HOME
QUICK SEARCH
AUTHORS
KEYWORDS
HELP
BUGSPHUSE EU 2024 Papers - Index
216 papers. Source: PHUSE (Pharmaceutical Users Software Exchange) EU Connect conference proceedings.
| # | Title× | Authors× | Abstract× | Keywords× | Extras | Pages |
|---|---|---|---|---|---|---|
| 1 | A Guided User Experience for Automated Mapping | Achilleas Zaras | The hype and promise of fully automated mapping have run counter to the reality of complex, messy source data. In this session, we'll present an approach that aims to bridge that gap - a guided user experience for automated mapping. Rather than being a black box turnkey solution, here the focus is on having a human in the loop expert involved in the most complex aspects, while easing the process for the user with an AI-guided experience. Most attempts at automated SDTM solutions have met the 80/20 rule, capable of automating 80% of mappings. This session will address and consider the challenges and limitations faced by these solutions and present an approach that closes in on the 20% so that we can arrive at a fully automated system that keeps the human in the loop. Let's take a moment to think about the USS Enterprise from Star Trek | sdtm programming, data transformation, audit trail, clinical trial, quality control, data cleaning, clinical data, clinical research, raw data, cdisc | source presentation source |
9 |
| 2 | A programmer's introduction to modelling with proc mixed, focusing on how to debug and validate, common mistakes and how to interpret the output | George Gibson | interaction effect, proc step, proc mixed, treatment arm, proc means, change from baseline, nominal variable, interval variable | source presentation source |
6 | |
| 3 | A Tutorial for JFM Analysis in SAS for Modelling Recurrent Events and Death | Edyta Winciorek | Time-to-event data have long been important in many clinical trial designs. Most analyses focus on the time to the first event, ignoring subsequent events or only considering them if they occur before censoring. Several statistical models have been proposed for analyzing multiple dependent competing risks. In this paper, we consider joint frailty models (JFMs) for simultaneously analyzing recurrent events and death. We demonstrate how to code and analyze these complex recurrent event data using SAS through a hands-on guide/tutorial format with real-world examples provided by Liu and Huang (2008) for frailty proportional hazards models. Referring to their work published in [5], we present a simple SAS code to conduct the JFM analysis via PROC NLMIXED |
likelihood function, frailty model, proc mixed, hazard function, proc nlmixed, censoring indicator, proc step, correlation, gamma distribution, clinical trial, in=, regression analysis | source presentation source |
9 |
| 4 | A way out of the PROC COMPARE labyrinth | Judith Kanz | Outputs for the various analyses need to be programmed while the trial is ongoing. But validation cannot wait until final data is available either. Double programming of the input dataset for the reporting procedure usually is the method of choice as validation can easily be repeated on updated data by executing the available validation programs. This poster shows not only how to facilitate programmatic comparison but also how to digest all information provided by the PROC COMPARE procedure step by step |
proc compare, proc step, data cleaning, temporary dataset, proc report, macro parameter, independent qc, in=, adam dataset | source poster source |
9 |
| 5 | Accelerating GSK's Open-Source Journey with The AI Programmer | Andy Nicholls, UK Christina Fillmore |
training set, data science | source presentation source |
7 | |
| 6 | Accelerating Trial Startup: Automated Central Monitoring Visualisations and Safety Package Management | Carsten Rødseth Barsøe, Charlotte Bergknut |
This paper introduces a streamlined approach to accelerate the setup of clinical trials, utilizing test data that seamlessly transitions into real data upon the first patient visit. Emphasizing the automation of central monitoring and safety packages, we showcase crucial information such as enrolment rates, patient demographics, adverse events, and lab results for efficient trial oversight. We simplify the process by defining a minimal set of key data elements applicable across trials. Test ADaMs, containin g these key data elements, serve as the foundation for trial startup visualizations and tables for central monitoring and safety packages. When the first patient visit occurs, the Test ADaMs easily transition into ADaMs with real data. This basic Trial startup package facilitates seamless integration of automated visualizations into trial workflows, ensuring smooth operations from setup to patient visit | adverse events, clinical trial, sdtm domain, cdisc | source presentation source |
4 |
| 7 | ADRG Builder Python Application | Pierre Dostie | In the world of statistical programming, the Analysis Data Reviewer's Guide often remains overlooked when it comes to automation and standardization. However, there are several sections within the ADRG where standardization and automation could significantly impact both the quality and consistency. Consider elements from Define.xml to be replicated within ADRG, serving as a starting point for automating creation. We've developed a Python TM application that offers users with a range of options, from building an initial version of the ADRG document to producing a delivery ready PDF file. Users can also capitalize on the standardization aspect of the tool by integrating standard analysis datasets subsection text, ensuring consistency across trials and projects. Integration of validation tool reports into the ADRG is also an automated process handled by this tool | adam dataset, clinical programming, data cleaning, excel, cdisc | source presentation source |
16 |
| 8 | Advancements in the Evolution of Our SMC Story: Process and Technology | Waldemar Miller, Melanie Leveque |
The imperative for expedited Safety Monitoring Committee (SMC) output delivery is crucial in improving drug development processes, especially given the high time pressure associated with creating SMC output. Our objective was to streamline workflows, improve deliverable usability to accelerate project timelines, and empower professionals to prioritize critical tasks. We reached these goals through standardized processes with minimal study adaptation, rapid data availability using source data, accelerated result disclosure by eliminating manual result transfer, and improved data insights through dynamic outputs. Our efforts culminated in the development of an internally created tool that evolved the conversion of SMC classical ASCII TLF format into interactive Patient Profile visualizations in HTML and TLFs directly within PowerPoint slides using advanced features of R | ods html, data cutoff | source presentation source |
7 |
| 9 | Advances in Clinical Table Creation - Python Updates | Philip Bowsher, Michael Chow |
ThispaperwillintroducetheGreatTablesPythonpackageforcreatingcomplextablesoftenusedinclinicaltrialsreporting.ThemainGreatTablessiteforreferenceishere:https://posit-dev.github.io/great-tables/articles/intro.htmlTheGreatTablespackagehelpsstatisticalprogrammerscreatewonderful-lookingtablesusingthePythonprogramminglanguage.ItisbasedonlearningsanddevelopmentsincreatingtheRpackagegt.TLGs(tables,listings,andgraphs- alsoknownasTLFs/TFLswithFforfigures)areanimportantpartofclinicaltrialreporting.ThegoalofthispaperistointroducestatisticalprogrammerstotheGreatTablespackagefortablecreationinPythonforsubmissionstoregulatorybodies.ThispaperisaimedatstatisticalprogrammersthatarebuildingclinicaltablesandhavewonderedhowtoapproachTLGswiththeGreatTablesPythonpackage.YoucanfindalloftheexamplesinthispaperaswellasthePhuseConferencepresentationslideshere:https://github.com/philbowsher/Clinical-Tables-in-Python-with-Great-TablesApresentationaboutusingGreatTablescanbefoundhere:https://www.youtube.com/watch?v=M5z | adverse events, csr | source presentation source |
12 |
| 10 | Advancing Kidney Assessments for Transplants with AI-driven Applications | Federica Citterio, Soundarya Palanisamy |
Adopting AI-driven decision making in the clinic depends on multiple factors: accuracy, transparency, safety and more The goal is to create a system that optimizes the interaction between humans and AI, striking a balance between the automated capabilities of AI models and the discernment of highly trained professionals. In this presentation we will show an AI-driven app, developed in partnership by SAS and the School of Medicine at University of Cambridge, to assess kidney's pre-implantation quality. The application focuses on the critical time window when histopathologists can review kidney biopsies to determine transplant viability and supports them by automating routine tasks thanks to computer vision models. The user-friendly nature of the application ensures that human professionals remain involved in the process, eliminating technological concerns and allowing doctors to concentrate on patient care | deep learning, sas viya | source presentation source |
8 |
| 11 | Agile Leadership: Inspiring Resilience, Open Collaboration, and Creative Problem-Solving for Outcome-Based Continuous Innovation | Bhavesh Patkar | clinical research, customer satisfaction | source presentation source |
6 | |
| 12 | Agile? Wow! How Agile Ways of Working Can Drive Value for Conventional Clinical Trial Reporting | Sheetal Patel, Michael S Rimler |
Agile teams have been successfully implemented in many industries and primarily used for software development. As our industry continues to evolve its understanding of agile ways of working, it is becoming clear that there exists unrealized value, derived from adoption of agile methods by sponsor companies and CROs. This paper will explore alternative implementation approaches of agile principles on conventional clinical study reporting, comparing the merits and risks of various resourcing models, as well as provide insight into the current GSK approach in this area | resource allocation, clinical trial | source presentation source |
6 |
| 13 | AI Got the Power: Streamlining Clinical Data Creation | Emily Yates, Matt Luppino, Andrew Burd |
clinical trial, edc, clinical programming, excel | source presentation source |
12 | |
| 14 | Aligning Good Transparency Practice with the ISO/IEC 27559 De-Identification Framework | Abby McDonell | Poster | clinical trial, clinical data | source | 1 |
| 15 | An Experiment on the Use of Altair SLC for Clinical Study Analytics | Mark Holland, Michael Byrne |
proc step, data science, clinical trial, out=, clinical programming | source presentation source |
9 | |
