques like clustering and computer vision. We highlight the potential of combining computer vision with natural language processing to further improve results. By enabling machines to better understand document structure and semantics, this research lays the groundwork for advanced downstream applications such as LLM-supported RAGs, metadata recommendation, process standardization, automated document generation, and SAS or R program skeleton cre- ation | sample, data cleaning, accuracy, clinical trial | source presentation source |
13 |
| 73 | From Hassle to Harmony: Unifying Code Quality Experience in R and SAS | Michael Thomsen | sas log, imputation, clinical data, data cleaning, clinical trial, macro definition | source presentation source |
9 | |
| 74 | From Mock to Output: Using Generative AI to Write SAS Code for Standards-Based TFLs | Kala Shivali, Bhuvana Venkatappa |
Manual authoring of SAS programs for clinical Tables, Figures, and Listings (TFLs) is repetitive and difficult to standardize across studies. This paper describes a metadata- driven framework that uses large language models (LLMs), guided by CDISC SDTM/ADaM structures and Analysis Results Standard (ARS) principles, to generate reviewable SAS code with end-to-end traceability. The design links a study metadata repository, leveraging Analysis Results Metadata (ARM) and TFL mock shells, to a prompt builder. It constrains generation with institutional macro libraries and static lint rules, and enforces determinism through low-variance decoding and fixed seeds. Governance considerations include model versioning, prompt and artifact registries, PHI redaction, and validation documentation aligned with audit expectations. The framework is intended to augment existing programming workflows while maintaining reviewer oversight, quality control, and compliance | data cleaning, quality control, mock shell, sas function, proc step, audit trail, proc freq, adverse events, clinical trial, demographics table, clinical programming, cdisc | source presentation source |
16 |
| 75 | From Outsourced to Open Source: Denalis R-Driven Transformation in Biometrics | Michael Thorpe, Thomas Neitmann |
In 2025, Denali Therapeutics, Inc., a biotechnology company based in South San Francisco, submitted its first Biologics License Application (BLA) to the US FDA via the Accelerated Approval pathway. This milestone not only marked Denali's inaugural BLA submission but also the company's first Clinical Study Report in which Biometrics deliverables were not fully outsourced. Instead, Denali adopted a hybrid operating model: the analysis workload was shared between an external vendor using SAS and Denali's in-house Biometrics team leveraging R within a validated Statistical Computing Environment. By choosing the open-source R language over proprietary SAS, Denali was able to build on the shoulders of giants, leveraging pharma-specific R packages developed and shared by the broader industry through initiatives like the pharmaverse | regulatory submission, data cleaning | source presentation source |
4 |
| 76 | From Overload to Insight: Personal Knowledge Management for Clinical Data Scientists in the Age of Generative AI | Sascha Ahrweiler | As knowledge workers in a regulated and rapidly evolving industry, clinical data scientists at all career levels are expected to learn fast, think clearly, and deliver reliably-often under pressure. This session introduces a modern approach to Personal Knowledge Management (PKM), combining classic systems like Zettelkasten and Building a Second Brain with the support of generative AI. Using practical examples, this paper explores how AI-powered search, summarization, and reasoning tools can augment daily tasks and support continuous learning. Scenarios from clinical programming show how these tools enhance insight generation, reduce cognitive load, and improve reproducibility-while maintaining data privacy and compliance. Beyond personal productivity, we explore how PKM practices can scale within teams-through shared knowledge bases, internal corporate knowledge sharing tools, and AI-assisted coaching models that promote collective learning | data science, clinical programming | source presentation source |
9 |
| 77 | From Programming Zero to programming Hero, how an internship shapes future programmers in a CRO | Marie Poupelin | Becoming a statistical programmer is a challenging leap, often compressed into a short onboarding process following graduation. But what if the learning process began 6 months earlier? Reflecting on my journey from intern to experienced professional, this paper takes a longitudinal perspective, examining the two-year impact of internships on skill growth and confidence building. Starting with training on dummy data to practice foundational techniques, programming from scratch and gradually transitioning to real studies with dedicated mentorship. I'll demonstrate how internships build essential skills while bridging academic theory and industry application. Well-structured programs prepare juniors for operational readiness and deliver significant organizational benefits. This paper is not about a career journey, it's an in-depth exploration of how internships create measurable value for both individuals and companies, seen through the lens of an intern-turned-professional | statistical programmer, pharmaceutical industry | source presentation source |
4 |
| 78 | From Schedules of Activities (SoAs) to USDM: Automating Protocol Extraction Using Large Language Models | Kerstin Forsberg, Johannes Ulander |
excel, clinical trial | source presentation source |
10 | |
| 79 | From SDTM to a Graph-Based, Multi-Purpose Interface for Data | Eva Kelty, Catharina Dahlbo |
As a clinical trial is completed, data is commonly organized and stored according to the SDTM standard, designed to standardize the organization and formatting of data to streamline collection, management, analysis, and reporting. However, because the data remains stored in a traditional table/relational database structure, it can be challenging to gain a comprehensive view of data. By moving data into a graph-based structure, new possibilities emerge for enhancing cross-functional collaboration. Unlike traditional table-based structures, where relationships between data points are manually defined, a graph- based interface connects related information, making complex relationships easier to visualize and explore through dynamic querying. By improving the accuracy and efficiency of data exploration, these tools support better decision- making and strengthen patient safety | adverse events, clinical data, sdtm domain, cdisc | source presentation source |
4 |
| 80 | From Standards to Strategy: Customising CDISC Open Rules for the TMF Reference Model | Irina Sargsyan, Sabine Jonas |
Presentation | life sciences, data cleaning, excel, data standard, xls | source | 20 |
| 81 | From Static Archives to Analytical Intelligence: Transforming Statistical Analysis Value Through Standardization | Ilan Carmeli | Current models driving the generation, delivery, and reuse of statistical analyses are ill-equipped to maximize the full strategic value of the insights generated in tables, listings, and figures (TLFs). Static displays lock value in the highly variable, non-standard structures of PDFs, RTFs, HTMLs, and other proprietary output formats, when a more structured storage approach would yield far greater returns on investment. Statistical output rendered into standardized, machine-readable formats allows for applications to more fully exploit the value generated in the statistical analysis phase of clinical trials. This paper explores practical implementations of AI-driven table analysis, standardized metadata frameworks, and automated generation capabilities that demonstrate measurable improvements in efficiency, traceability, and analytical reusability | data cleaning, data science, adam dataset, cdisc | source presentation source |
8 |
| 82 | From Statistical Programming to AI Automation in Data Management: A Cross-Departmental Tech Journey | Irene Luna, Marcia Vaz |
clinical trial, data science, clinical data, clinical programming | source presentation source |
5 | |
| 83 | From XPT to Dataset-JSON: Enabling Real-Time Validation and Streamlined Submissions | Hugo Signol, Sebastià Barceló |
Context Dataset-JSON overcomes XPT limitations. It eliminates the long-standing restrictions of SAS Transport (XPT) format, its compact footprint makes it well-suited for API-based workflows, reducing data management complexity. By leveraging APIs, Dataset-JSON removes the need for file-based data handling, enabling direct use of databases as a single source of truth. Furthermore, APIs allows efficient data filtering before loading, reducing unnecessary data transfer and improving overall efficiency. Moreover, this format combined with APIs accelerates the regulatory submission process by fostering early collaboration between sponsors and regulatory authorities. Data can be shared as soon as it is collected and transformed into standardized formats such as SDTM, SEND, or ADaM. This early exchange of information enables regulators to start their review process sooner, leading to faster approvals | data cleaning, excel, cdisc | source presentation source |
8 |
| 84 | FUN WITH FIGURES | Miriam Amor | Emojis, memes, pictures, diagrams, etc: images are increasingly used in the way we communicate, and clinical trials are not the exception. With this new trend, other programming languages like R are being used to create figures. However, when it comes to SAS, the community is facing a lack of resources with deep knowledge in graphics, specifically in the Graphic Template Language (GTL). It is not uncommon that project leads are on the hunt for skilled programmer in those SAS procedures across their teams. Graphics and not only a powerful tool for data-driven decision making, but also an artistic way of displaying data. Though GTL can be scary at the beginning, like any art, one can easily learn and develop new technical skills with a few tips and a bit of practice. This presentation aims to share some tips and tricks to start having fun with figures in SAS | proc step, proc template, proc sgplot, by group processing | source presentation source |
10 |
| 85 | GLMM and GEE for Categorical Repeated Measures - Linking SAS and R | Miriam Amor | Since R started being used in clinical trials, hybrid SAS and R programming has become a reality. Collaborative initiatives, such as like PHUSE's CAMIS working group, are leading the way in facilitating the alignment between different languages. In statistics, repeated measures are frequently analysed, with MMRM models being well-known models to analyse continuous variables. However, when the outcome is categorical, different methodologies are needed. Statistical methods that rely on a link function such as GEE and GLMM (using Laplace approximation, PQL or GHQ) become key alternatives. While SAS provides procedures like GENMOD and GLIMMIX, R offers functions as glmer, glmmPQL and geeglm, and the combination of both is the perfect toolkit to analyse categorical variables with repeated measures. This presentation aims to show how to use R and SAS to program different models when the outcome is a categorical variable (binomial and multinomial) with repeated measures | proc step, proc mixed, correlation, proc glimmix, covariance, fixed effects, proc genmod, proc calis, likelihood function, degrees of freedom, covariance structure, multinomial distribution, repeated measures, nominal variable, response variable, binomial distribution | source presentation source |
16 |
| 86 | gridify: Enrich Figures and Tables with Custom Headers and Footers | Maciej Nasinski, Alexandra Wall |
safety population, analysis set, pharmaceutical industry, treatment arm, clinical trial | source poster source |
10 | |
| 87 | Guiding the direction of open-source development for faster end-to-end HTA submissions | Christian Olsen | data wrangling, clinical programming | source presentation source |
10 | |
| 88 | Harnessing the Power of Interactive SAS Outputs Through SASpy in a Flask | Jamie Pilkington | Python is equipped with a suite of robust tools that facilitate the creation and deployment of web applications. Building upon prior research that has already demonstrated the interoperability between Python and SAS using SASPy, this presentation aims to demonstrate the development of a Python-based web application, using a package called Dash, capable of hosting dynamic SAS outputs. These outputs can be interactively updated through the user interface of the web application. The paper will be divided into three sections: an introduction to SASPy, an overview of the fundamental concepts of a Dash web application, and a live demonstration. This paper will feature a graphical output that exemplifies the potential modifications, alongside the corresponding SAS code generated on the backend | macro variable, dataset, proc step | source presentation source |
5 |
| 89 | Hash-ing It Out: Bringing SAS to the Next Level | Lara Suckut, Veronika Schmidt |
As data structures and utilization become larger and more complicated every day, efficient, time-saving solutions are needed to extract targeted information from multiple, possibly differently structured data sets. SAS hash tables provide quick and flexible ways to merge, sort, and look up data by taking information directly from the source instead of generating unnecessary data duplication that slows down processing and clutters up storage space. Additionally, using hash tables can be beneficial by saving time and reducing data steps during programming. We provide a thorough introduction to the basic definitions, concepts, and benefits of hash objects. By using examples that are practical and applicable to the work of a statistical programmer in the pharmaceutical industry, we aim to provide the audience with a firm grasp of the opportunities offered by this method and ideas on how to use hash tables to solve their current data challenges | hash table, data step, concomitant medications | source presentation source |
7 |
| 90 | How Can Agile R Package Validation Workflows Be Achieved? | Sophia Stahl-Toyota, Maximilian Kreienbaum |
Getting from feature requests or bug reports to the next validated package release in an agile fashion is the vision of many R package product owners, developers, validation managers and users. Add further colleagues from quality departments, deployment pipeline operations and management to the list of stakeholders and we see that the topic heavily relies on alignment of people and processes in addition to tools. We present how the tooling landscape for managing software development can be set up with sophisticated use of the technical possibilities offered by platforms such as GitLab, Jira, Microsoft Azure DevOps and GitHub. They include workflows, issue statuses, labels, scoped labels, tags, components and timing related iterations, sprints, and milestones. However, efficient configuration and use of those options is the art to master for effective tracking of development and validation activities that involve many review steps | data cleaning, excel, r package | source poster source |