| 16 | Analysing Your SAS Log with User-Defined Rules Using an App or Macro | Philip A. Mason | SAS provide some pretty basic help with logs that are produced, typically just linking to errors and warnings. Many people build log checkers to look for particular things of interest in their logs, which usually involves capturing the log and then running some SAS code against it. I made a way to define rules in JSON format which can be read by a SAS macro and used to look for things in a log. This means different rules can be used for different use cases. They can be used via a macro or via a web application I build. The web app can switch between rules, provides summaries, draws diagrams of the code, provides performance stats, and more. Hopefully this functionality might one day be built into SAS, but in the meantime it works well as an addition | sas log, macro variable, out=, macro definition, cpu time | source presentation source |
11 |
| 17 | Analytical Risk-Based Monitoring (ARBM): From Targeted to Zero Source Data Verification (SDV) - Are Teams Ready for the Revolution? | Plamena Kostadinova | Shifting the focus from SDV on 100% of data to targeted SDV on critical data that matters most to the study objectives is currently a well-accepted approach in clinical trials. By applying it, study teams can direct monitoring and oversight on areas more likely to affect patient safety and data quality. Is the next step removing SDV altogether? This paper presents a case which first implemented targeted SDV for a subset of subjects. To ensure data quality and integrity, several ARBM methods were implemented. After further observation on real-time data point changes, SDV was further reduced to complete shutdown | data cleaning, clinical trial, data integrity, data review | source presentation source |
5 |
| 18 | Anatomy of a Table - Our Continued Journey to Automating Output Delivery Through Metadata | Edwin van Stein | To improve traceability, transparency and consistency, GSK is implementing automation of analyses and output creation through standardised metadata. Tool-agnostic analysis results metadata (A highly- enriched adaptation of CDISC ARM) is used to produce analysis results data. Following the WORM principle (Write Once, Read Many), tool-agnostic analysis output metadata can then be used to render those results into one or more displays (tables, figures and listings, both post-text and in-text), other outputs (e.g. XML files for disclosure) and systems (in whatever format those systems require) without reproducing the numbers. This leads to a process where it's easier to validate and reuse analysis results and to build tools in a multilingual setting. This poster is a follow-up to my presentation from the PHUSE EU Connect 2023 and focuses on visualising the metadata for a single table and how it all fits together in our journey to standardising our | analysis set, safety population, cdisc | source poster source |
10 |
| 19 | Another Year, Another Migration: Planning the Move From Entimo entimICE to SAS LSAF | Panteleo Nacci | To better cope with the increasing needs and demands of the evolving pharma scenario, AstraZeneca (AZ) has launched Redefining Clinical Data Flow (RCDF), an ambitious project to completely revamp how clinical research is implemented in the company. As part of this larger project, it has been decided to change our current Statistical Computing Environment (SCE) from entimICE-AZ to SAS Life Sciences Analytical Framework (LSAF). This change will require the gradual migration of all clinical study data, programs, deliverables, etc. from the existing platform to a completely different new one, posing a series of issues with maintaining the normal flow of activities | macro variable, automatic macro variable | source presentation source |
7 |
| 20 | Artificial Intelligence Potential Use Cases in Clinical Data Engineering | Deepak Pakalapati, Mounika Pillamarapu |
Clinical data engineering deals with aggregation, cleaning, reconciliation, and standardization of data from various sources to ensure data veracity and readiness for use in clinical trial reporting. Artificial intelligence (AI) as a technology has just begun its logarithmic phase and can contribute immensely to almost all stages of drug development, more so for those involving data engineering, processing, and reporting. Applying AI techniques to data engineering requirements can speed up the processes, saving both time and money to the sponsors. Deduplication, data augmentation, real world data sanitization, outlier detection, data standardization, missing data and error correction are just a few examples of areas where AI techniques like machine learning, deep learning, feature extraction and natural language processing can have a focused impact. These use cases are discussed in considerable detail | clinical trial, imputation, missing value, data cleaning, machine learning, clinical data, artificial intelligence, clinical research, sensitivity analysis, outlier, data integrity, neural network, power analysis, sample, adverse events, data science | source presentation source |
8 |
| 21 | Associated Patient Mapping in SDTM | Sadiya Ait M'Bark | The poster will delve into the use and representation of non-study subject participants. Study subjects' information/data is collected and mapped in Study Data Tabulation Model (SDTM) datasets. However, in some clinical studies, information about persons other than the study participant may also be collected and these needs to be included in SDTM models. These non-study subjects can include, but is not limited to a family member, an organ donor, a caregiver, etc. They are classified as Associated Persons (AP). CDISC has specific guidelines that deals with AP data and this poster will share some best practices in handling this type of data | clinical trial, sdtmig, cdisc | source poster source |
2 |
| 22 | Attracting Young Talented Programmers to the Pharmaceutical Industry in a Post-COVID World: Adapting to the Evolving Working Landscape | Lewis Turner | The COVID pandemic changed the world, and the Pharmaceutical Industry was not exempt from that, but perhaps in ways that were not immediately obvious, and the effects of which are now coming to fruition. The spike in popularity with remote-based working has created an office vacuum which may prove difficult to fill when the idea has spread in its appeal to junior programmers and university graduates despite their lack of industry experience, connections, and an employer's uncertainty as to their return on investment. History would argue that people junior in their careers benefit from being in an office, which evokes the questions, how do we recruit talent and build them up for their future when their fueled desire to remotely work may hinder them and the company? How does the industry adapt to changing tides and place trust in its less supervised workforce, whilst still trying to achieve client satisfaction? | pharmaceutical industry, clinical programming | source poster source |
8 |
| 23 | Auto-Generating the List of Tables from a Statistical Analysis Plan - A Case Study | Jaya Simha Inampudi, Krishna Gopal |
Presentation | data science, adverse events, analysis set, clinical trial, crf, proof of concept, safety population | source | 14 |
| 24 | Automate Your Metadata: Reading and Modifying Information from PDF Files Using Python | Salomo Benedikt Sparna, Renate Scheiner-Sparna |
eCRF forms are normally exported from the data capture systems in PDF, in this format they are annotated according to CDISC SDTM standard. The annotations are meta-information that can be further processed for programming specifications or other purposes. This presentation shows how to export the annotations to a pre-structured Microsoft EXCEL file using Python. In addition to exporting the annotation fields to EXCEL rows, information can be added, e.g. for standard variables that are always present in CDISC SDTM standard. Reading and modifying data from PDF files is also useful for cleaning messy files that contain more than one generation of annotations. Some older annotations are no more in text boxes but embedded in the PDF body. This can be harmonized using Python. Modified information can also be written back to the PDF file. The code will be available on GitHub (https://github.com/ssparna/PHUSE-EU-Connect-SM15) |
clinical trial, excel, cdisc | source presentation source |
6 |
| 25 | Automating Real-World Data Analysis Using R Packages and Shiny: A Three-Stage Workflow | Stavros Oikonomou, Louise Raiteri |
Real-world data (RWD) plays a crucial role in healthcare and research, yet its accessibility is hindered by legal and privacy constraints. Researchers often develop programming code on synthetic data, with only summary results available when applied to real data. To address this, the IQVIA Biostatistics analytics hub have devised an automated, three-stage approach for conducting RWD studies. The first stage involves a data-source specific R package for standardised, validated transformation and cleaning of RWD. The second stage employs a data-source agnostic R package for cohort selection, summary output creation, and visualisations based on individual patient data, incorporating statistical disclosure control. The final stage utilises a Shiny[1] application to provide visualisations based on summarised data created by the R package | stage 3, nominal variable, data science, proc means, load, stage 1, stage 2, excel, data cleaning | source presentation source |
8 |
| 26 | Back to the Drawing Board - Accelerating RShiny Application Development with Effective User Interface Designs | Barbara Mikulasova | data cleaning, data science | source poster source |
4 | |
| 27 | Be the Cosmic Guardian of Your Codebase with GIT | Shuai Wu, Brian M. Lang, Carl Herremans, Christian Koul |
Changes in computing environments, automation, and programming languages, particularly in drug development, have driven the widespread adoption of open-source technologies. Among these, version control systems like Git have become highly popular. However, fully understanding and using Git's extensive capabilities can be challenging, especially for beginners who often stick to basic commands like git add, git commit, and git push. This paper addresses this gap by using a metaphorical approach, drawing parallels between Git and the Sacred Timeline from Marvel's Loki. By likening the Git main branch to the Sacred Timeline and feature branches to alternate timelines, this paper provides a relatable framework that enhances programmers' comprehension of Git's branching, merging, and commit history. This analogy helps illustrate the importance of maintaining code integrity and resolving conflicts, much like the Time Variance Authority (TVA) safeguards the multiverse's coherence | source presentation source |
9 | |