11 |
| 91 | How Does a Computer Really Process Your SAS Code? | Ben Barnaby-Pass | One of the most fascinating aspects of programming is how the machine you are working with takes in and runs your code. SAS is a complex language and is just like learning any other different language, be it Spanish, English or German. But do you really understand any language without looking at the logic behind it? Learning the logic of a language is the main reason why I believe learning SAS is wonderful! Understanding how a computer can take a very robust language and turn it into SDTMs/ADaMs/TFLs is eye-opening and really opens your mind to how you can manipulate SAS code into being the most efficient code you can produce. Be it drop/keep or a simple IF statement, you can make SAS create the most wonderful things! | where statement, drop statement, if-then-else, keep statement, cpu time | source presentation source |
5 |
| 92 | How To Become a More Productive Programmer Using LLMs (Lots of Limb Movements) | Thomas Bold | Presentation | source | 42 | |
| 93 | How to Identify and Operationalize GenAI Use Cases Which Really Help You | Christoph Bergen, Luis Wirth |
Generative Artificial Intelligence (GenAI) has emerged as a transformative technology. Distinguishing its value within regulated and complex domains such as the pharmaceutical industry remains challenging. This paper presents an introduction to guide decision-makers in evaluating where and how to apply GenAI effectively. We begin by contrasting classical, discriminative AI approaches with generative models, outlining the types of problems best suited for each paradigm. Through real-world pharmaceutical examples, we illustrate how GenAI models can be leveraged and highlight infrastructure and processes required to translate a model into a productive and adopted solution. Furthermore, we discuss strategies for operationalizing GenAI within enterprise environments, ranging from self-service AI platforms to distributed systems of GenAI tools | machine learning, artificial intelligence, deep learning, model training, training set, pharmaceutical industry, data cleaning | source presentation source |
8 |
| 94 | How to map lab data into the LB domain and ADLB | Bert Parmet | Capturing laboratory data in compliance with CDISC standards can be challenging due to various factors. Understanding the data requires a basic knowledge of medical concepts to assess validity. It is necessary to convert the diverse units used by different sites into single standard units to ease comparing the data and facilitate data analysis. Some lab values must be derived from existing measurements, like the Creatinine Clearance. Flags must be applied to out-of-range values, sometimes involving data from multiple time points. In oncology studies, lab values need to be graded according to the NCI CTC standard. This presentation offers a detailed guide on mapping lab test results to CDISC standards, starting with a medical context overview, followed by mapping lab results to SDTM. Finally, creating ADLB is discussed, covering key tasks like deriving calculated lab values, setting flags for clinical concerns, and applying NCI CTC grading | controlled terminology, clinical database, laboratory data, sample, data science, ctcae grade, adverse events, cdisc | source poster source |
6 |
| 95 | Human in or Human out of the Loop: Impact and Things To Consider While Deploying AI | Olivier Bouchard | clinical research, pharmaceutical industry, artificial intelligence, machine learning, clinical trial, life sciences | source presentation source |
8 | |
| 96 | Human in the Middle: Accelerating with AI, Anchored in Quality | Melanie Hullings, Christopher H. White |
The integration of artificial intelligence (AI) in regulated environments presents unique challenges, particularly regarding data integrity, traceability, and accountability. In industries governed by strict compliance standards, such as life sciences, finance, and healthcare, leveraging AI requires careful consideration of how decisions are made, validated, and documented. This talk explores the concept of "Human in the Middle" as a practical framework to bridge the gap between AI capabilities and regulatory requirements. By maintaining human oversight in critical decision-making processes, organizations can harness the efficiency and power of AI while preserving compliance and ensuring transparency. The session will examine real-world applications of this model, highlighting how "Human in the Middle" approaches can enable trustworthy AI deployment without compromising on regulatory expectations | data cleaning, data integrity, life sciences, clinical research | source presentation source |
6 |
| 97 | Hybrid Way To Develop Patient Narratives | Loskutov Volodymyr, Oleh Buno |
Presentation | data cleaning, statistical programmer, adverse events, csr, serious adverse event, machine learning, missing value | source | 16 |
| 98 | I Propose Aspose! Automating PDFs and Microsoft Office Files | Dominik Habel | This paper looks at the commercial software Aspose which offers numerous file level APIs to work with Microsoft Office file formats as well as PDFs and many more. It discusses how you can use Aspose for Java in a SAS environment and shows several practical examples on how using Aspose can help you with your daily tasks | proc groovy, proc step, variable | source presentation source |
7 |
| 99 | IDD'AI: Automating the Tedious So You Can Focus on the Genius | Lionel Debecq, Samantha Cambier |
In the fast-paced world of clinical research, who has time to manually write SAS programs from mock tables? Enter our AI-powered automation tool, IDD'AI, designed to make your life easier and your work faster. This innovative tool aims to use Natural Language Processing (NLP) to read mock tables like a pro, picking out the stats you need and variables from the ADaM metadata. And here's the kicker: our LLM will rank potential matches by confidence level, so you spend less time hunting for the right code and more time doing what really matters. By automating the boring bits, we aim to help you get the job done quicker and with fewer mistakes. This proof of concept (PoC) by Euranova and IDDI will evaluate efficiency gains, AI accuracy, and regulatory compliance (GDPR, European AI Act) while keeping human oversight for reliability. Say goodbye to tedious programming and hello to a smoother, more efficient workflow! | adam dataset, proof of concept, proc means, adverse events, proc step, data science, cdisc | source presentation source |
8 |
| 100 | Implementing ADNCA in the PK Process: A Cross-Departmental Approach | Daphne Vuurstaek, Frédéric Saad |
PK-related data formatted as ADaM datasets are being requested more frequently by health authorities. This requires study teams to prepare submission -ready ADaM PK-related data, on an adhoc basis without prior planning. Over the years, the International Society of Pharmacometrics (ISoP) & CDISC have partnered closely to standardize Non- Compartmental Analysis (NCA) data formats establishing the new regulatory standards which have been adopted by J&J. The new ADaM BDS data structure, ADNCA, necessitated process improvements to optimize the delivery of the required datasets. To address challenges and minimize last-minute re-work moving forward, we initiated a collaborative effort across multiple departments to reform our NCA data flow. This poster presents insights into the modifications we implemented utilizing an ADNCA dataset as input for the NCA analysis | clinical programming, adam dataset, compartmental analysis, cdisc | source poster source |
4 |
| 101 | Interactive Exploration of Top Line Results: Integrating DaVinci with Teal | Fabian Roethleithner, Isabel Glauss |
Exploring top line results after database lock in clinical trials is crucial for quickly assessing effectiveness, safety, and statistical significance to evaluate success of a medication. Using modular R/Shiny apps for interactive data visualization enhances this process significantly by enabling dynamic and trial-specific analyses. This presentation introduces an approach for interactive exploration of top line results, based on a combined approach of DaVinci and Teal modules. DaVinci integrates seamlessly with Teal modules allowing for a quick and comprehensive visualization of top line results, facilitating storytelling and decision-making in the clinical trial process | clinical trial, data visualization | source presentation source |
5 |
| 102 | Interactive Insights: Leveraging R Shiny for Enhanced Data Visualisation in Clinical Trial Readouts | Carolyn Fisher, Elisabeth Deutschmann |
Presentation | phase ii, primary endpoint, forest plot, survival analysis, phase iii, data science, clinical trial | source | 20 |
| 103 | Inverse Probability Weighting in Multiple Groups: Balancing Baseline Characteristics and Improving Precision of Average Treatment Effect Estimation | Dimitris Karletsos | propensity score, sample, sample size, multinomial logistic regression, logistic regression, regression analysis, baseline characteristics, logistic distribution, treatment effect, hazard ratio, power analysis, proc phreg, randomization | source presentation source |
11 | |
| 104 | Is There Really Any Benefit to Stratified Randomisation Compared to Simple Randomisation with Covariate Adjustment? | Pavankumar Bhagat | Randomized controlled trials (RCTs) are the gold standard for evaluating interventions in clinical research. Stratified randomization is a common technique used to balance prognostic factors across treatment groups, aiming to increase the precision of treatment effect estimates. However, its effectiveness and efficiency compared to simple randomization with covariate adjustment remain a subject of debate. A simulation study as part of master's thesis at university of Manchester was performed to compare the performance of stratified randomization and simple randomization with covariate adjustment under various conditions, including different sample sizes, association levels between covariates and outcomes, stratification levels, and treatment effects. By employing a linear regression model, key performance metrics such as power, type I error rate, empirical standard error, bias, and coverage probability were assessed | treatment effect, type i error, significance level, sample size, sample, proc calis, clinical trial, confidence interval, randomization, proc reg, regression analysis | source presentation source |
9 |
| 105 | JazzAIR: Advancing Clinical Study Reporting Through Modern Development Practices and Testing Strategy | Jakub Sobolewski, Chen Yang |
Presentation | quality assurance, clinical trial | source | 34 |
| 106 | JSON for Submissions: Why You Should Care, What You Need To Know, and What To Do Now | Bill Qubeck, Sridhar Vijendra |
The FDA has published a Federal Register Notice seeking public comment on the potential adoption of CDISC Dataset-JSON v1.1 as a new exchange standard for regulatory submissions marking a pivotal moment in the evolution of clinical data delivery. Dataset-JSON, a structured, machine-readable format, is being considered as a long-term replacement for the legacy SAS XPT format. For sponsors, CROs, and technology providers, this signals a shift that will require new tools, revised processes, and a deeper focus on metadata governance and validation strategies. This session explores why Dataset-JSON matters, what the format offers, and how organizations should prepare. Building on insights from the recent PHUSE/CDISC pilot, we will examine Dataset-JSON's key attributes, including support for complex metadata, native compatibility with modern programming languages (e.g., R, Python), and its potential to enable real-time validation and automation | data cleaning, proc json, proc step, missing value, dataset, regulatory submission, sas programmer, data integrity, clinical trial, ods json, clinical data, sysdate9, sample, clinical research, cdisc | source presentation source |
13 |
| 107 | Keeping it Real: A Tiered Approach to Real-World Data Quality by Leveraging Synthetic Data | Lisa Murch, William Kuan |
Data quality remains one of the most significant barriers to robust real -world evidence (RWE), particularly in Europe , where fragmented healthcare systems, local coding standards and data formats , and GDPR restrictions complicate access. We outline a tiered approach to improving real-world data (RWD) quality with SAS Viya, a data and AI platform. We begin with data profiling and rule -based checks to identify common data quality issues such as missingness, temporal inconsistency, and non-standard codes across EHR, claims, and registries. Next, we explore the potential of synthetic data for addressing the challenges of data quality, and discuss how synthetic data can be used to fill the data gaps, mitigate bias and enhance privacy compliance. Finally, we explore the challenges of integrating proprietary and open source tools while maintaining compliance with regulatory standards and the goals of European RWD networks such as DARWIN EU | data cleaning, machine learning, sas viya, data audit, life sciences, missing value, sas studio, unsupervised learning, lookup, data wrangling, outlier | source presentation source |
10 |
| 108 | Landing Moonshots | Ross Farrugia | In 2024, at Roche/Genentech we achieved an ambitious goal of moving 90% of our teams and 1,000+ data scientists to a completely new platform and way of working with our inhouse built | data science, artificial intelligence | source presentation source |
8 |
| 109 | Large Language Models as Medical Writers: Challenges and Solutions | Ioannis Reklos, Alexandrina Lambova |
The creation of Systematic Literature Review (SLR) reports and Health Technology Assessment (HTA) dossiers is crucial in medical writing, often requiring extensive expert effort. We examined the use of Large Language Models (LLMs) to automate key steps, specifically information extraction and medical text generation. We identified core challenges in AI-assisted document creation. For information extraction, issues arose when information appeared in text and tables. We addressed table-related problems, such as multi-page splits and recognition errors, through custom solutions, achieving substantial improvements. Text-related challenges included topic-specific prompt tuning and Retrieval Augmented Generation (RAG) failures. We mitigated these with topic classifiers for improved retrieval. Extraction accuracy was assessed against human-generated gold standards. In text generation, we faced challenges with content synthesis and granularity control | logistic regression, regression analysis, logistic distribution, data cleaning, normalization | source presentation source |
10 |
| 110 | Launching CDISC CORE v1.0: A New Standard for Automated Study Conformance | Tomás Sabat Stofsel, Nick De Donder |
Presentation | proc sql, regression analysis, data cleaning, data validation rules, controlled terminology, cdisc | source | 44 |
| 111 | Leading through talents | Karolina Filar | The part of the project lead role that I enjoy the most is recognizing the talents in the team, often faster than the talent owners themselves. I have created a mentoring program, giving people the exposure and variety of the tasks that they needed to glow. The outcome has been amazing; my mental load as a project lead has dropped significantly whereas people have gained a safe environment where they could experiment and learn without putting the quality and timing of the deliveries at risk. I find this approach towards managing a team energizing. I would like to share my story and inspire team leaders to look at the new joiners way broader than through their years of working experience. In my presentation I will go through specific examples a nd recommendations for developing talent -focused management style | data wrangling, clinical trial | source presentation source |
8 |
| 112 | Leading with Emotions Positive Leadership and the Power of Optimism | Goekhan Oezcan | Emotions play a central role in leadership, and leadership itself has increasingly become a matter of building relationships. Positive Leadership integrates scientifically grounded insights from Positive Psychology with a leadership style rooted in trust, meaning, appreciation, and optimism. This presentation explores how emotions can be used effectively in leadership to foster motivation, engagement, and psychological safety within teams. At its essence, optimism serves as a powerful force - an inner mindset that enhances clarity, builds resilience, and empowers decisive action. Participants will gain practical insights into cultivating a leadership culture that is both positive and emotionally intelligent - and discover why, particularly in challenging times, leading with empathy and focusing on strengths is crucial for success | source presentation source |
8 | |
| 113 | Lets Not Get Lost in Translation: Open-Source Adoption at Novartis | Mark Baillie, Orla Doyle |
Open-source technologies are gaining traction in clinical trial analysis and reporting. At Novartis, we are modernizing our approach through two key efforts: (i) the increased adoption of open-source tools ("the furniture") and (ii) the implementation of a next-generation Statistical Computing Environment (SCE) ("the house"). These efforts are driven not only by technological opportunity but also by the evolving scientific and regulatory landscape. For example, recent guidance such as the ICH E9(R1) addendum on Estimands [3] and the FDA's covariate adjustment [4] recommendations have introduced new analytical expectations that often require novel methods, many of which first emerge in open-source ecosystems or must be developed internally. This context has motivated a value-driven approach that avoids mere "code translation" and instead focuses on improving the quality, flexibility, and efficiency of analysis and reporting | clinical trial, phase iii, clinical data, independent qc, data cleaning | source presentation source |
6 |
| 114 | Leveraging AI for Seamless SAS to R Migration: Challenges and Opportunities | Milena Kraus, Corrina Stoeckinger |
While in the past SAS has been the de facto standard for the statistical analysis of clinical trials, in recent years a growing portion of the industry has had their eyes on open-source alternatives. Arguably the most popular is the programming language R. To support Boehringer Ingelheim's ClinPharm task force for the transition from SAS to R a team of experts from HMS was brought in. Here, we present our joint effort for the migration of a collection of SAS macros and templates in the context of PK/PD analyses into R using GenAI to streamline the process. We discuss the different approaches we piloted including direct translation as well as utilizing existing packages of the pharmaverse and the pros and cons of each | data cleaning, adam dataset, clinical data, clinical trial, macro definition, data science, population pk, sdtm domain, cdisc | source presentation source |
11 |
| 115 | Leveraging Generative AI for SAS Macro Testing and Test Data Preparation | Denis Gavriusev | GenAI is transforming clinical SAS programming by automating macro testing and test data preparation. Developers can leverage Generative AI to auto-generate comprehensive SAS macro test cases, spanning routine executions, boundary scenarios, and invalid parameter combinations, to ensure robust validation. GenAI can also generate code that converts real or synthetic datasets into test-ready data: anonymizing and masking sensitive fields, and enriching datasets with both routine and edge-case values to meet testing requirements without exposing confidential information. This integrated approach streamlines testing workflows, accelerates quality assurance, and upholds data privacy standards, empowering teams to deliver reliable and compliant analytical tools more efficiently | data cleaning, macro definition, data wrangling, ctcae grade | source presentation source |
10 |
| 116 | Leveraging LLM for R Shiny App Code Development: A Case Study with Event Prediction | Yirong Cao | Presentation | training set, data science, clinical programming, histogram, randomization, data cleaning | source | 15 |
| 117 | Leveraging on R packages for HTA Analysis & Reporting: Our Developer Journey | Jean Muller, Wenyu Li |
In the rapidly evolving landscape of Health Technology Assessment (HTA), R provides early access to new and/or advanced peer-reviewed methodologies essential for timely and efficient Analysis & Reporting (A&R) deliveries. This presentation will showcase our development of customized R packages designed to enhance our HTA A&R toolkit. The emphasis is on the importance of maintaining a systematic structure and adhering to good programming practices to establish standards and ensure long-term code maintainability. We discuss the advantages and drawbacks of the R package format, with a particular focus on the trade-off between package functionality and interdependencies. By sharing our journey in R development, the aim is to provide valuable insights to assist other R developers in traceable, reproducible, and reliable HTA A&R deliverables | data cleaning, proc means, r package | source presentation source |
8 |
| 118 | Leveraging Open-Source Python Tools to Improve Efficiency and Transparency in Clinical Trials | Sohan Lal, Michael Olds |
In clinical trials, it is important to have clear, efficient, and repeatable processes that ensure data accuracy and speed up clinical research. Python is a widely used programming language that supports these goals with its versatility and many helpful tools. This paper explains how Python improves clinical trial workflows. Using libraries like Pandas, NumPy, PyMuPDF, and SciPy, researchers can organize and analyse data more effectively and efficiently. With machine learning tools like Scikit-learn, researchers can also automate tasks and gain faster insights. Python works well with different file types and tools, allowing for extraction of data from PDFs, managing large datasets, and automating documentation. This makes it easier for teams to collaborate and make better decisions. This paper will also highlight existing Python-based solutions that support clinical trials in various ways | clinical trial, clinical research, machine learning, clinical data | source presentation source |
8 |
| 119 | May the Code Be with You: Entering the pharmaverse Galaxy | Kirsty R. Lauderdale | Presentation | data cleaning, clinical trial | source | 17 |
| 120 | Micro-Decisions, Macro Impact: The Role of Agile Thinking in Every Line of Code | Eswara Satyanarayana Gunisetti | In the world of statistical programming, agility isn't just about stand-ups or sprints, it's woven into every tiny decision a programmer makes. From how we name variables, design macros, structure datasets, or write comments, every line of code either builds flexibility or bakes in future friction. This talk explores how agile thinking: clarity, simplicity, adaptability, and value orientation can be applied directly at the code level. We'll discuss how small programming habits, like modularizing code, documenting intent (not just process), and writing for readability, have outsized impacts on reusability, QC efficiency, and audit readiness. Through real -world examples from SDTM, ADaM, and TLF workflows, we'll show how these micro -decisions lead to macro gains: faster change implementation, clearer handovers, and a stronger sense of ownership. This is not about doing more, it is about thinking agile while doing what we already do, better | statistical programmer, proc sort, data cleaning, cdisc | source presentation source |
5 |
| 121 | Milestones Matter: Understanding Key Milestones in Clinical Trials | Satyapal Gudla | As statistical programmers within the pharmaceutical industry, we primarily work on analysis datasets, tables, listings, and graphs that contribute to a wide variety of milestones in the journey of clinical trials. These milestones are critical for making important decisions about the study, from inception to final analysis, spanning phases from First-in-Human trials to the drug's market release, including post-marketing analysis. This poster/paper will provide an overview of the diverse milestones we contribute to, highlighting the purpose and scope of analyses, key deliverables, stakeholders, timelines, and important decisions or outcomes associated with each milestone | statistical programmer, csr | source poster source |
21 |
| 122 | Mind the Gap: Imputing Missing Dosing Times in QD & BID Regimens Using ADPPK Guidelines | Aschwin van der Giesen, Gerly van der Vleuten |
Missing dosing times in pharmacokinetic datasets present a frequent challenge in clinical data programming, potentially affecting dataset consistency and interpretability. The CDISC ADPPK guidelines provide a standardized framework for imputing these missing values while preserving data integrity and traceability. This paper explores practical imputation strategies for once-daily (QD) and twice-daily (BID) dosing regimens, including interval-based methods, protocol-defined administration window alignment, and PK sampling time-based approaches. A case study illustrates the application of these techniques within a reproducible programming workflow, demonstrating improved dataset completeness and regulatory compliance. By adhering to ADPPK standards and documenting imputation decisions explicitly, programmers can enhance dataset consistency, minimize ambiguity, and support scientifically robust pharmacometric analysis | imputation, sample, sensitivity analysis, missing value, data cleaning, adam dataset, cdisc | source presentation source |
8 |
| 123 | Modernising Clinical Data Science: Best Practices for a Unified, Governed SCE | Matthew Tendler | Presentation | data cleaning, audit trail, data pipeline, data science, independent qc, raw data, excel | source | 13 |
| 124 | More Science, Less Bookkeeping: Our Path to Hyper-Efficient ADaM Program Generation via AI, Process Improvements, and Good Old-Fashioned Programming | Matthew Phelps, Nicolai Skov Johnse |
data cleaning, clinical trial, adam dataset, raw data, clinical data, cdisc | source presentation source |
10 | |
| 125 | Multiple Imputations Overview of Missing Data Problem, Solution Methods and SAS Implementation | Dmytro Kalinin | In any study, there is often data that was intended to be collected but was not, due to various reasons such as patients not completing the study, missed visits, equipment failures, and other unforeseen circumstances. Missing data not only diminishes the p ower of statistical tests but can also lead to erroneous conclusions. There are numerous imputation methods available, such as Last Observation Carried Forward (LOCF), worst-case scenario, group averages, etc. However, the most reliable and universally app licable method is multiple imputations (MI). In this presentation, we will explain how to utilize multiple imputations in your analysis, the appropriate methods to select, and how to implement these techniques in SAS. Moreover, we will share our experience s in creating ADaM datasets for MI results. This article does not aim to present scientific research or novel discoveries | imputation, proc step, missing value, proc mi, multiple imputation, proc mianalyze, proc calis, missing not at random, proc mixed, adam dataset, out=, missing completely at random, normal distribution, logistic distribution, missing at random, regression analysis, baseline characteristics, in=, estimate, wilcoxon test, binomial distribution, anova, sample, proc freq, data science, correlation, proc genmod, cdisc, last observation carried forward | source presentation source |
13 |
| 126 | Navigating Anonymous Employee Feedback: A Managers Guide to Constructive Action | Lynn Clipstone | In today's dynamic workplace, anonymous employee surveys are vital for understanding team sentiments. However, transforming this feedback into actionable strategies, while remaining detached from the sometimes critical feedback, poses significant challenges. This presentation will draw from my years of experience to help managers effectively review and act on survey results. I will share tips on motivating your team, as a manager of managers, to develop actionable strategies. We will focus on the primary themes and different lenses to look at this with. We will also look at cultural differences within and between teams to understand and harmonise global results. Join me to discover how you can turn anonymous feedback into a catalyst for positive change, enhancing both individual well-being and organizational success | data visualization, heatmap | source presentation source |
4 |
| 127 | Navigating Experimental Arms in Subtype Studies: Strategies for Precision and Clarity | Narek Torosyan, Edgar Sargsyan |
As clinical trials continue to evolve alongside the principles of precision medicine, subtype studies designed to assess treatment effects within biologically or clinically defined subgroups are gaining prominence. These designs offer critical insights into which patients benefit most from specific interventions. However, they also present significant challenges, particularly in managing multiple experimental arms tailored to different subpopulations. Incorporating multiple experimental arms raises questions around comparability, statistical power, multiplicity control, and regulatory expectations. Without careful planning, these complexities can undermine interpretability and limit regulatory acceptance. This paper presents structured strategies for navigating experimental arms in subtype studies, focusing on design precision, bias prevention, and regulatory clarity. Real-world examples from oncology, rare disease, and immunology will be used to illustrate best practices | power analysis, clinical trial, hierarchical testing, randomization, sample | source presentation source |
4 |