| 28 | Behind the Blind: Enhancing Clinical Data Management with Python | Niels Schrauwen | Efficient and profound management of unblinding data streams is indispensable to a clinical trial's success. This is achieved by a seamless conversion of external unblinding data into SDTM-conformant datasets. This abstract explores the use of Python as a tool for converting those complex data into high-quality SDTM datasets. We will share our experiences with using Python to streamline the SDTM conversion process, enabling efficient reconciliation, dataset crea tion for preliminary analysis, and study -specific data validation. Python's versatility and readability facilitate code programming , debugging and review. Its rich libraries, integration capabilities, and strong community support make it a compelling choice as a programming language. These features will help significantly in achieving specific study needs. Moreover, we will emphasize Python's scalability to align with future trends in managing large volumes of clinical data | clinical trial, data cleaning, sdtm programming, sdtm domain, clinical data, clinical research, proc sql, data governance, cdisc | source presentation source |
5 |
| 29 | Biomarkers Unleashed: A Programming Pipeline from Chaos to Clarity in Genomic Data | Saskia Ständler, Lea Zittlau |
Biomarker analyses are becoming increasingly important in diagnostics and clinical trials, not only in oncology but also in other fields, as the gained insights pave the way to personalized therapies. The generic term "biomarker" includes a great variety of biological molecules, and this diversity also applies to the type of their analysis methods. One promising case is the use of genetic data which can be obtained via next generation sequencing (NGS). Working in the genomics field, statistical programmers face the challenge of handling a huge amount of primarily unstructured raw data. Here, we show an exemplary pipeline which details the processing of NGS data to create a final analysis set. Thereby, we address questions of data standardization, harmonization, and annotation of genomic data by using public databases. Sharing insights about such a workflow hopefully increases the awareness of the challenges in biomarker analyses and their solutions | statistical programmer, sample, adam dataset, clinical trial, biomarker, raw data, cdisc | source presentation source |
10 |
| 30 | Biomedical Question Answering with Large Language Models | Matthew Moody | Question answering (QA) systems have the potential to improve biomedical research and the quality of clinical care by providing users with the latest and most relevant evidence. Usually, the input of a QA system is a question, potentially alongside a knowledge base containing terminologies and a corpus of scientific publications; and the output contains the answer(s) to the question. Explanation is important, i.e. by showing where the answers come from and the reasoning path to the answer. In this paper, I will explore a biomedical QA dataset and implement a QA system through the use of Language Models, utilizing key techniques such as Retrieval Augmented Generation (RAG), and I will explore how the explanation of an answer can be improved by providing provenance to a publication. Keywords-Question answering, Natural Language Processing, Large Language Models | experimentation, training set | source presentation source |
10 |
| 31 | Biometrics Transformation - The Path to Business ROI | Jason Housley, Alison Banda |
source presentation source |
2 | ||
| 32 | Blink and You Will Miss It! | Daniel Lindenbaum | Hash table programming opens doors to under-explored and powerful areas of SAS but many programmers avoid it due to quirky syntax and perceived difficulties. For those brave enough to roll up their sleeves and accept the challenge, there are great rewards to be uncovered. Even at an elementary level, hash objects offer novel and concise solutions to many daily programming tasks. Data operations are carried out in memory and minimize disk I/O. Memory operates in the nanosecond region as opposed to the milliseconds required for disk access - one thousand times faster. Sort, Merge (without pre-sorting), Lookup and Modify datasets serve as an introduction to the possibilities that are available when adopting this efficient element of the SAS language | hash table, sysrc, hash iterator, out= | source presentation source |
9 |
| 33 | Block by Block: introducing blockr, the Flexible Open-Source Dashboard Builder | Karma Tarap, Jean-Philippe Coene |
This paper introduces blockr, an open-source R package designed to streamline the creation of data analysis pipelines and interactive dashboards. By providing a visual, block-based interface, blockr caters to both end-users and developers, bridging the gap between traditionally distinct roles in data science. We describe the architectural framework of blockr, explore its applications in clinical research, and examine its potential for broad adoption across various industries. The analysis demonstrates how blockr can streamline workflows, enhance collaboration, and democratize advanced analytics. A comparison with commercial software solutions, such as JMP Clinical and Spotfire, underscores the accessibility and flexibility of this platform in rapidly evolving research environments | clinical research, data science, data transformation | source presentation source |
7 |
| 34 | Bootstrap Simulation for Outlier Detection in Operational Site Metrics Using Patient Reassignment | Björn Koneswarakantha | bootstrap, likelihood function, outlier, sample, adverse events | source presentation source |
3 | |
| 35 | Bridging Platforms: Integrating RStudio POSIT and SAS Grid into the Same Environment | Sebastià Barceló | Integrating RStudio POSIT into the same environment where we have SAS Grid posed several challenges due to disparate platforms - SAS Grid installed on Windows server and POSIT installed on Linux server. This presentation outlines the hurdles encountered during the integration process and the innovative solutions devised to overcome them. From addressing compatibility issues to streamlining access to the same drives from both platforms, we navigated through complexities to ensure seamless integration. Through collaborative efforts and strategic planning, we successfully bridged the gap between Windows and Linux environments, enabling efficient utilization of RStudio POSIT and SAS Grid accessing the same data. Join us to explore our journey, learn from our experiences, and gain insights into effectively integrating diverse technologies within a unified environment | sas grid, sas base, data science | source poster source |
5 |
| 36 | Bridging Statistical Programming and Data Management: A Job Rotation Experience | Frederikke Høyer | data governance, clinical programming, clinical trial, statistical programmer, database lock, data science, clinical data, data cleaning | source | 4 | |
| 37 | Bridging the Gap: Role Transition from SAS Programmer to Data Scientist | Mrityunjay Kumar, Shashikant Kumar |
data science, sas programmer | source presentation source |
7 | |
| 38 | Bridging the Virtual Gap: Enlivening Our Global Programming Team | Daphne Vuurstaek, Pekka Heikkilä |
As our Johnson & Johnson global programming team from all across the world continues to expand and remote work becomes more common, the physical distance between colleagues is rapidly increasing. Our company has been actively exploring ways to bring people closer together and one of them is the ENLIVEN initiative. Its members, programmers from different regions, are organizing face-to-face and online activities, fostering friendly competition, enabling creative collaboration and building a virtual community. The meaning of "ENLIVEN" is to make something more entertaining, engaging and appealing, which is exactly what we aim to do. We want to build a stronger feeling of community by encouraging open communications, facilitate authentic connections, and promoting cross-boundary learnings. We target to create ONE team that is inclusive, offers support, and empowers every individual | sas engineer, statistical programmer | source presentation source |
6 |
| 39 | Bringing DevOps to the pharmaverse: The PHUSE PharmaDevOps Working Group Project's New Website | James Black, Benjamin Straub |
Poster | source | 1 | |
| 40 | Building CORE Strength: An Implementation Exercise | Addie Nina Olsen, Stine Winkler Bertelsen |
Validation of clinical data using CDISC conformance rules is essential for ensuring consistency, quality and interoperability of data. The CDISC Open Rules Engine (CORE) project provides both machine-readable rules and an open-source engine for data validation. We present our learnings from implementing the CORE engine in an automated SDTM generation flow on a SAS Viya platform: the challenges that we have encountered and the promising aspects of the implementation. In addition, we outline plans for enhancing our CORE implementation with explanations and problem-solving tips that can assist the end user in aligning to CDISC standards | data cleaning, clinical data, sas viya, cdisc | source poster source |
3 |
| 41 | Building for the future: Strategies for nurturing and developing new line managers | Eniko Bront | Transitioning from programming to line management presents a unique journey for each programmer embarking on this path and comes with its own set of challenges and insights. This paper leverages the knowledge and guidance of experienced leaders and mentors to propose strategies to ensure new line managers feel supported effectively across the different focus areas. Beyond making use of conventional training methods, mentoring/coaching programs and knowledge-sharing forums, the paper also explores the utilization of tools and data that would empower new managers to develop, while balancing programming and line management. Through the implementation of these strategic approaches, both new line managers and their development teams can effectively tackle daily challenges while laying the groundwork for fostering the next generation of leaders | source presentation source |
4 | |
| 42 | BY-lieve in better BY group tables | Chris Spence | When tasked with creating tables using a BY variable one of the biggest challenges is displaying correct denominator labels for each treatment within each BY group. For this, most programmers will resort to macro techniques. The most common method is to simply wrap the entire program in a macro and call it for each BY group with denominator labels retrieved from the variable labels for each run. Alternatively, only the PROC REPORT is wrapped in a macro, with the denominator labels defined using macro variables within the PROC REPORT, often with consecutive ampersand resolutions. This paper will show an alternative method in which ACROSS variables are leveraged within a single PROC REPORT. We demonstrate with this method how denominator labels can be correctly assigned in such a report. Additionally, we show how to expand the technique to include any required spanning headers and finally how this facilitates enhanced programmatic Quality Control (QC) |