| 128 | Neither Magic Wand Nor Terminator: Finding the Sweet Spot in AI-Assisted Coding | Josua Böser | Remember debugging that cryptic bug until midnight, only to discover a missing semicolon? Now imagine explaining that problem to an AI that alternates between writing flawless algorithms and confidently suggesting you name your variables "myAwesomeString42". Welcome to the bizarre reality of AI-powered coding! While tech prophets either promise coding nirvana or predict developer extinction, most of us just want help with regular expressions without selling our minds to the AI companies. This talk cuts through the hype to show how AI coding tools work in the trenches - when they're brilliant, when they're confusing, and how to tell the difference. We'll demonstrate practical techniques for using AI as your coding co-pilot: one that helps navigate unfamiliar frameworks, suggests better approaches, and occasionally needs its enthusiasm tempered with human wisdom. Learn to harness AI's strengths while retaining the satisfaction of solving problems yourself | independent qc, error message | source presentation source |
12 |
| 129 | NeuroDriven Mentorsip: Recognizing and Teaching Different Kinds of Thinkers | Ellada Abrahamyan, Armenia Arman Melkonyan |
This paper presents a cognitive-informed approach to mentoring clinical programmers, grounded in both psychological theory and field experience. It defines six thinking styles-linear, visual, pattern-driven, conceptual, experimental, and emotionally attuned-that shape how individuals learn, code, and respond to feedback. The | source presentation source |
8 | |
| 130 | New Multiple Imputation Approaches with SAS | Andrea Parodi | This paper presents a practical implementation of the Jump to Reference (J2R) approach using conditional mean imputation, comparing it to traditional multiple imputation. We describe the methodology, provide SAS code examples, and summarize results from a simulation study. The goal is to offer a reproducible and computationally efficient alternative for handling missing data in clinical trials, with a focus on accessibility for programmers and statisticians | imputation, mean imputation, multiple imputation, type i error, clinical trial, missing value | source presentation source |
2 |
| 131 | No Patients? No Problem Define Your Study Data with Python at the CORE | John McDade | In clinical trials, access to study data is often delayed due to challenges around patient recruitment. Slow patient recruitment can challenge downstream activities which are often more time consuming such as complex ADaM and TFL creation. Test data from an EDC is often counterproductive for downstream programming as makes little clinical sense. This paper presents a python application that can generate CDISC-compliant SDTM data to allow downstream work to progress at the convenience of the study team. The application leverages a study define.xml to use study specific metadata to drive SDTM generation - or simply configure the number of patients if only generic data is required. The data generated can be in many different formats including dataset JSON and has real time validation using CDISC CORE [1], taking advantage of recent industry initiatives. Keywords: CDISC, Dataset-JSON, SDTM, Synthetic Data, CDISC CORE, Define-XML, Data Standards, CDISC 360i | data cleaning, adverse events, excel, clinical research, outlier, cdisc | source presentation source |
11 |
| 132 | Oh Dear, Another Crisis? A Programmers Guide to Surviving and Thriving Through the FDA ODAC | Zach Dorman | You've put in several months of hard work to complete your datasets, TLFs and submission materials, worked through hours of regulatory response, only to now be told that your study is subject to an Oncologic Drug Advisory Committee (ODAC) meeting held by the FDA. Is it time to panic, or time to shine knowing that you've prepared your team for just this eventuality? In this paper, I will talk through my experiences of going through an ODAC as a programming lead and oversight - from the initial stages of not knowing what to do or expect, right through to the day of the ODAC itself. I hope to demystify the ODAC process, share some successes and lessons learnt, and give you some tools for how to thrive both as an individual and programming function in this pressured environment, should you be lucky enough to be involved in one in the future | interim analysis, line chart | source presentation source |
12 |
| 133 | On the Right Track: Managing Platform Modular Studies with a Central Metadata Switchboard | Dubravka Georgijevic, Carla Santillán |
A platform modular study is a type of adaptive trial that uses a master protocol to evaluate multiple, targeted therapies within a single disease setting. These trials may not have a planned end date, and treatment arms (and/or modules) can be added, or removed, based on predefined rules. For statistical programmers, managing these studies often feels like handling multiple studies within one database. A key challenge is the ongoing series of module-specific analyses, each requiring unique derivations and output modifications. This paper introduces a metadata-driven solution using a central file that controls which derivation, or output, adjustments should be applied per study module, and delivery. This file is read into SAS to generate macro variables used in SDTM, ADaM, and TLF programs for conditional programming | macro variable, interim analysis, analysis set, sdtm domain, data science, clinical trial, cdisc | source presentation source |
7 |
| 134 | One sample Log-Rank test for one sample survival studies: SAS code vs R code | Giulio Pacini, Menarini Ricerche, Giulia Tonini, Menarini Ricerche, Simona Scartoni, Menarini Ricerche |
In single-arm clinical trials with survival outcomes, the Kaplan-Meier estimator and its confidence interval are widely used to assess survival probability and median survival time. The intent of this work is to write a code to compare the survival curve of a single treatment group to that of an historic control. Such is often the case in clinical phase-II trials with survival endpoints. Currently there is no specific PROC that can be applied to such analysis in SAS. There is availability of a library in R, to calculate the One Sample LogRank test (OneArmTTE). We are here presenting a calculation of the One sample LogRank test in SAS, which reproduce the exact results of the R library OneArmTTE. We also analyze through simulations the performance of such testing |
sample, rank test, sample size, confidence interval, null hypothesis, log-rank test, weibull distribution, time-to-event endpoint | source presentation source |
5 |
| 135 | Open-Source Practicalities and the Legal Side | Katja Glaß | The rising popularity of open-source within the pharmaceutical sector presents a multitude of opportunities. However, next to its benefits there are uncertainties surrounding permissible usage. Did you know that source code in a paper default to copyright protection? Moreover, the prevalence of copy- left licenses in R raises questions about the obligation for R-based software solutions to also adopt copy- left open-source practices. This presentation aims to demystify the legal aspects of leveraging open-source in pharmaceutical solutions while ensuring compliance with the respective open-source licenses. It will offer insights, shedding light on how to appropriately utilize open-source solution through deriving or enhancing to a commercial or open - source solution. Attendees will gain a comprehensive understanding of how to effectively harness open - source tools while adhering to legal frameworks, fostering innovation and maximizing the potential of open- source | source | 11 | |
| 136 | Our Journey in Integrating External Control Arms (ECAs) and Real-World Data (RWD) for Rare Disease Trials | Cyril Sombrin, Gautham Selvaraj |
External Control Arms (ECAs) are increasingly applied in rare disease trials where Randomized Controlled Trials (RCTs) may be impractical. Recent guidance from the FDA, CDISC, and PHUSE highlights the importance of data reliability, bias reduction, adherence to standards (SDTM, ADaM), and the regulatory use of Real -World Data (RWD). This poster summarizes best practices and presents two case studies: constructing an ECA from natural history studies and legacy RCT data and leveraging publicly available RWD and RCT sources such as the Critical Path for Alzheimer's Disease (CPAD) database. We address common challenges in data integration, standardization, and statistical approaches such as propensity score matching, alongside practical strategies for regulatory communication and dataset conformance | missing value, clinical trial, phase iii, sensitivity analysis, adam dataset, data cleaning, cdisc | source poster source |
4 |
| 137 | Overcoming Challenges in Collaborative Spreadsheet Editing with Shiny, SpreadJS and JSON-Patch | Kamil Foltynski, Sebastià Barceló |
excel, audit trail, clinical programming | source presentation source |
5 | |
| 138 | Oversight of Vendor Deliverables: A Layered Approach for Statistical Programming | Pramod Chittam, Camille Raets |
Effective oversight of vendor deliverables in clinical trials is critical for every sponsor to ensure compliance with regulatory standards, maintain data integrity, and quality. We developed a structured and layered approach for reviewing statistical programming outputs, designed to minimize risk, enhance quality, and optimize resource utilization. The framework comprises three levels of oversight-initial automated verification, semi-automated detailed review, and optional focused review for high-risk studies. By integrating automation and continuous improvement principles, this approach supports early error detection, systematic verification, and adaptability across diverse study contexts | clinical trial, clinical programming | source poster source |
3 |
| 139 | Pharmacokinetics Is Simple... Sometimes | Vincent Buchheit | sample, statistical programmer, pk parameter, clinical trial, compartmental analysis, key, population pk, missing value, data science, data cleaning | source presentation source |
12 | |
| 140 | PHUSE Working Groups Highlights | Chris Price | Poster | clinical research, clinical trial | source | 1 |
| 141 | Plunging into Pools: Dive into a Comparative Analysis of Pooled TLFs with R | Sophie Khan | As we delve into the realm of innovative programming methodologies beyond SAS, we venture into the ever-evolving landscape of R programming. R emerges as a versatile tool, offering enhanced efficiency in data analysis and visualization, and opening new avenues of exploration and discovery in the pharmaceutical industry. This paper presents a comparative analysis of pooled TLFs using R, focusing on a pooling report prepared for regulatory submission that integrates findings from multiple studies across many indications. The methodology involves importing RTF documents to construct dataframes, extracting related R code from an EXCEL document, and utilizing the 'compare_df' package for dataframe comparisons. Results include a comprehensive suite of HTML outputs for each pooled TLF, facilitating informed decision-making. The pooled TLFs were segmented into distinct components, with data comparisons revealing identical matches across individual study TLFs |
excel, ods html, pharmaceutical industry, sample | source poster source |
9 |
| 142 | Pooling Studies Through Code Traceability | Hamdi Chouikha, Kornelius Neufeld |
Study pooling is a critical component of regulatory submissions, specifically for Integrated Summaries of Safety and Efficacy (ISS/ISE). Yet the process remains largely manual, error-prone, and constrained by non-standard study structures, legacy formats, and inconsistencies in derivations and value-level encodings. At Bayer we are reimagining study pooling through a traceability-driven approach. Enabled by the Verisian platform we are taking a blue ocean approach by shifting away from traditional programming towards code traceability-driven transparency. Here we demonstrate how we map variables, compare derivations, and harmonize value-level data, all independent of CDISC compliance. Using graph and AI technologies we can identify and explain differences across studies, understand underlying logic, and resolve discrepancies based on a chosen reference study | regulatory submission, clinical trial, cdisc | source presentation source |
8 |
| 143 | Poseidon: Streamlined Regulatory Screening of HA Websites with AI/ML to Extract Updates and Reduce Manual Efforts | Maria Giovanna Zilli, Fabian Wahren |
Regulatory compliance in the pharmaceutical industry requires continuous monitoring of health authority updates, with subject matter experts at Merck KGaA currently spending considerable time manually screening international regulatory websites for news, guidelines and regulations updates. This manual approach consumes significant resources and risks missing critical updates, as the increasing rate of publications makes it challenging to confidently identify relevant information that could impact the business. We are developing Project Poseidon to address these challenges through an artificial intelligence /machine learning system that aut omates regulatory surveillance via two core phases: automated crawling for content screening, and intelligent triage for scoring, prioritization and stakeholder assessment | machine learning, artificial intelligence, pharmaceutical industry | source presentation source |
8 |
| 144 | Poster PP11 | Pragmatism Meets Precision, Andrea Parodi |
Pragmatic trials (PTs) provide an invaluable means of evaluating real-world interventions, but their inherent complexity demands careful statistical planning. This poster explores key statistical considerations in PTs, offering insights into best practices for study design, data analysis, and interpretation of results | randomization, missing value | source poster source |
4 |
| 145 | Powering Up Programming Milestones: From Chaos to Cohesion | Maciej Wysokiński, Hitesh Thacker |
Join us to explore a transformative approach to simplify management of programming milestones in clinical trials. We address the need for precise programming milestones data, which can counter inefficiencies from fragmented, manual processes as these often lead to poor resource utilization. Our team drives this change through standardization, automation and streamlined communication by employing latest tools that are readily available. Our demo showcases how we use SharePoint, Power Automate and Power BI to collect, store and track programming milestones data with automation. We illustrate how operational dashboards support leadership reviews and decision -making, alongside integrating th ese tools with other databases for enhanced functionality. We'll also share insights into the new skills our team developed throughout this process. Our approach can be used as a template for future innovations and expanded upon to meet the needs of other departments beyond programming. 2 | resource allocation, data cleaning | source presentation source |
8 |