by group processing, proc report, proc freq, proc step, create table, adverse events, proc sql | source presentation source |
10 |
| 43 | CDISC SDTM/SEND Generation Without Statistical Software: A Software Demonstration | Jozef Aerts | The presentation consists of a demonstration of SDTM-ETL, a GUI-based, CDISC-standards savvy software for semi- automated mapping to and dataset generation of CDISC SDTM and SEND submission datasets | statistical software, proc sql, cdisc | source | 3 |
| 44 | Challenges and Wins of Pooling Internal and External Alzheimer's Data for Secondary Use | Michaela Mertes, Iska Gans |
This presentation will give an overview of a pooled dataset including internal and external Alzheimer's study data for secondary use. Detailing the process to get all the data into a standardized and harmonized structure and content explaining some of the challenges. The final goal of this dataset is to set the foundation for elaborating and optimizing the analysis for scientists. So far 35 studies are included with over 75 000 subjects (including screen failures) | data mart, clinical trial, sample | source presentation source |
9 |
| 45 | Challenges Faced When Creating Dosing Records for a Population PK Dataset | Marianne van Vugt, Gerly van der Vleuten |
The ADaM datasets ADEX and ADPC are the two most important source datasets required for the creation of population pharmacokinetic (PopPK) datasets. The key piece of information here being the date and time at which either drug administration or PK sampling took place. One of the primary challenges in constructing a PopPK datasets is ensuring the accuracy of dosing records, especially when dealing with missing dosing times. This paper explores the structure of PopPK datasets, methods for capturing dosing information, and the challenges faced in imputation of missing dosing data. We examine standard imputation strategies, as recommended by CDISC, and address special considerations for complex dosing regimens, such as BID (twice-daily) dosing. By applying standardized rules and thorough documentation, programmers can provide an accurate analysis dataset for population pharmacokinetic modeling | imputation, sample, population pk, cdisc | source presentation source |
3 |
| 46 | Changing ODS Outputting Methods to Streamline the TFL Output Process | Richard Fryer, Rose O'Nians Michalowicz |
The process of converting TLF outputs into their final format, normally RTF to PDF, and compiling them for delivery is time consuming and often riddled with issues. This task is required during a period on the critical delivery path where time is limited. With intricate bookmarking and multiple format requirements to adhere to, the efficient set-up of in-program replaying of outputs straight to the final destination is often overlooked. The resultant default PDF output with automated bookmarking saves significant time, and memory usage, in the TLF process and comes with a list of additional benefits, including ease of compiling. The use of ODS Document has been covered in a multitude of papers in the past, however this paper will document the real-life practical implementation of the procedure, including the necessary process updates required to navigate issues which are not well documented in previous papers |
proc step, proc report, macro variable, output delivery system, by group processing, macro call, ods pdf, ods rtf | source presentation source |
8 |
| 47 | Choosing the Right Tool: A Comparative Evaluation of SAS, R and Python for Sankey and Swimmer Plot Generation in Clinical Data | Julien Frenehard, Neelam Yadav |
Understanding intricate healthcare datasets and making informed decisions is made easier through clinical data visualization. Sankey plots are renowned for their proficiency in illustrating flow and relationships within clinical data, while swimmer plots offer invaluable insights into treatment courses, including specific patient events | swimmer plot, bar chart, proc sgplot, proc step | source presentation source |
11 |
| 48 | Clinical Disease Endpoint Flag Implementation for Vaccine Studies | Charumathy Sreeraman | Before being approved for use in people, every novel vaccine(s) must undergo stringent safety, immunogenicity, and protective efficacy evaluations. Assessment of a vaccine's reactogenicity -its ability to suggest that a drug has the potential to elicit a typical unfavorable response when ingested-is a critical safety step in the development of vaccines. Another aspect that comes under viability is the antibody's immunogenicity. The ability of a foreign substance, like an antigen, to cause an immune response in a human body is known as immunogenicity. The degree to which vaccination protects recipients from adverse health outcomes, including infection, serious illness, hospitalization, and death, in the actual world. The clinical disease endpoint flag is recommended by the regulatory bodies as an approach for evaluating the efficacy of treatment | phase iii, clinical endpoint, clinical trial, phase ii, phase i | source presentation source |
7 |
| 49 | Combining the Power of RShiny and SAS Viya | David Weik | RShiny is a flexible tool to build out data visualization and analytical applications. SAS Viya complements this by offering a powerful analytical platform that not only performs advanced data processing but also ensures enterprise- level security and data access management. This presentation will explore the synergistic potential of combining RShiny with SAS Viya, illustrating how this integration enhances functionality, accelerates iteration cycles, and expands scalability within institutions. Attendees will gain insights into practical strategies for leveraging these technologies to streamline workflows, maintain secure and authorized data access, and ultimately drive forward innovation | sas viya, access control, data wrangling | source presentation source |
4 |
| 50 | Comparative Analysis: Creating Kaplan-Meier Plots Using SAS vs Python - A Step-by-Step Approach | Mrityunjay Kumar, Ephicacy Lifescience |
In the realm of biomedical research, Kaplan-Meier (KM) plots are quintessential for survival analysis. This abstract proposes a comparative exploration of constructing KM plots using two widely adopted programming languages, SAS, and Python. The presentation will offer a systematic, step-by-step demonstration of the process, highlighting the unique methodologies, syntax, and functionalities of each language based on hypothetical data. Attendees will gain insights into the advantages and limitations of both SAS and Python in generating KM plots, empowering them to make in formed decisions regarding their preferred analytical tool. Through this comparative analysis, researchers and analysts will enhance their proficiency in survival analysis and expand their interest of programming skills by bridging the gap between SAS and Python methodologies | hazard ratio, regression analysis | source presentation source |
13 |
| 51 | Comparison of the Wilcoxon-Signed Rank Test with the Hodges-Lehmann Estimator in SAS, R and StatXact as Part of the PHUSE CAMIS Project | Agnieszka Tomczyk, Lyn Taylor |
The Wilcoxon signed-rank test is a non-parametric test which can be used instead of the paired Student's t-test when assumptions regarding a normal distribution are not valid. The test compares if the median difference between paired samples is significantly different to 0. Whilst the test itself is generally straight forward to implement in software such as SAS, R and StatXact and the same results can be obtained, the calculation of a corresponding Hodges-Lehman estimate and confidence interval in different software can be tricky! This paper covers implementation of the Wilcoxon signed-rank test with its p value, Hodges-Lehmann estimator and corresponding confidence interval in different software. Various methods including exact & non-exact ones, non-exact with & without continuity correction, disregarding & considering "0" differences or ties are discussed and compared. Analysis is performed on an asthmatic cross-over study dataset | p-value, rank test, wilcoxon signed-rank, confidence interval, sample, proc step, normal distribution, crossover study, exact test | source presentation source |
7 |
| 52 | Connecting Standard from Design to Analysis to Submission: Finishing the CDISC 360 Vision | Peter Van Reusel | Presentation | phase ii, data pipeline, phase i, proof of concept, data science, data wrangling, data standard, cdisc | source | 25 |
| 53 | Considerations for the Submission of RWD using CDISC with Insights from HL7 FHIR and OMOP | 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 four 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 an example CDISC SDTM submission and highlighting updates needed to the SDTM to better accommodate RWD. Specifically, we examine challenges and considerations for the Exposure domain within this paper. Considering how other models such as HL7 FHIR, OMOP, and PCORnet handle certain elements, we propose new solutions that are needed to support RWD. These ideas are intended to help sponsors submit RWD and to start the conversation with SDOs such as CDISC around the updates needed for more efficient regulatory review | raw data, insurance claims, clinical data, clinical trial, clinical research, data standard, cdisc | source presentation source |
13 |
| 54 | Coral Reef- A FAIR metadata layer to provide clinical study data at your fingertips | Javier D. Fernández, Nelia Lasierra |
protocol deviations, csv | source presentation source |
5 | |
| 55 | Data Curation Beyond Today's Standards | Catharina Dahlbo, Eva Kelty |
Clinical trial data is inherently complex and sourced from various systems, including electronic health records, patient surveys, compliance tools, and standard or specialty laboratories. This diverse origin and storage across different databases pose significant challenges to data consistency. Therefore, data curation is essential for maximizing the utility of clinical trial data. While standards like CDISC are utilized for specific purposes, challenges persist due to tabular formats and inconsistencies across domains. To fully leverage clinical trial data for scientific exploration and knowledge enhancement, additional processing is necessary. Adopting a model optimized for exploring data stored in a graph database offers significant benefits. Graph databases provide flexibility with an easily expandable data model, organizing data as a mind map, thus facilitating comprehensive data exploration and visualization | clinical trial, clinical data, curated data, adverse events, raw data, cdisc | source poster source |