| 146 | Qualitative ADaM Datasets and Define-XML Without Using ADaM Specs | Pieter Coppens | Both the structure and derivation rules for Analysis Data Model (ADaM) datasets are typically defined in ADaM specifications before statistical programming begins. However, at such an early stage, key documents are often still under development and clinical data are still incomplete. Moreover, the data are not fully cleaned yet. As a result, assumptions must be made which introduces the risk of incomplete or inaccurate ADaM specifications, in turn leading to mistakes in ADaM datasets. In this paper we present an alternative approach that involves delivering a testrun analysis without the need for predefined ADaM specifications. This method uses actual study data and is guided by a central ADaM model. Statistical programmers develop the datasets based on a stable version of the statistical analysis plan (SAP), while an independent validator performs derivations in parallel to ensure unbiased results. Any discrepancies are resolved through discussion and refinement | adam dataset, final analysis, data cleaning, controlled terminology, database lock, data science, cdisc | source presentation source |
4 |
| 147 | Quality as an Innovation Catalyst: Transforming Tech-QA Dynamics | Megan Lafferty | quality assurance, liquidity gap | source presentation source |
3 | |
| 148 | R Adoption and Change Management A Large CRO Perspective | Martin Brown, Lynn Clipstone |
Organizations are increasingly leveraging open-source tools to enhance their statistical analysis, reporting, competitiveness and agility. For CROs, this transition can be challenging, as much of the current work is traditionally conducted in SAS, supported by well-established processes. This presentation will address the complexities and resource constraints associated with such a transition. Attendees will gain valuable insights into best practices for implementing change management strategies, enriched by a detailed case study of how change management is being implemented in a large CRO Biostatistics and Programming department. This case study will demonstrate how R and other open- source tools can drive efficiency and foster innovation right now while moving as fast as possible towards a future state | data science, proc iml | source presentation source |
8 |
| 149 | R-Burst: The User Support Approach Shaping GSK's Open-Source Strategy | Andy Nicholls | GSK Biostatistics has an objective to ensure 50% of our code is written in R by the end of 2025. After six years of developing open-source training and tooling, the Data Science Innovation and Engineering team shifted our strategy to focus entirely on our end users through the "R-Burst" initiative. The initial three-month burst of user hyper-care addressed immediate challenges in adopting open-source today. We supported Programmers, Statisticians, and Data Scientists by saying "yes" to all support requests. Our only requirement was that we work together on solutions, and share the resulting insights through our daily blog for the benefit of the entire Biostatistics community. This user-focused strategy has helped us overcome adoption hurdles and informed our open-source innovation beyond 2025 | data science, likelihood function | source presentation source |
5 |
| 150 | Re-Inventing the Deliverable Tracker - from Excel to Dashboard | Megan Harries | Deliverable Trackers are a critical tool in biometrics workflows and are used for everything from project planning, resource management, through to QC and validation tracking, and as a quality audit record. The standard approach is to use Microsoft EXCEL- this offers a well understood user interface, however it relies on manual entry that is often out of date, and data is duplicated in multiple places with limited ability to link and synchronise data from other systems. This presentation demonstrates a new approach to implementing a deliverable tracker that uses an interactive user- interface backed with a database populated by harvesting metrics directly from the SCE and from other systems via API calls |
excel, audit trail, error message, clinical programming, data cleaning | source presentation source |
8 |
| 151 | RECISTance Is Useless, A Novel Visualisation Of RECIST Data Using Bubbles, On A Time-line | Daniel Lindenbaum | macro call, baseline | source presentation source |
10 | |
| 152 | Redefining the role of the Statistical Programmer in an AI-Driven World | James Shipley-Burgess | As AI transforms the life sciences, the role of the statistical programmer is evolving. Traditional coding tasks are becoming automated, prompting the question: what's next for programmers? This poster explores how a reimagined drug development model places programmers at the centre of innovation. With rising clinical trial costs and increasing pressure to streamline processes, programmers are uniquely positioned to make a strategic impact. Beyond coding, they are becoming cross-functional collaborators by guiding AI outputs, optimising data workflows, and uncovering insights hidden in trial data. Rather than being replaced, programmers are emerging as essential drivers in a human-in-the-loop model. This shift highlights a new era where programmers are not just technical contributors, but key enablers of smarter, more efficient development | statistical programmer, clinical research, clinical trial, data transformation, clinical programming | source poster source |
4 |
| 153 | Reproducibility, Lineage and Control: Building a Traceability Framework for the Modern Clinical Trial | Bill Qubeck, Shilpa Sood |
clinical trial, audit trail, 21 cfr part 11, clinical data, adam dataset, data integrity, access control, regulatory submission, raw data, data cleaning, cdisc | source poster source |
6 | |
| 154 | ResponDEX - Response Data Exploration Dashboard | Sevak Kamalyan | This paper introduces ResponDEX, an interactive R Shiny-based dashboard designed to facilitate comprehensive analysis of solid tumor response using RECIST criteria and industry-accepted clinical endpoints. Built on SDTM datasets, ResponDEX provides seamless access to essential metrics such as Best Overall Response (BOR), Objective Response Rate (ORR), Disease Control Rate (DCR), Progression-Free Survival (PFS), Overall Survival (OS), Duration of Response (DoR) and many others. The dashboard features a rich selection of interactive visualizations, including Kaplan-Meier survival curves, Tumor Diameter Change Waterfall Plots, Swimmer Plots, and Forest Plots for Hazard Ratios. Users can explore tumor progression over time with an innovative human body map visualization, dynamically tracking lesion locations and changes. Additionally, ResponDEX offers a comprehensive set of summary tables and listings, providing structured insights into patient response and study outcomes | change from baseline, disease control rate, cdisc | source | 9 |
| 155 | Rethinking Automation in the AI Era | James Zee, Liz Cole |
While artificial intelligence continues to shape conversations around clinical data transformation, many teams risk overlooking the immediate opportunity of expanded use of automation intelligence across the end-to-end data lifecycle. The integration of process automation, rule-based logic, and domain expertise enhances decision-making and improves consistency- complementing, not replacing human input. This session explores how thoughtfully embedded automation intelligence - used alongside targeted AI- realizes tangible value across use cases from statistical programming to pharmacovigilance. Drawing on practical applications from a CRO environment, we'll look at how automation has been used to streamline data flows and reduce manual rework, ultimately contributing to more efficient delivery. The aim is to share not just outcomes but also pose questions and spark discussion as we look ahead to a future of integrated AI and intelligent automation | clinical research, artificial intelligence, clinical data, data cleaning, excel, clinical trial | source presentation source |
4 |
| 156 | Revolutionising Clinical Trial Monitoring and Data Analysis with AR- and VR-Driven Immersive Analytics | Tej Narayan Yadav, Saurabh Das |
Clinical trial monitoring is critical for study quality, safety, and efficiency. Current trial monitoring practices rely on expensive onsite visits or complex remote monitoring of massive datasets using analytics platforms with 2D dashboards and visualizations. These static visualizations limit comprehending complex clinical trial data, recognizing patterns, understanding context and interactivity. Augmented reality (AR) and Virtual Reality (VR) present new opportunities to transform and augment clinical trial monitoring processes through immersive 3D data visualization and analytics. These technologies have already shown value in medical training, rehabilitation, and mental health treatment. To address this, the paper discusses an innovative AR/VR solution enabling clinical trial monitors to visually interact with multidimensional study data in 3D objects such as charts, graphs, 3D models by using real -time tabletop experience | clinical trial, data visualization, clinical data, protocol deviations, sample, adverse events | source presentation source |
7 |
| 157 | risk.assessr: extracting OOP function details | Edward Gillian, Hugo Bottois, Paulin Charliquart |
Extracting details from Object Oriented Programming (OOP) functions in R has proved challenging as there are different types of these functions (S3, S4, R6, special) which have different methods for encoding and diverging standards for writing them. Firstly, to extract details, these functions need to be identified in the NAMESPACE and S3 functions have different NAMESPACE constructs to other OOP functions. These functions can be constructed with different and unique classes that must be checked. These functions can be documented in diverging ways using usage descriptors such as ""defunct|deprecated|deprec|superseded|imported|re-exported|reexports". Secondly, these different function types require considerably different methods to extract the details ranging from functions called from the methods package and base R to custom functions to create extraction inputs. These diverging encoding methods and standards had implications for creating code that can be executed in various processes | data cleaning, sample | source presentation source |
10 |
| 158 | Roadmap to an ODAC: The Journey to and Through an FDA Oncologic Drug Advisory Committee as a Study Lead Programmer | Charles Wright | An FDA ODAC can be a daunting proposition for any programming team, especially when they are unexpected and at short notice. As a study lead programmer who is responsible for the day-to-day programming activities on a study, how is it possible to manage the programming of an enormous influx of additional adhoc output requests to support an ODAC, whilst continuing to make sure everything else at a study level runs smoothly? In this poster, I will walk through the journey of planning and preparing a programming team to execute and fulfill the requirements needed prior to an FDA ODAC, what efficiencies we built into the process, as well as what lessons we learned. Additionally, I will share my experience of what is required of a programmer on the day of the ODAC itself and being prepared to provide programmatic outputs at the drop of the hat | interim analysis, quality control | source poster source |
2 |
| 159 | Robust Programming Strategies for Exposure to ADaM Datasets | Susann Grill | Building robust ADaM datasets for drug exposure is crucial for accurate safety and efficacy analyses in clinical trials. Dependingonstudydesign,dosingschedules,thepotentialfortreatmentchangesandinterruptions,aswell as the need to combine multiple sources of data, deriving exposure variables can become increasingly complex anderror-prone. Thispaperexploresanticipatorystrategiestoidentifydatainconsistenciesoranomalies,suchasmissing,overlap- ping, or contradictory records. It focuses on minimizing errors through early assessment of input data quality, the applicationof logicchecks to safeguard complexderivations, and cross-checkswith other ADaM datasets. These defensive approaches ensure data quality, enhance program stability, and promote efficient workflows in clinicaltrials by proactivelyaddressing potentialissues that may ariseacross a variety ofstudyscenarios | adam dataset, in=, treatment period, sdtm domain, cdisc | source poster source |
13 |
| 160 | rock and roll - Deliverables driven by standardized data specifications | Svetlana Freiberga, Mounika Suddapalli |
Creating data deliverables from scratch is repetitive, error-prone, and difficult to scale. To support data programmers in this process, while enforcing adherence to standards and maintaining high quality, Pharmetheus has developed rock, an R Shiny application. rock streamlines everything from drafting data specifications to generating ready-to-use programming environments. It helps standardize workflows and ensures reproducibility and full traceability while retaining flexibility. rock brings programmers and modelers together in a shared, version-controlled framework that adapts to evolving project requirements. It's not just about automation; it's about building smart tools people actually want to use. rock is a practical solution designed around day-to-day challenges, making workflows more efficient, transparent, and scalable | raw data, internal audit | source poster source |
8 |
| 161 | SAIL.R: Navigating Change Together in Clinical Analytics! | Annemie Van Dijck | SAIL.R is a cross-functional initiative at Johnson & Johnson dedicated to implementing R for the development of both static and interactive analysis output in clinical data exploration and reporting. The initiative revolves around key streams such as regulatory submissions, the R reporting framework, interactive visualizations, and effective change management. This paper emphasizes the importance of people-centric approaches during this transition for statistical programmers and statisticians from SAS to R. Engaging employees throughout the change process allowed for valuable feedback and promoted a sense of ownership. One fundamental principle was to show the tangible benefits of the change to the team. We prioritized providing robust support through training, access to subject matter experts, and a safe environment for questions. We will also cover strategies for stakeholder management, effective communication, overcoming resistance, and evaluating success metrics. Overall, the SAIL | regulatory submission, sas programmer, likelihood function, clinical trial, data science | source presentation source |
11 |
| 162 | SAMBA: Bridging the Gap How GenAI Transforms Analysis Plans into ADaM Specs | Thiago Figueiredo, Richard Becker |
Presentation | data science, phase i, phase ii, clinical trial, regulatory submission, statistical programmer | source | 19 |
| 163 | SAS Code Hidden in Plain Sight | Bartosz Jabłoński | Sometimes there is a need to share a SAS ○R macro in a way that the execution of the macro is possible, but at the same time the source code is not "readable" (e.g., the code is a proprietary solution). In this article, a programming-based solution utilizing encrypted user-written functions (FCMP) and SAS data sets, which gives requested effect, will be explained. Furthermore, the solu- tion (in contrary to operating-system-dependent SAS catalogs) allows for code exchange between different operating systems! As a cherry on top, the GSM (Generate Secure Macros) package, which allows users to create secured macros stored in SAS PROC FCMP functions and share them between different operating systems without showing their code, will be presented. Table of contents |