5 |
| 56 | Data integration on CDISC ADaM Level: Multi-purpose Structures of ADSL and BDS Datasets | Renate Scheiner-Sparna | A data integration has to cover different requirements of the sponsor (ISS, ISE, publications, requests from authorities, management decisions). This presentation compares different structures of ADSL and BDS datasets to meet these expectations. ADSL has one line per subject and study. The integrated period can be represented either by integrated variables, or by a separate line, which covers all the single studies. As third possibility, a separate analysis dataset can hold the information. A use case of a rare disease unit is presented. The identifiers allow for either drawing information on the whole period, or on one or more of the single studies. The entities of the single studies can be used as a whole, or one or more can easily be selected. The dataset structure will be compared to the ones presented in other publications (Zhong et al. [1]; Phelan et al. [2]; Minjoe [3] | adam dataset, data integration, treatment period, cdisc | source presentation source |
6 |
| 57 | Data Pipelines in Clinical Trials Leveraging Apache Airflow for Automated Task Management | Kostiantyn Zvieriev | Currently, there exists a wide array of tools available for data processing and analysis, including programming languages with data analysis packages, big data processing solutions distributed across clusters, and various data storage systems from local files to cloud databases. The main challenge lies in effectively integrating these interconnected systems to function as a cohesive data processing pipeline. Apache Airflow is an open-source platform designed for orchestrating complex workflows and data pipelines. With an intuitive interface, it simplifies task management and scheduling, automating workflows efficiently. Airflow represents workflows as Directed Acyclic Graphs (DAGs), aiding visualization of dependencies and task monitoring. Airflow provides extensive support for various data sources and integrations, facilitating interaction with popular tools and technologies in the data ecosystem | data pipeline, clinical trial, data wrangling | source presentation source |
16 |
| 58 | Data Traceability Map with R Shiny | Alexander Mkrtchyan, Sevak Kamalyan |
sdtm domain, cdisc | source presentation source |
8 | |
| 59 | Database Lock Is Only the Beginning: The Long, Winding Journey of Secondary Data Reuse at Roche | Tamara Senior | Traditionally in the industry, once a database is locked and study closed, it's considered the end of the road. However in recent years, some pharmaceutical companies have breathed new life into their study data by taking a novel approach and thus coining the buzzword "secondary re-use". At Roche this work has led to critical insights, expanding disease area knowledge and furthering drug development, but it's not without its challenges. By utilising data-driven approaches and tools such as R, PythonTM, LLMs, and cloud-based storage solutions, our secondary reuse of data involves gathering, curating, harmonising, and integrating data into valuable data assets. This presentation will address the challenge of harmonising data from trials conducted years ago with missingness or collected using different data standard models | data mart, data standard, clinical data, sample | source presentation source |
3 |
| 60 | Dataset-JSON as an Alternative Transport Format for Regulatory Submissions | Stuart Malcolm, Samuel Hume |
Poster | source | 1 | |
| 61 | DaVinci: A Modular Framework for Interactive Data Exploration and Visualisation in Clinical Research | Isabel Glauß, Stefan Döring |
DaVinci is a dynamic, modular framework designed for interactive data exploration and visualization in clinical research. It is particularly intended to support exploratory biomarker analysis and safety review during study conduct. The framework's modules function as building blocks that can be combined to create customized R/Shiny applications, tailored to specific clinical trial and research project needs. DaVinci offers a variety of functionalities, including the ability to plot, for example, heatmaps, waterfall, and eDISH plots, as well as generate tables and listings. Currently, it boasts twelve output modules, with ongoing development to expand its capabilities. In a move towards community-driven growth and refinement, DaVinci has been open-sourced recently. It also integrates with modules from other open-source frameworks like Teal, further diversifying its range of functionalities | biomarker, clinical research, clinical trial, data review, adverse events, concomitant medications | source presentation source |
5 |
| 62 | De-Identification of Participant Data for Public Disclosure | Nicola Perry | k-anonymity is a property possessed by certain anonymized data. The concept of k-anonymity was first introduced by Latanya Sweeney and Pierangela Samarati in a paper published in 1998. A release of data is said to have the k- anonymity property if the information for each person contained in the release cannot be distinguished from at least k - 1 individuals whose information also appear in the release. According to Health Canada k should be 11. A public sharing threshold of 0.09 indicates that a person's information cannot be shared unless it cannot be distinguished from at least 10 other individuals within the same release. This paper aims to explain how GSK have applied this threshold in data displays that have a number of participants with direct identifiers (age/race/ethnicity) for studies in scope of public disclosure to minimize the possibility of re-identification of participants | clinical trial, proc means | source poster source |
7 |
| 63 | Dealing with the poor performer | Stephen Prust | Nobody wants to be bad at their job but sometimes it happens. When managing teams this situation will occur and can be one of the most difficult things to resolve. The presence of the virtual world has perhaps made this aspect of managing even harder. This presentation takes a systematic approach to the issue. It is not sufficient to solely identify the problem and say it must be resolved. There are many considerations. We have a duty of care to our staff and a responsibility to the team as a whole. Practical approaches are discussed and backed-up by personal experiences.. Also discussed are references to applicable models including change management, team development, and learning styles. The objective of the presentation is to suggest processes and practical methods for the office | source presentation source |
7 | |
| 64 | Debugging Leadership: Cracking the Code of Statistical Programming Teams in Pharma | Jessica Hirsch, Vitali Gering |
In the dynamic world of pharmaceutical statistical programming, effective leadership is essential for maximizing the potential of diverse teams. This presentation explores leading statistical programming teams in pharma, emphasizing the blend of technical expertise and human dynamics. We'll examine the unique challenges of statistical programmers, balancing precision work with team management. From the analytical mindset to collaborative drug development, we'll dissect key leadership considerations. The talk will provide actionable insights into fostering a culture of innovation, continuous learning, and adaptability within statistical programming teams. Drawing on real- world examples and best practices, we will discuss strategies to enhance communication, mitigate challenges arising from varied skill sets, and cultivate an environment where every team member feels valued and motivated | clinical programming, clinical research | source presentation source |
2 |
| 65 | Developing Validated R Packages for Standardised and Efficient Statistical Report Generation at Roche Diagnostics | Dorothee Childs, Christoph Ehret |
data cleaning, data science, clinical trial, r package | source presentation source |
9 | |
| 66 | Development of a Customised Statistical Computing Environment for Clinical Trial Data Analysis | Christoph Frank, Mario Lozina |
data cleaning, sas studio | source presentation source |
5 | |
| 67 | Development of USDM through translation of human-readable protocols | Jasmine Kestemont, Stijn Rogiers |
With the introduction of the Unified Studies Definition Model (USDM), there is scope to translate diverse human- readable study protocols into standardized format and develop a library that can be used to write future protocols. The information can be extracted using Natural Language Process (NLP) Language Interpretation for textural Information (LITI) rules. I n order to ensure LITI rules can work for variation in protocols text, fair number of rules need to be developed, a time-consuming and complex process. Similarly information can be extracted also using Large Language Models (LLM). Developing an end-to-end process to extract information can be complicated and LLMs have risk of hallucination. However, using combination of NLP and LLMs can reduce number of LITI rules and can also help validate LLM response their by reducing hallucination risk | clinical trial, proof of concept | source presentation source |
6 |
| 68 | Digital Data Flow: Achieving Protocol Digitalisation and Clinical Research Interoperability Through Multi-Stakeholder Collaboration | Mikkel Traun, Lakshmi Sankar |
Presentation | proof of concept, clinical research, clinical trial, data standard | source | 21 |
| 69 | Do Secondments Work? For You? For Your Team? For Your Organisation? | Saqwindar Singh | Presentation | source | 25 | |
| 70 | Driving Efficiency and Automation in TFL Generation: A Case Study of Establishing an Oncology TFL Library and Analysis Results Metadata Repository | Salil Parab, Bhavin Busa |
Poster | machine learning, mock shell | source | 1 |
| 71 | Dynamic Lookups in R | Duong Tran | Poster | proc format, proc step, lookup, if-then-else, proc catalog | source | 1 |
| 72 | Effective APIs for SAS Language Applications | Oliver Robinson, T. Randall Betancourt |
Pharmaceutical firms have large numbers of validated SAS programs. These programs are run repeatedly with alterations to produce output. An improvement in this process is passing parameters to validated autocall macros. Programs are hand-edited, tested, then output is copied and pasted to distribute results. These multiple, manual steps are expensive, error-prone and, importantly in a regulated environment, hard to reproduce reliably. The use of SAS language autocall macro libraries provides a method for effective reuse of SAS programs. Defining REST APIs allow firms to control and automate this logic as well as expand process automation to encompass R, Python, SQL, and shell scripts. We describe consuming SAS autocall macros embedded with R and Python scripts through API calls passing macro parameters to automate program execution. This abstraction enables consumption of validated SAS programs without the client/consumer needing to understand details of the underlying program logic | rest api, proc http, proc step, autocall macro, macro variable, proc sql, excel | source presentation source |