data step, macro definition, cpu time, macro put, macro variable, proc step, proc fcmp, proc sql, catalog, macro loop, sas packages | source presentation source |
28 |
| 164 | SAS Viya and Posit Team Integration: Git-Enabled, AI-Driven Clinical Reporting with Azure and Open-Source Technologies | Maksym Prykolota, Vaibhav Kulkarni |
This paper introduces a Git-enabled integration of SAS Viya and Posit Team, anchored by a Central Data Repository (CDR) built on Azure Cloud technologies. Designed for regulatory- compliant, high-volume clinical reporting, this modern workflow supports cross-functional collaboration, automation, and audit readiness. It delivers a pragmatic approach that balances real-world constraints with innovation, enhancing productivity while ensuring compliance with GxP principles - covering traceability, access control, change management, and data integrity. The solution is platform-agnostic, offering seamless extension from SAS Viya to R and Python through open-source integration. It provides practical insights into how AI-enabled development, open-source tooling, and cloud-native infrastructure can be applied to optimize clinical data workflows | sas viya, sas visual analytics, statistical programmer, data cleaning | source presentation source |
9 |
| 165 | Shots to Stats: AI's Take on Alcohol Units | Liam Clothier | The Timeline Follow Back (TLFB) method is an approach to quantifying alcohol intake. The data collected with the tool can be difficult to convert into analysable data, using up trial resources. The development of AI offers an opportunity to speed up data cleaning in this context. However, the accuracy of AI tools to convert, for example "two glasses of wine", into number of units is unknown. This study uses data from the AlcoChange clinical trial to test three different open source AI software and look at their consistency and accuracy compared to traditional human review. We report an updated analysis that also demonstrates development of AI over the last 12 months | data cleaning, clinical data management, clinical data, clinical trial, data governance, excel | source poster source |
6 |
| 166 | Showcasing a Shiny Tool for Evaluating Biomarkers for Derisking Clinical Trials | Stefan Thoma | The development of new therapies is frequently hampered by clinical trials that are long, costly, and require large patient populations, particularly for chronic conditions with long-term endpoints like multiple sclerosis. This paper addresses the challenge of minimal savings in time and patient numbers when using only traditional endpoints, such as composite confirmed disability progression (cCDP), for interim and final readout. We propose the strategic use of an intermediate endpoint-an earlier marker that is reflective of clinical benefit-to provide confidence and inform the critical go/no-go decision to proceed to Phase III, or stop the Phase II early. To support this strategy, we are developing a unified, quantitative framework for the systematic evaluation of candidate intermediate endpoints in the form of a shiny application. This approach ensures that intermediate endpoints developed by different groups are fairly compared using a unified methodology | treatment effect, clinical endpoint, interim analysis, correlation, conditional power, phase ii, phase iii, futility analysis, effect size, clinical research, randomized controlled trial, confidence interval, clinical trial | source presentation source |
9 |
| 167 | SI Add More MSG to Your Annotations | Davy Baele, Nicolas Deravet |
Presentation | medical history | source | 19 |
| 168 | Simplify, Automate, Submit: An In-House Framework for Seamless Metadata Flow from (Pre)Production to Submission | Malathi Hari, Thomas Gulholm-Hansen |
Presentation | source | 13 | |
| 169 | Small Application for Big Change: Data Integrity and Compliance Through Automated Decision Tracking with Shiny | Olena Olishevska, Mika Mäkinen |
Ensuring data integrity and compliance in clinical and nonclinical studies is challenging due to the diverse systems and processes employed by companies. To address this, we introduce a prototype of web application designed to track derivation and formatting decisions, thereby enhancing transparency and accountability in data flow. Proposed solution accounts for harmonizing input data, logging of user actions and versioning of standardization documents, providing a full traceability of data flow. Using real-world example, we show how our implementation of the proposed design using Shiny (R framework) creates a decision database in accordance with CDISC Controlled Terminology, which is later used for automated production of SEND-compliant domains. Early user evaluation indicates major reduction in data processing time and improved transparency of processing steps for further automation | raw data, dataset, data wrangling, data transformation, clinical trial, data pipeline, cdisc | source poster source |
9 |
| 170 | Smarter SAS logging-with a macro AI hasn't mastered (yet) | Jules Van der Zalm | Program logs are vital for debugging code in any language. While SAS generates solid log files, they're not always great for pinpointing the exact location of problematic code-especially when it's inside a macro. To facilitate easier code debugging, it's good practice for SAS macros to announce themselves in the log. By printing their name, it becomes much simpler to trace the source of an error message. All you have to do is to scroll upwards in the log and find the latest macro announcement. These announcements often include the macro name as well as parameter names and values. These announcements are embedded in the program code. Maintaining these announcements is tedious, error-prone and outright distracting. They're usually a series of %PUT statement with the macro name and parameters hardcoded into them. They're often copied from program to program, leading to inconsistencies between what's logged and what the macro is actually doing |
macro variable, macro parameter, macro put, macro definition, put statement, sas log, macro call, sample, excel, macro loop | source presentation source |
11 |
| 171 | Snakemake for Statistical Programmers: Automation and Reproducibility in Clinical Studies | Nassim Sleiman, Mathieu Cayssol |
Snakemake is a workflow management system that automates complex data pipelines by specifying rules and dependencies. It ensures reproducibility, optimizes runtime through parallel execution, and minimizes manual intervention. At Roche, we have developed a standardized Snakemake pipeline (integrating essential dependencies) to automate the creation of SDTMv, ADaMs, and TLG outputs using SAS and R. Designed specifically for statistical programmers, this approach streamlines the process of generating clinical datasets and analytical reports, improving transparency and consistency. By leveraging Snakemake's capabilities, statistical programmers can reduce processing time and maintain robust, traceable workflows. This poster will highlight our pipeline structure and the transformative impact of adopting Snakemake for clinical data reporting tasks | statistical programmer, adam dataset, clinical data, clinical trial, cdisc | source poster source |
11 |
| 172 | Statistical Endpoints and Analysis in Diagnostic Imaging: The Matrix Has You and Your Data | Srinivas Lanka, James Sykes |
Diagnostic imaging studies are essential for assessing the accuracy, reliability, and diagnostic performance of new imaging technologies, particularly in oncology. As the demand for improved imaging grows, the ability to accurately detect lesions and support early clinical decisions becomes increasingly important. Effective imaging can lead to faster, more appropriate treatment pathways. These studies often rely on endpoints derived from a standard 2 x 2 matrix-most notably, sensitivity and specificity, which are key co-primary endpoints in Phase 3 trials. A well-defined standard of truth is also critical for evaluating diagnostic accuracy. But how do we efficiently collect and summarize these endpoints? This paper explores how to map diagnostic imaging data to the SDTM structure, integrate it into ADaM efficacy datasets, and derive relevant performance statistics | primary endpoint, key performance indicator, confidence interval, missing value, cohen's kappa, type ii error, type i error, negative predictive value, precision, primary efficacy analysis, clinical trial, efficacy summary, data science, proc step, hypothesis testing, proc sql, adam dataset, analysis set, proc freq, cdisc | source presentation source |
17 |
| 173 | Status and Updates of OpenStudyBuilder | Mikkel Traun | Poster | cdisc | source | 1 |
| 174 | Succeed with gridify: Enriching R Visualisations for Regulatory Submissions | Alexandra Wall, Jennifer Winick-Ng |
pharmaceutical industry, clinical trial | source presentation source |
10 | |
| 175 | Survival analysis for AdVerse events with VarYing follow-up times (SAVVY)-estimation of adverse event risks | Regina Stegherr, Claudia Schmoor, Jan Beyersmann, Kaspar Rufibach, Valentine Jehl, Andreas Brückner, Lewin Eisele, Thomas Künzel, Katrin Kupas, Frank Langer, Friedhelm Leverkus, Anja Loos, Christiane Norenberg, Florian Voss, Tim Friede |
Background: The SAVVY project aims to improve the analyses of adverse events (AEs), whether prespecified or emerging, in clinical trials through the use of survival techniques appropriately dealing with varying follow-up times and competing events (CEs). Although statistical methodologies have advanced, in AE analyses, often the incidence proportion, the incidence density, or a non-parametric Kaplan-Meier estimator are used, which ignore either censoring or CEs. In an empirical study including randomized clinical trials from several sponsor organizations, these potential sources of bias are investigated. The main purpose is to compare the estimators that are typically used to quantify AE risk within trial arms to the non-parametric Aalen-Johansen estimator as the gold-standard for estimating cumulative AE probabilities. A follow-up paper will consider consequences when comparing safety between treatment groups. Methods: Estimators are compared with descriptive statistics, graphical displays, and a more formal assessment using a random effects meta-analysis. The influence of different factors on the size of deviations from the gold-standard is investigated in a meta-regression. Comparisons are conducted at the maximum follow-up time and at earlier evaluation times. CEs definition does not only include death before AE but also end of follow-up for AEs due to events related to the disease course or safety of the treatment. Results: Ten sponsor organizations provided 17 clinical trials including 186 types of investigated AEs. The one minus Kaplan-Meier estimator was on average about 1.2-fold larger than the Aalen-Johansen estimator and the probability transform of the incidence density ignoring CEs was even 2-fold larger. The average bias using the incidence proportion was less than 5%. Assuming constant hazards using incidence densities was hardly an issue provided that CEs were accounted for. The meta-regression showed that the bias depended mainly on the amount of censoring and on the amount of CEs. Conclusions: The choice of the estimator of the cumulative AE probability and the definition of CEs are crucial. We recommend using the Aalen-Johansen estimator with an appropriate definition of CEs whenever the risk for AEs is to be quantified and to change the guidelines accordingly. | regression analysis, kaplan-meier estimator, proc freq, sample, survival analysis, clinical trial, confidence interval, adverse events, data science | source poster source |
13 |
| 176 | Swarming Success: Optimising Efficiency in Statistical Programming | Lea Zittlau, Saskia Ständler |
Documentation of tasks and study-related information for statistical programmers is often scattered across multiple locations, such as EXCEL trackers, OneNote pages, Word documents, or Teams channels. This has proven to be a time-consuming process that lea ds to information being overlooked as well as difficulties in communicating with team members. In today's fast -paced world, it is increasingly important to organize work efficiently - especially when working on numerous projects at the same time. Agile methodologies, which have been common practice in software development for more than 20 years, have been shown to address this issue. In this paper, we demonstrate how statistical programmers can integrate agile tools into their daily work and collaborate wit h each other to coordinate programming activities. We also address pertinent questions related to motivating and training of project members to use agile tools |
statistical programmer, data cleaning, clinical programming, excel | source presentation source |
10 |
| 177 | Synthetic Data - the Game Changer in Drug Development | Soundarya Palanisamy, Anette Dalsgaard Jakobsen |
Drug development relies on a limited volume of relevant clinical data which is restricted in use, costly to obtain and often insufficient and underrepresented. Augmenting real -world data with synthetic data accelerates drug development through lower costs and faster execution. Synthetic data surfaces valuable trends latent in patient data while avoiding compromise of privacy. This session provides an overview of considerations for robust and governable synthetic data generation, using a new SAS offering, Data Maker as an example. It provides an overview of available methods, such as Generative Adversarial Networks (GANs), SMOTE, and other emergent methods. We also cover quality assessment and evaluation of synthetic data prior to application. We also highlight potential risks from synthetic data and suggest how to increase acceptance in drug development | clinical trial, clinical research, clinical data, sample, oversampling, proc step, data cleaning | source presentation source |
7 |
| 178 | Talk to Your Data: Accelerating Insights with AI-Powered Dashboards | Emily Yates | adverse events, data science, feature selection | source presentation source |
8 | |
| 179 | The Art of Timely Deliverables: How We Learned To Stop Worrying and Love the Deadline | Søren Fjelstrup | In the dynamic environment of Health Technology Assessment (HTA), timely data delivery is critical to meeting the demands of governmental bodies. This paper outlines our approach to managing HTA data requests efficiently. We will explore our centralized document system that tracks incoming requests, enabling effective prioritization, and by leveraging a sprint board for task management, we assign tasks rapidly. To further streamline the process, we use an LLM agent to translate incoming requests into a CDISC friendly format. This ensures that standard requests can be processed through a series of independent, modular programs, while custom programs are developed only for unique requirements. This bifurcated task management method reduces turnaround times and enhances productivity and allows for parallel workflows | excel, quality assurance, resource allocation, survival analysis, audit trail, data science, regression analysis, cdisc | source presentation source |
12 |