17 |
| 73 | End-to-end Automation Using a Model-Driven Development (MDD) Workbench | Stuart Malcolm | To achieve the CDISC 360 vision of end-to-end standards-based automation requires a paradigm-shift in the way we work with metadata. This paper introduces model-driven development (MDD) as an agile approach to the creation, management and processing of standards-based clinical trial metadata, and presents examples of how MDD can be used for automation using CDISC Standards | clinical research, clinical trial, cdisc | source presentation source |
7 |
| 74 | Enhancing Quality of External Data | Pascale Youinou | During clinical study set-up, both sponsor and CRO have a lot to consider, especially when collaborating with many different vendors (EDC, IWRS, ePRO, etc.). Sponsors still may choose not to integrate all databases in one system. When your normal process considers having DM and Stat teams involved upfront to proactively minimize the issues for data cleaning, SDTM, analysis and submission, working with primary output external to EDC might affect data quality. Often multiple vendors participate in the study and the DM team must deal with several databases' reconciliation in parallel | clinical database, edc, data integration, data cleaning, clinical trial, cdisc | source presentation source |
6 |
| 75 | Enhancing Standard Use by Creating a New Pivotal Role: The Standard Implementation Lead | Marc-Antoine Prodhomme, David Parkinson |
At Johnson & Johnson Innovative Medicine, we are in a continuous process of improving our R&D strategies. This vision is also reflected in our Clinical Data Standard (CDS) organization. Based on a gap analysis, we have identified areas of improvement of standard adoption and consumption. With this presentation, I would share a new initiative where we are focusing on changing the culture towards the use of data standards by introducing a new pivotal role who acts as liaison between study team and CDS organization. The newly created Standard Implementation Lead (SIL) role is integrated in study team to promote the Enhancing Standard Utilization initiative 's following 3 pillars. First advising and promoting the adoption of the new simplified connected standards. Then playing a central role in a new governance strategy. And finally, being a central actor to evaluate and increase standards compliance | clinical trial, ecrf | source presentation source |
3 |
| 76 | Ensuring Governance and Traceability of RWD from Source to Regulatory Submission: Use Case | Michael Kolzet, Berber Snoeijer |
proof of concept, regulatory submission | source presentation source |
12 | |
| 77 | Estimands: How to Make ADaM Implementation Your Friend | Jaclyn Amberg | The concept of estimands introduced in ICH E9 (R1) [1] is more and more used in clinical trials and required by authorities. Nevertheless, at the time of first implementations of this concept, no established guidance about details of implementation in data standards, especially in ADaM, was available. Motivated by examples from real clinical studies, we present approaches that we implemented, as well as challenges we faced, during set-up and conduct of statistical programming activities. The focus is especially on strategies to address intercurrent events (ICEs) for analysis, on data selection for different estimands in ADaM, and on the implementation of different imputation approaches. We also show our solutions in more complex settings like multiple ICEs by subject or varying strategies for different estimands | imputation, missing at random, missing value, intercurrent event, multiple imputation, missing not at random, hypothetical strategy, clinical trial, adam dataset, data science, while-on-treatment, cdisc | source presentation source |
10 |
| 78 | Evaluating the 'Fit for Purpose' of Real-World Data Sources for Regulatory Submissions | Berber Snoeijer, Lakshmi Mantha |
Poster | source | 1 | |
| 79 | Exploring Microsoft Power Apps for creating an analysis definition capture tool | Quincy Kaka | In today's pharmaceutical industry, the need for efficient, user-friendly, and scalable solutions is paramount. Microsoft PowerApps, a low-code platform known for ease of use and rapid development capabilities, benefiting from a built-in database and seamless integration with existing systems, can enable healthcare professionals to create tailored applications. Presented here is a summary lessons learned from exploring Power Apps' viability for producing a stakeholder analysis definition metadata capture tool with capture, read, update and delete functionalities. Key areas discussed are ease of learning and adoption of the platform, integration and interaction with both internal and external databases, as well as challenges faced during proof-of-concept development. This user experience revealed Power Apps' potential for streamlining internal workflows, aiding maintenance of standards, and providing overviews of project performance metrics | excel, key performance indicator | source presentation source |
4 |
| 80 | Exploring the Strengths and Challenges of Neurodiversity Within the Pharmaceutical Industry | Melis Blackford | pharmaceutical industry, data science | source poster source |
2 | |
| 81 | Failed Clinical Trials: How to Benefit From Negative Experience | Anna Amosova | Failed clinical trials can cost sponsors more than a billion dollars, and they can waste years developing a drug that will never reach the market. There are many ways for things to go wrong. These could include lack of efficacy, safety concerns, insufficient funding, or patient recruitment, enrollment, and retention. This paper analyzes various reasons behind failed clinical trials, including internal and external factors contributing to trial outcomes. Also, we would like to highlight indicators that predict potential study difficulties or even failure, and if the decision of study closure was made, what are the next steps from the patient and study management team perspective? Furthermore, this paper raises the question of whether some proactive strategies and best practices can help mitigate these risks and increase the likelihood of future trial success | clinical trial, sample size, sample, interim analysis, clinical research, likelihood function, adverse events, phase i, adaptive design, data cleaning, primary endpoint, protocol deviations, futility analysis | source presentation source |
13 |
| 82 | FAIRification of preclinical data in radiation oncology through data standardization and formalized representation | Olga Giraldo, German Cancer |
In radiation oncology, the data generated from preclinical trials serve as initial validation for treatment effectiveness and optimizing clinical approaches by unraveling molecular mechanisms underlying different treatment responses. Therefore, it is impor tant to standardize the practice in managing preclinical trial data to ensure consistency and reproducibility across studies, promoting collaboration, and facilitating regulatory review. The primary goal of this work is to standardize and formalize data collected from preclinical radiation oncology studies to facilitate knowledge discovery. To achieve this goal, data is stored in relational databases and organized according to the CDISC SE ND standard, and formalized by utilizing ontologies and semantic web technologies. FAIRdifying preclinical data could enhance personalized medicine initiatives and drive innovation in therapeutic development | data cleaning, relational database, cdisc | source presentation source |
5 |
| 83 | FDA Business Rules and CDISC Open Rules: The Road to Adoption | Nick De Donder | data validation rules, clinical trial, cdisc | source presentation source |
4 | |
| 84 | Finding Out Subject Compliance Data in Clinical Trials | Bhakti Gadekar, Pravat Pradhan |
Compliance in clinical trials is often overlooked, leading to underestimated impacts of patient noncompliance. Improved analysis of patient adherence during drug testing could enhance trial conduct. Subjects are typically required to adhere to protocol visits, complete forms between visits, and report side effects. These activities gauge subject adherence, crucial for trial reports. However, missing data from incomplete visits hinders accurate safety and efficacy assessments, potentially compromising participant compliance. To address this issue, it's imperative to monitor and assess subject data compliance with the study protocol throughout the trial. Our presentation will delve into various methodologies for ongoing study assessment, including monitoring c ompliance with procedures such as CT scans, bone scans, and Patient-Reported Outcomes (PROs) | clinical trial, proc freq, subject visits, statistical programmer | source presentation source |
10 |
| 85 | Fostering excellence: What it takes to build a world-class team of R and Shiny Developers | Aga Rasinska, Vedha Viyash |
Building a top-tier team of developers is about more than just gathering skilled individuals. It's about fostering an environment that promotes growth, collaboration, and innovation. At Appsilon, we have a team that delivers cutting- edge solutions while also contributing to open-source projects. Effective leadership plays a key role in this by nurturing potential, building trust, and empowering individuals to take ownership of their development. This paper explores the key elements that drive team excellence, including career pathways, Individual Development Plans (IDPs), and the alignment of individual aspirations with the organization's strategic goals. It highlights the importance of clear communication and a balanced approach to personal well-being and performance. By promoting a culture of openness, continuous learning, and innovation, leaders can empower their teams to EXCEL and contribute meaningfully |
source presentation source |
4 | |
| 86 | Four Languages, Three Years: An Agile Approach to Data Science | Shannon Haughton | Presentation | data science, clinical programming | source | 11 |
| 87 | From Contributor to Leader: How Open Source Nurtures Innovative Leadership | Mark Bynens | pharmaceutical industry, clinical data | source presentation source |
9 | |
| 88 | From Senior Programmer to Head-of-Function: taking that 'leap-of faith' | Roxanne McGuire | clinical trial, clinical research | source presentation source |
5 | |