| 180 | The Chicken, the Egg and the ADaM: Cracking the PRE-ADSL vs ADSL Sequence | Carla Santillán, Dubravka Georgijevic |
ADaMIG v1.2 introduced the concept of Pre-ADSL, an ADaM OTHER dataset intended for use when certain ADSL variables are better derived in other ADaM datasets. Our collaborator adopted this concept and formalized it as ADCORE in their ADaM standard. We share our experience from a study where ADCORE was the first created ADaM domain and included all standard ADSL variables except those based on efficacy endpoints, which were derived using downstream efficacy ADaMs. We reflect on past studies where ADCORE was not used but, in hindsight, we feel it would have improved workflow and traceability, and we present a case where its use was limited - specifically, when ADSL analysis set flags needed early availability, but derivations depended on complex relative dose intensity calculations. This presentation aims to provide a balanced perspective on integrating ADCORE into ADaM workflows and when its use may - or may not - be beneficial | adam dataset, dose intensity, imputation, sdtm domain, data science, cdisc | source presentation source |
8 |
| 181 | The Human Element: Synergising Emotional and Physical Intelligence in Statistical Programming Leadership | Harris Kampouris, Lino Jardim |
In statistical programming teams, where technical expertise is paramount, emotional quotient (EQ) and physical quotient (PQ) emerge as crucial leadership tools. Leaders with high EQ EXCEL in fostering collaboration, managing conflicts, and creating supportive environments. By understanding emotions and empathizing with team members, they build trust and psychological safety. PQ complements EQ by focusing on physical well -being, enhancing cognitive function, and stress management. The synergy of EQ and PQ in leadership enhances decision -making, balancing logical rigor with emotional awareness and physical resilience. As statistical programming evolves in fast -paced AI- driven environments, integrating EQ and PQ ensures sustainable success. This holistic strategy harmonizes technical excellence with human -centred leadership, optimizing team performance, fostering innovation, and driving efficient code development |
clinical programming, key performance indicator, regulatory submission, data science | source presentation source |
12 |
| 182 | The Key Features Every Statistical Computing Environment Should Have and Often Lacks. A Presentation About Almost 20 Years of Experiences with SCEs. | Michael Weiss, Elena Glathe |
data cleaning, dataset | source presentation source |
5 | |
| 183 | The Kids Are All Right: How Investing in Career-Young Talent Can Create Engaged Future Leaders | Sean Miles | This presentation shares a personal journey through a Graduate Training Programme (GTP) and subsequent career growth to highlight the long-term value of investing in early career talent. It discusses the importance of solid foundations and supported, accessible learning in forming well-rounded and confident programmers & statisticians in the world of clinical data science. While recognising some of the challenges in implementation, this presentation will offer practical insights into how early engagement and meaningful investment in individuals' development can build not only technical capability, but also the confidence and commitment essential for growing future leaders in clinical data science. From student to master: exploring the route taken, essential skills gained and milestones along my journey from inexperienced programmer to ownership and coordination of a Graduate Training Programme | clinical data, data standard | source presentation source |
3 |
| 184 | The Power of Early Careers: Investing in the Next Generation | Miles Almond, Jun Se Kim |
In today's competitive landscape, attracting and retaining top talent to bring valuable business impact is no easy feat, but we strongly believe long term investment in early careers is key to achieving this. This presentation explores the transformative impact of placement students and interns within our Data Sciences Pipeline. Through interviews with former students, now senior programmers and leaders, we'll explore how early exposure to industry has been instrumental in their career development. We'll also spotlight several impactful business projects led by students, showcasing measurable benefits such as process improvements, automation tools, and fresh perspectives that brought lasting value to the department. We hope to make a compelling case for why companies, especially those with limited early careers initiatives, should invest in these programmes | pharmaceutical industry, data science | source | 13 |
| 185 | The Power of Embracing the Unknown for Personal Growth | Gitte Konradsen | In an ever-changing work environment, leaders face the challenge of navigating uncertainty while fostering personal growth within their teams. This presentation dives into the concept of embracing the unknown as a catalyst for resilience and innovation. By stepping outside comfort zones, individuals can develop new skills, gain diverse perspectives, and enhance their adaptability. Practical strategies from recent and ongoing changes within Biometrics at Novo Nordisk such as agile ways of working, use of R, and full-service outsourcing will be presented. Focus will be on how to cultivate a positive mindset, support teams in facing uncertainties, and transform challenges into opportunities for development while acknowledging the individual level of curiosity as well as resistance to change. Through real-life success stories, the presentation aims to inspire and empower leaders to embrace change as a pathway to growth, ultimately benefiting both individuals and organizations | source presentation source |
2 | |
| 186 | The Show Must Go On: Best Practices for Submitting SDTM Data for Ongoing Studies | Kristin Kelly | Though the CDISC SDTM Implementation Guide provides advice on how to prepare SDTM datasets for completed studies, there is little guidance on what to do when the study is ongoing, leading to varied implementation practices across the industry. At times, it may be difficult for a regulatory reviewer to readily determine that a study is still in progress without looking in the Clinical Study Data Reviewer's Guide (cSDRG). The recent addition of the ONGOSIND (Ongoing Study Indicator) parameter in the FDA Study Data Technical Conformance Guide (sdTCG) allows sponsors to clearly specify within the data whether a study is ongoing. In this paper, some considerations for preparing domains such as Demographics (DM), Disposition (DS), and Trial Summary (TS) for an ongoing study as well as strategies to ensure data transparency across the SDTM submission package will be discussed | data cleaning, trial summary, clinical trial, sdtm domain, cdisc | source presentation source |
8 |
| 187 | The Utilization of an Interface Based on a Graph Database | Catharina Dahlbo, Eva Kelty |
A graph database-powered interface provides dynamic, connected views of clinical trial data, tailored to specific roles and trial stages. It gives users direct access to a rich data environment, removing the need for constant report or graph requests. By enabling independent data exploration, it empowers each role throughout the trial lifecycle. During the planning, design teams explore historical data to identify patterns and optimize study design. During monitoring, site coordinators and data managers track enrolment, deviations, and adverse events. Investigators and safety managers gain contextual insights through interactive timelines and safety signal queries. In the analysis stage, medical writers ensure accurate reporting, streamlined preparation for regulatory submission and final study filing. This platform transforms complex data into actionable insights, revealing hidden relationships and supporting informed, confident decisions | adverse events, clinical trial, clinical data | source poster source |
8 |
| 188 | The Value of Clinical Trial Data: From Submission to Innovation | Anne Katrine Alsing | clinical trial, clinical data, bronze layer, data standard, silver layer, data transformation, data science, gold layer, adverse events, missing value | source presentation source |
8 | |
| 189 | The {teal} Framework Updates: How Open Source and Cross-Industry Collaboration Benefit Everyone! | Peyman Eshghi, Paul Jenkins |
In today's data -driven landscape, effective data visualization is crucial for deriving insights and supporting informed decision-making. This paper presents recent advancements in the {teal} framework, an open -source R package that has gained significant adoption across pharmaceutical organizations for clinical data analysis and visualization. We highlight the importance of data transparency and accessibility, especially in light of increasing regulatory interest, and discuss how the {teal} framework aligns with these expectations. Insights from industry leaders, gathered through forums like the PHUSE Working Group, provide context to our discussion. We present examples of applications of the {teal} framework, showcasing how Johnson & Johnson utilizes the framework and its new features to create interactive visualizations. Furthermore, we touch o n the {teal} roadmap and future feature prioritization to illustrate ongoing developments | access control, data visualization, bar chart, clinical data | source presentation source |
12 |
| 190 | Think like a Quarterback - Deliver Projects, not fumbles | Eoin Byrne | Leading a clinical trial demands the strategic mindset and flexible thinking of a top quarterback. Just as a quarterback reads defences, adapts to shifting plays, and leads a team under pressure, leading a clinical trial requires us to interpret complex data, adjust to regulatory and logistical changes, and guide cross-functional teams toward a common goal. Both roles require decisive decision -making, anticipation of obstacles, and effective communication. A quarterback trusts the playbook but knows when t o call an audible. Similarly, when leading a clinical trial, we follow all necessary documentation but must adapt when real -world challenges arise. Success hinges on vision, leadership, and the ability to execute under uncertainty. These engaging parallels highlight the dynamic and proactive thinking require d in clinical research, encouraging professionals to adopt a quarterback's blend of strategy, agility, and collaboration | clinical trial, adam dataset, cdisc | source presentation source |
9 |
| 191 | Towards Managing Macro Libraries in an SCE: A Vision for Reproducibility Through Version Locking | Shilpa Sood, Priyanka Dharuvuri |
Macro libraries are foundational to modern statistical programming. This session explores a forward-looking approach to managing macro libraries within a Statistical Computing Environment (SCE), with the goal of improving access consistency, enhancing traceability, and reducing compliance risk. By envisioning a framework that incorporates modular release practices and version control directly into macro execution, organizations could enable systems where programs automatically capture and reuse the exact macro versions used at runtime-even if newer versions exist in the global library. Though not yet standard practice, this concept presents an opportunity to strengthen reproducibility, streamline deliverables, and align macro management with regulatory traceability expectations through future SCE capabilities | macro definition, clinical programming | source presentation source |
6 |
| 192 | Transforming Study-Specific Reporting (SSR Trends) Through an Innovative Data Pipeline Architecture Approach | Archana Nayak, Jill Ophoff |
In Analytical Risk-Based Monitoring, centralized real-time data reviews are vital for effective trial oversight. At J&J Innovative Medicine, dashboards with standard Key Risk Indicators are supplemented with customized Study Specific Reports tailored to the unique needs of each trial. Generating these reports promptly upon initial subject data receipt remains challenging. To address this, we developed a re-purposeful data pipeline architecture that incorporates study-specific configuration files separately from the main pipeline, balancing ease of set up, flexibility and efficiency. This design employs a standard data structure, enabling near real-time mapping of raw data from external vendors and EDC systems. We are currently testing this approach with a report on pharmacokinetic sampling compliance, utilizing a Biospecimen Events SDTM-like domain to detect potential sample collection errors. Preliminary results from a large Immunology Phase 3 trial are promising | clinical trial, data review, cdisc | source presentation source |
3 |
| 193 | Tried and Tested Principles for Building, Managing and Leading High-Performing Teams Through Challenging Initiatives | Caroline Phares, Oliver Robinson |
net promoter score, clinical research | source presentation source |
5 | |
| 194 | Turbocharge Your Trials: Transform CRF Creation and the CDISC standardization process with Xbiom's Smart Metadata Repository and AI/ML Integrations | Harinarayanan Gopichandran | data cleaning, clinical trial, metadata management, regulatory submission, data standardization, audit trail, clinical data, data governance, edc, crf, clinical data management, access control, adam dataset, data science, data validation rules, cdisc | source presentation source |
8 | |
| 195 | Turning Data into Deliverables: Automating Biometrics from Metadata to TFLs | Inderjeet Arora | The clinical trial industry has long relied on well-defined macros and tools for generating Tables, Listings, and Figures (TLFs). However, the broader biometrics workflow i.e., from transforming raw/source data to SDTM, then to ADaM, and ultimately to TLFs remains complex and involves multiple manual steps. Despite automation tools, significant effort is still required to manage data mappings, transformations, and validations across these stages. A unified platform that automates the end-to-end process i.e., mapping raw data to SDTM, SDTM to ADaM, and ADaM to TLFs - can significantly streamline this workflow. By incorporating AI algorithms, such a system can intelligently manage data transformations and TLF generation, while built-in audit trails and governance features ensure controlled access, transparency, and traceability. leading to reduction in repeated communication, and manual execution | mock shell, raw data, clinical trial, clinical data, data science, data cleaning, adam dataset, data transformation, machine learning, adverse events, audit trail, regulatory submission, data wrangling, sdtm domain, csr, data integrity, controlled terminology, cdisc | source presentation source |
9 |
| 196 | Twice as Much Aint Twice as Good: Is Double Programming Really the Gold Standard We Think it Is? | Vitali Gering | data cleaning, independent qc, clinical trial, quality control, statistical programmer, data science, clinical programming, pharmaceutical industry | source presentation source |
6 | |
| 197 | Unlocking Efficiency Through Digitisation: Plotting the Journey with Atlas for Clinical Trial Output Generation | Daphné Grasselly, Eddy Foster |