| 89 | From Source to Submission: Getting the Best of Multiple Standards | Jules Van der Zalm, Berber Snoeijer |
clinical trial, raw data, controlled terminology, data standard, data cleaning, cdisc | source presentation source |
4 | |
| 90 | Future Proof Your AI Investment | Alistair Dootson, Jon Gabriel |
source presentation source |
4 | ||
| 91 | Generative AI: A Clinical Programmer's Friend! | Karl Kennedy, Daniel Ghali |
This paper discusses the benefits of incorporating Generative AI into a clinical programmer's daily routine. It demonstrates these benefits through an R-based Shiny app that showcases various coding features that programmers can utilize by using Generative AI. The app illustrates how to use Generative AI natural language prompts to comment, test, optimize, debug, explain, and refactor R code. The use of these features can lead to increased productivity, efficiency, and quality in programming tasks while also reducing the learning curve for programmers new to R | clinical programming, laboratory data, dataset, missing value, exception handling, line chart, excel, data science | source presentation source |
10 |
| 92 | Genomic Biomarkers: Development of a Pipeline for Distilling Patient Mutation Status | Alexander Tchourbanov, Karl Brand |
proc freq, clinical trial, normalization | source presentation source |
9 | |
| 93 | Getting a Prompt Response Using ChatGPT and SAS | Stephen McCawille | This paper demonstrates the integration of ChatGPT, a state-of-the-art AI language model, with SAS 9.4 using Application Programming Interface (API) keys and PROC HTTP to automate and simplify common programming tasks. By utilising SAS Macros, prompts related to code optimisation, debugging, and explanation are converted into API requests, enabling programmers to receive immediate feedback within SAS. These prompts are condensed into single-word macro calls (e.g., debug, optimise, explain), with flexible options such as insert_error and insert_code for further customisation. This integration could potentially enhance productivity and reduce the tim e spent on routine tasks, leading to improved outcomes in clinical programming |
proc step, proc http, macro call, data cleaning, data step, proc report, error message, proc fcmp | source presentation source |
10 |
| 94 | Guidance Beyond the SDTM Implementation Guide | Kristin Kelly | A common misconception among preparers of SDTM data seems to be that it is sufficient to just follow the SDTM Implementation Guide when creating the datasets. The truth is that it is more complicated than that. A preparer of SDTM datasets needs to be aware of all the industry guidance available when preparing for regulatory submission, from CDISC and the regulatory agencies, but also from other organizations as well. This presentation will discuss some of the lesser-known guidance in the industry and why they should be referenced, as well as some of the impacts of not using these documents in the creation of SDTM datasets | sdtm domain, sdtmig, clinical trial, data cleaning, medical history, clinical endpoint, data validation rules, cdisc | source presentation source |
10 |
| 95 | Harmonising FDA and PMDA Submission Requirements | Luca Semeraro | Ensuring compliance with the electronic submission guidelines set forth by the Food and Drug Administration ( FDA) and the Pharmaceuticals and Medical Devices Agency (PMDA) for data submitted in a standardized CDISC format has become increasingly crucial. Currently, the task of submitting identical electronic data packages to both agencies poses certain challenges. However, understanding these regulatory requirements is essential for effectively planning, developing, and submitting study data deliverables. To shed light on this complex landscape, the paper highlights both the similarities and differences between the FDA and PMDA guidelines for CDISC -standardized data submission. Following this comparative overview, the paper offers a detailed analysis of specific PMDA requirements. Additionally, the paper includes examples from two case studies that demonstrate the practical application of these guidelines and the challenges faced during package preparation | data cleaning, adam dataset, data validation rules, clinical trial, clinical data, controlled terminology, sdtm domain, data standard, efficacy summary, vital signs, cdisc | source presentation source |
12 |
| 96 | Harmonization and Interoperability between OMOP CDM and CDISC SDTM | Dhruba Sikdar | The growing interest in observational data, a.k.a Real World Data (RWD) to assess safety, effectiveness and economic value of healthcare solutions have created some practical challenges of using this in research, whereby disparate data sources make it difficult and time consuming to compare and exchange between various systems. Conversely, the ever-increasing use of RWD in support of regulatory submissions has created similar challenges for the Health Authorities across the globe. This problem prompted the need for a common data model (CDM) to help researchers and to aid regulatory decision making. While Clinical Data Interchange Standards Consortium's (CDISC's) Study Data Tabulation Model (SDTM) is the de facto standard for submission to US FDA and many other major global regulatory authorities, it has its limitation while dealing with observational data | sdtm domain, adverse events, cdisc | source presentation source |
8 |
| 97 | Hi. I Am a Chatbot. How Can I Help You? Enhancing the Reliability and Traceability of Large Language Models | Vicky Poulsen, Anja Vrecer |
The capability of Large Language Models (LLMs) to process large volume of text, retrieve valuable information and generate human-readable text and their versatility have offered the pharmaceutical industry numerous new opportunities, from optimising existing processes to accelerating drug discovery and development. In Novo Nordisk (NN), we are exploring the possibilities of harnessing the power of LLMs in various areas. In this presentation, we will demonstrate our attempt in developing a chatbot that provides end-user support to the day-to- day operation of SDTM automation. Other than providing reliable SDTM related information based on internal and trusted external sources, our ambition is to build a chatbot that can act as first-level support in triaging SDTM generation issues. We will also share how we intend to overcome the known limitations of LLMs and incorporate transparency in the solution to boost confidence in its use | machine learning, neural network, cdisc | source presentation source |
9 |
| 98 | How Agile Can you Be? Looking at Statistical Programming as a Software Development Process | Oliver Wirtz, Peter Bonata |
Agile methods and practices have been around in software development for more than 20 years. Surprisingly, agile principles have not found their way into day-to-day statistical programming work. Instead, statistical programming, especially in clinical studies, is still very much relying on planning work towards fixed milestones, following finalized specifications. This paper will challenge this common practice and outline how an agile mindset can be established in teams using software development tools such as Azure DevOps and thus emphasizing team collaboration, incremental delivery, and continuous planning and learning with built-in quality | clinical programming, clinical trial, database lock, adam dataset, cdisc | source presentation source |
6 |
| 99 | How R Pharma? An Overview of How Our Industry Is Adopting the R Programming Language | Ross Farrugia, Ari Siggaard Knoph |
clinical trial, sdtm programming, cdisc | source presentation source |
5 | |
| 100 | How to Handle Huge Amounts of Output Tables - A Systematic Approach | Nico Schreiber | In the context of clinical programming, a key objective is the consolidation of all tables, listings and figures (TLF) into one or more comprehensive documents. Managing potentially hundreds or even thousands of output tables, particularly in the context of Health Technology Assessment (HTA) analysis, poses a considerable organizational challenge. To address this issue effectively, this paper presents a systematic approach to handle this huge amount of outputs in SAS 9.4. This methodology utilizes a set of macros designed to manage table metadata, facilitating the seamless and automatic creation of the final TLF documents. Although it was originally designed for a large number of outputs, it can also be easily adapted to a smaller number of outputs. Alongside the substantial time savings in the process, this systematic approach additionally reduces the risk of errors associated with manual execution, thereby amplifying the efficiency and reliability of clinical programming workflows | clinical trial, excel, output delivery system, data cleaning | source presentation source |
7 |
| 101 | How to Optimise Your SAS Code: Tips and Tricks | Vismaya Kondassam Veedu | The SAS language serves as a widely utilized statistical analysis programming tool. This paper aims to highlight various helpful tips and tricks that can streamline your programming experience. This paper will explore a collection of functions and statements designed to enhance coding efficiency, minimize runtime, and facilitate easier error debugging. It will also discuss methods for programming that facilitates seamless updates in the future, along with certain macro functions aimed at reducing the manual workload involved in producing RTF output | dataset, macro variable, excel, proc report, proc step, sas log, data wrangling, proc sort, data science, sas output, data cleaning | source presentation source |
16 |
| 102 | How to Use Regular Expression for Complex Searches in Data | Romain Gruber | Perl Regular Expressions are tools for pattern matching and string manipulation. They are essentially sequences of characters that form a pattern, useful for various tasks including searching, replacing, or splitting text. These expressions are particularly useful when dealing with data where exact values aren't known, but a specific pattern might be present. Perl Regular Expressions allow us to search for and manipulate such patterns. While SAS provides numerous functions for string manipulation, Perl Regular Expressions offer several benefits. For standard string processing tasks, such as searching for a specific substring, filtering alphabetical or numeric characters, or replacing one word with another, they can be more concise than conventional SAS functions. Moreover, they facilitate pattern matching and replacement, which may not be achievable with standard SAS functions. Perl Regular Expressions adhere to a specific syntax, using metacharacters to define patterns | sas function, character variable, prxmatch | source poster source |
5 |