Presentation | clinical trial, data science, parallel processing, sas expression | source | 20 |
| 198 | Unlocking Healthcare Insights: Graph-Based Modelling and Analysis of Synthetic Patient Data | George Panagiotakis, Nikolaos Kountouris |
In the pharmaceutical domain, real-world data (RWD) spans a large, intricate web of information drawn from electronic health records, patient registries, insurance claims, and other healthcare sources. Although traditional relational databases are well -established for structured analytics, their rigid tabl e-based schemas often create challenges for handling the complexity and dynamically evolving nature of those data. This paper explores the use of graph theory & databases for modeling synthetic data that approximate real -world patient data, without breachi ng any regulatory policies such as Health Insurance Portability and Accountability Act (HIPAA), and the General Data Protection Regulation (GDPR). Graph-based modeling could act as an alternative tool by explicitly capturing the naturality within relationships to healthcare data, enabling scientists to perform analyses effectively | data science, relational database, normalization, sample, insurance claims, structured data | source presentation source |
18 |
| 199 | Untangling the Dependencies Web | Rhys McCrosson | Clinical submissions often contain hundreds of datasets (split across raw, SDTM, ADaM and TFL) with a complex dependency flow linking datasets both within and between the different levels. Traditionally these links are recorded by programmers in a QC tracker or program header, but this is prone to human error and only provides a fragmented view of the full web of dependencies. In this paper we explore ways of programmatically scanning dataset programs (including any external macros that are called) and specs to build a complete network of dependencies, which can be interrogated to answer a range of questions, including: Do the dependent datasets listed in the spec match the datasets that the production program is reading in? Is the QC program using the same datasets as production? * Given a subset list of TFLs for an additional analysis, what subset of raw/SDTM/ADaM datasets are required? | sdtm domain, adam dataset, macro call, concomitant medications, macro variable, out=, key, sample, cdisc | source presentation source |
12 |
| 200 | USDM Synthetic Data | Generation Stuart Malcolm | clinical trial, quality assurance | source presentation source |
7 | |
| 201 | Using CDISC USDM and BCs to select data from alternative data sources for regulatory purposes | Berber Snoeijer | As the use of alternative data sources (ADS), including real-world data (RWD), grows in regulatory decision-making, standardized methods are needed to assess their suitability. The CDISC Unified Study Definitions Model (USDM) and Biomedical Concepts (BCs) provide a structured approach to identify and evaluate relevant data. This paper describes how USDM and BCs support two regulatory use cases: study feasibility assessments and RWD studies. USDM enables alignment between protocol requirements and available data, facilitating faster, reproducible feasibility evaluations. BCs ensure consistent interpretation of clinical concepts across sources, enhancing the validity of RWD analyses. Together, these tools help streamline data selection, improve transparency in data lineage, and support regulatory-grade evidence generation | clinical trial, data lineage, cdisc | source presentation source |
6 |
| 202 | Using the full power of SAS to resolve Algorithmic MedDRA SMQs | Stephen Prust | hash table, proc sql, sas log, sas function, proc step, adverse events, macro definition | source presentation source |
18 | |
| 203 | Using Visual Dashboards to Improve Data Accuracy | Daniel Smith | Data lies at the heart of any clinical trial, making its accuracy and reliability a primary objective for all stakeholders. Errors can arise in various ways, from typographical mistakes to incomplete data capture. Traditional approaches often rely on manual checks or reviews of derived datasets, which can increase the risk of human error or mask issues after analysis. This paper highlights the power of using simple visualizations based directly on source data to enable earlier and more accurate identification and correction of data errors. Visual dashboards aggregate and simplify large amounts of data, making it easier to spot anomalies and unexpected patterns. By implementing these visualizations, teams can improve inter-team communication, reduce the need for extensive data review, and minimize the risk of costly mistakes. Modern clinical trials have access to an increasing number of data visualization tools, making the adoption of a visual | adverse events, data visualization, clinical trial, data review, sample, phase i, raw data, study day, data cleaning, concomitant medications | source poster source |
8 |
| 204 | Validated Statistical Computing: A Kubernetes Framework for Regulatory-Compliant R, Python and SAS Deployments | T. Randall Betancourt, Tim Fantuzzo |
data cleaning, audit trail, sas grid, 21 cfr part 11, proc sql, resource allocation, data science, batch job, cloud storage, connection string, job scheduling, rest api, sql server, validation documentation, excel, dataset, sample | source presentation source |
53 | |
| 205 | Visualising the Invisible: Patient Journey with Waterfall, Spider and Swimmer Plots | Varsha Mithun Patil, Sunita Poojary |
Patient-level plots are essential in clinical data analysis, providing detailed insights into individual treatment responses across therapeutic areas such as oncology and neurology. These visualizations help researchers and clinicians assess variability in patient outcomes, identify trends, and make informed decisions. Waterfall plots illustrate individual patient responses by displaying changes in tumour size or clinical metrics. Spider plots track longitudinal variations, highlighting shifts over time in response to therapy. Swimmer plots visually map patient trajectories, representing treatment duration, response patterns, and key clinical events. This paper highlights the significance of these plots in evaluating treatment efficacy and disease progression. It also demonstrates their implementation using Python on Databricks platform with synthetic data, offering a practical perspective on their use in clinical trials | swimmer plot, waterfall chart, clinical trial, proc means, procedures, regulatory submission, change from baseline, complete response, standard deviation, treatment arm, treatment effect, outlier, clinical data | source presentation source |
14 |
| 206 | Visualising Trends and Bottlenecks: A Python Web App for QC Progress Monitoring | Spencer Renyard | Tracking the progress of QC efforts across multiple programs can be challenging without clear, actionable visualizations. To address this, we developed a Python web application that accesses QC tracker data via the Monday.com API, using both live data and the activity log to determine the status of each program over time and when changes occurred. The application generates graphical representations of progress, including a weighted percentage based on the number of programs in each status category. By automating the extraction and analysis of QC tracker data, the application reduced manual effort and provided a historical view of program progress. Its visualizations enabled teams to identify trends, track bottlenecks, and make data-driven decisions, improving oversight of QC activities. This paper will include a technical walkthrough of the application's development, covering Python libraries, the Monday.com API, and visualization techniques | excel, line chart | source presentation source |
7 |
| 207 | What a Good One Looks Like Building a Team of Internal R Experts | Mike Smith | Presentation | data science | source | 24 |
| 208 | What We Gained (and Lost) by Using Python in SDTM Conversions | Katarzyna Bzymek | Presentation | clinical research | source | 25 |
| 209 | Which Macros Are Used in the Study? | Guido Wendland | Knowing which macros are used in a program or study can be useful for various reasons: To populate the program header To identify the need for program updates To increase consistency across programs To identify dependencies between macros To identify potential issues or estimate the impact of macro updates Etc. The macro %_mac_usage provides a detailed EXCEL report of all macro invocations. This presentation offers insights into the involved techniques to derive and report this information |
macro call, macro variable, macro definition | source presentation source |
5 |
| 210 | Why I can't see myself coding that much anymore | Angelo Wibbeler | Various presentations/papers have shown that simple standardizations of specification documents can automate many programming tasks, including deriving simple variables. This reduces duplication of information, enhances consistency, and speeds up development. With the advent of Large Language Models (LLMs), it is now possible to use less standardized information from specification files and generate sample code for the remainder of programming tasks that can't be automated yet. This can be used as a starting point of the programming process. This presentation, using the ADaM programming process as an example, will discuss: Which programming tasks can be automated and which code should not be written How to leverage LLMs and some simple scripts to write entire ADAM programs without manual coding * How to stay flexible despite adding standardizations | clinical programming, independent qc, data cleaning, cdisc | source presentation source |
11 |
| 211 | Why Its Important To Build a Global Team Spirit And Some Ideas on How To Do it | Paul Dick | Building a strong team spirit is essential for meeting deadlines because it fosters collaboration, efficiency, and motivation. When team members trust and support each other, they communicate more effectively, reducing misunderstandings and delays. A positive team environment encourages individuals to share ideas, solve problems together, and divide tasks efficiently, ensuring that work progresses smoothly. Moreover, strong team spirit boosts morale, keeping employees engaged and motivated to meet their goals. When people feel valued and connected to their team, they are more likely to take responsibility for their tasks and help colleagues who may be struggling. This collective effort ensures that projects stay on track. In high-pressure situations, a well- bonded team remains resilient and adaptable, finding creative solutions to challenges rather than getting discouraged | source presentation source |
7 | |
| 212 | Why Wait for Real Data? The Power of Synthetic Data in Modern Clinical Trials | Abhijeet Arora | Clinical trials often struggle to access diverse, high-quality data due to delays in data collection, privacy restrictions, and recruitment biases. These challenges slow down study progress and make it harder to build and validate statistical programs early in the process. Synthetic data generation presents a powerful alternative by producing realistic, privacy-preserving datasets that closely mimic real clinical trial data. With open-source Python libraries or advanced platforms, teams can create synthetic datasets to develop and test edit checks, analysis scripts, and models well before real data becomes available. This approach accelerates development, minimizes rework, and reduces dependency on sensitive patient information. Beyond improving efficiency, synthetic data supports more inclusive and unbiased analyses by simulating a broader range of patient profiles | data cleaning, clinical trial, data validation rules, controlled terminology, outlier, data science, stress testing, clinical research, cdisc | source presentation source |
6 |
| 213 | Your Outputs Deserve a Second Set of Eyes - Meet ELARA | Mane Sargsyan | Validating clinical outputs can be slow, repetitive, and vulnerable to human oversight - especially when reviewing layout details like titles, footnotes, page numbers, and axis labels. ELARA (Expert Layout Analysis and Reporting Assistant) is an R-based solution that uses image recognition to automate this visual checking process. It compares two versions of a report, scanning page by page to detect mismatches in structure, layout, and text placement. Instead of manually reviewing outputs line by line or figure by figure, users receive a clear visual summary of what changed - saving time, reducing errors, and increasing confidence in report delivery. ELARA integrates easily into existing workflows and allows reviewers to focus on content, not formatting. Whether validating development vs. QC outputs or checking final deliverables before submission, ELARA simplifies the process and ensures that no page, no label, and no detail is overlooked | mock shell, data cleaning | source presentation source |
7 |
| 214 | {AdmiralMetabolic}: A Collaborative Journey To Develop an Admiral Extension Package for Metabolic Data | Anders Askeland, Olga Starostecka |
In recent years, the field of pharmaceutical data science has witnessed a significant shift towards the R programming language for regulatory analysis and reporting1-3 . While initially focused on outputs, comprehensive tooling now enables R utilization across the entire clinical trial data workflow, including SDTM transformation using {sdtm.oak}4, ADaM dataset creation using {admiral}5, results reporting using tools such as {rtables}6 and {gtsummary}7, and dynamic data visualization facilitated by packages like {teal}8 and {davinci}9. Despite the maturity of {admiral} (ADaM In R Asset Library), its actual implementation remains sparse in specialized therapeutic areas like metabolic diseases, gaps in existing tooling and documentation make implementation daunting. To address this, collaborators from Novo Nordisk, Roche, Boehringer Ingelheim, and Novartis developed {admiralmetabolic}, an extension package for metabolic-specific ADaM datasets | adam dataset, clinical trial, clinical data, cdisc | source presentation source |
5 |
This site serves as a convenient hosting location and does not work as a publisher. All copyrights belong to their respective owners.
The information on this site is provided for general reference purposes only. Whilst every effort is made to ensure that the site is up to date and accurate, no warrant, nor any responsibility or liability for, the accuracy or completeness of the content or for any loss which may arise from reliance on information contained in this site is provided. Links to other sites are provided for your convenience but no responsibility or liability for the content of those sites or of any external site which links to this site is accepted. References to any specific commercial product, process, service, or corporation are also for information only, and do not constitute endorsement, recommendation or favoring.
SAS is a registered trademark of SAS Institute Inc. SAS, Statistical Analysis System, and all other SAS Institute Inc. product or service names are registered trademarks or trademarks of SAS Institute Inc. in the USA and other countries. ® indicates USA registration.