| 103 | ICH M11 Clinical Electronic Structure Harmonised Protocol (CeSHarP) and CDISC: Making the Electronic Protocol a Reality | Peter Van Reusel | Presentation | phase iv, cdisc | source | 29 |
| 104 | ID-SPECTACKLE: Making ADaM specifications "Fun" | Sophie Dekeyser | ID-Spectackle is a user -friendly graphical interface to guide statistical programmers. This tool is meant to help and prepare the ADaM Specifications and create the Define-XML for submission. It helps to limit a lot of manual input by automating the metadata collection and provides guidance to fill in the specifications. As the basis of the specifications, a SAS Macro collects metadata, with support of the ADaM IG and NCI Controlled Terminology. This automated process presents multiple advantages: - it avoids error-prone copy-pasting of redundant information among sources, - it ensures that specifications and datasets are aligned, - it provides consistency checks with completion rules, - the evolution of standards can be implemented directly in the system, - it ensures that the specifications format is compliant with the one from Pinnacle 21 (P21), - it clarifies and deploys a standard process company wide | adam dataset, excel, data cleaning, data standard, cdisc | source presentation source |
11 |
| 105 | Implementing CDISC CORE in a Cloud-Native, Regulated Environment to Support Rare Diseases Submissions | Stijn Rogiers, Emmy V Pahmer |
The focus of our presentation is to share with the CDISC community our journey implementing the CDISC Open Rules Engine (CORE) within the SAS Life Science Analytics Framework (LSAF) environment at argenx. CORE is a free open-source tool for validating CDISC and regulatory conformance. LSAF is a closed system which provides a 21 CFR Part 11 compliant, web-based, Statistical Computing Environment (SCE) and Structured Data Repository where study data can be securely stored with versioning and audit history capabilities. Argenx is a global immunology company, committed to improving the lives of people suffering from severe autoimmune diseases. We will highlight challenges encountered and insights gained during the implementation process. We also plan to perform an evaluation of CORE functionality, running the rules for rare disease studies. We will also share potential plans for the future in supporting the development of the CORE engine | macro definition, life sciences, cdisc | source presentation source |
4 |
| 106 | Incorporating Longitudinal Co-Medication Data into Population Pharmacokinetic (POPPK) Analysis Datasets: A Programmer's Perspective | Aschwin van der Giesen, Gerly van der Vleuten |
This paper explores the integration of time-varying comedication data into a Population Pharmacokinetic (PopPK) dataset. PopPK analysis is critical for understanding variability in drug concentrations within a population receiving clinically relevant doses. The inclusion of comedications, especially those that influence drug metabolism, is essential for accurate modeling of pharmacokinetics (PK) in populations. This paper discusses the methodology for identifying relevant comedication data, addressing common challenges such as incomplete start- and end dates of comedication records, and incorporating these data into a PopPK dataset to improve the precision of drug interaction modeling | adam dataset, concomitant medications, population pk, clinical trial | source presentation source |
5 |
| 107 | Innovation of Causal Inferential Frameworks for Biostatistics in Real-World Data and Evidence | Anwesha Roy | causal inference, clinical trial, potential outcomes, sensitivity analysis, inferential statistics, residual, population, treatment effect, sample | source poster source |
8 | |
| 108 | Insights of Using {admiralonco} at BMS: Facilitating Oncology Efficacy ADaM Datasets | Yirong Cao, Yi Yan |
The R package {admiralonco} has facilitated the generation of the oncology ADaM datasets by providing essential tools and generic vignettes. This paper will introduce the practical experience on how BMS develops company specific efficacy vignettes and some utility functions for an oncology study. There are three main components: response analysis as per RECIST 1.1 (ADRS), time to event analysis (ADEFTTES) and summary of lesion / lesion changes (ADEFTM). The ecosystem also includes dependency ADaMs such as concomitant medication dataset (ADCM) to calculate subsequent therapy date for the enhanced definition of lesion changes analysis. The whole development is conducted with the BMS platform - DaVinci, integrating Posit Workbench (R programs development), GitHub (version control), Posit Connect (documentation) and package manager (storing the package for internal users) | missing value, adam dataset, best response, clinical trial, cdisc | source presentation source |
10 |
| 109 | Interactive reporting using FDA Standard Safety Shells and CDISC Analysis Results Standard | Megan Harries | Safety reporting is common across all clinical trials, and recent initiatives such as the FDA Standard Safety Shells and CDISC Analysis Results Standard offer the possibility of significantly reducing the time it takes to setup safety reporting at the start of a trial and make it easier to analyse safety signals across studies. It is possible to create a 'building-block' approach to implement the FDA Safety Shells which allows the outputs to be created and validated quickly while also supporting study-specific customisations. In addition to producing traditional static TFL, the outputs produce analysis results datasets based on the CDISC Analysis Results Standard (ARS), which can be used to produce hyperlinked outputs in HTML and interactive reporting in R Shiny | building block approach, system organ class, cdisc | source presentation source |
5 |
| 110 | Interactive Session (PM) - Exploring Effective Leadership Styles Through Interactive Storytelling | Caroline Phares, James McDermott |
source | 16 | ||
| 111 | Interactive Session - The State of Open Source Technology in Clinical Data Analysis, Reporting, and Submissions - A PHUSE Manuscript | Katja Glass, Michael S Rimler |
Presentation | data cleaning, clinical data, clinical trial, data science, data pipeline, csr, raw data | source | 27 |
| 112 | Interpreting Data Cutoff Methods Using R | Lusine Barseghyan, Vahagn Barseghyan |
To safeguard participants in clinical trials, regular safety profile analysis is essential, relying on precise data cut-off (DCO) management. The R programming language offers robust solutions for clinical programmers, featuring accessible functions and libraries. This paper proposes an efficient DCO implementation method using R. A dedicated package applies predefined DCO methods to specified datasets and variables, including date variables and those affected by DCO. Users can define preferred methods and variables via a customizable template or utilize separate package functions. This approach streamlines DCO execution, ensuring compliance and accelerating safety profile analysis, thus enhancing clinical trial management efficiency | clinical trial, excel | source presentation source |
4 |
| 113 | Introducing Graph Structure in Real-World Data Analyses | Georgios Panagiotakis, Maria Puri |
In the realm of pharmaceuticals, real-world data (RWD) comprises an expansive and intricate web of information drawn from diverse sources. These include electronic health records, insurance claims, and patient registries. Traditional analytical approaches often fall short when attempting to fully harness the potential of this interconnected data landscape. Graph theory, with its emphasis on relational structures, emerges as a promising framework to unlock the full value of RWD. However, the practical applications of this theoretical approach in the pharmaceutical sector remain largely unexplored, presenting both opportunities and challenges for r esearchers and industry professionals alike. Graph theory's strength lies in its unique ability to map out and examine complex relationships in health data. This approach can shed light on common health patterns and key intervention moments that often slip through the cracks of traditional analysis | relational database, structured data, clinical trial, insurance claims | source poster source |
10 |
| 114 | Introduction to the ADNCA ADaM | Charlotte Hampson, Florence Horspole |
Pharmacokinetics (PK) is an integral part of drug development as it is the process of analyzing the movement of drugs within the body. In order to describe, characterize, and quantify this effect, PK parameters are required to be calculated. This poster will provide an overview of the ADNCA (Data for Non-Compartmental Analysis) ADaM Dataset and how it is an effective way of generating a dataset that will contain the required data for calculating the PK Parameters. It will compare the ADNCA method to older processes used to aid PK Parameter derivation, and how it will improve efficiency during a study | adam dataset, key, in=, pk parameter, clinical research, compartmental analysis, cdisc | source poster source |
4 |
| 115 | It's Not the Destination, it's the Journey: Unlocking Growth and Realising Potential with Continuing Professional Development | Diana Stuart | Presentation | source | 10 | |
| 116 | It's not the X's and O's, it's the Jimmys and Joes | Eoin Byrne | Building a successful team can be demanding and we all face a similar challenge; Do we build our team based on a set structure and place team members into set roles, or do we fit the roles around our team members already in place? Leading a project team through this can be a tough balancing act. However, we can draw inspiration from the world of sport where such decisions are commonplace, none more obvious than American Football , w here Head Coaches bring in their own scheme and players. We all want the best players on our team, and we all know the way we want to play the game, however truly great Head Coaches find the marriage between both the game plan and their players' skill sets. Find the inner NFL Coach within you and begin to lead your "Star Players" and remember, your role doesn't end when the final whistle blows | final analysis, data governance | source presentation source |
6 |
| 117 | It's Not You; It's My Algorithm: How Prompt Engineering Can Save Your Relationship with AI | Josua Böser, Vanessa Kapsona |
As technology advances, Artificial Intelligence (AI) is becoming increasingly popular and helpful in various fields of work. However, it is essential to learn effective ways to interact with AI. One such skill is prompt engineering, which involves translating complex instructions into structured commands - or prompts - to guide the AI to achieve specific results. This paper explains the principles, versatility, and importance of prompt engineering. We will describe prompt techniques, provide guidelines for local experimentation with different AI models using Ollama, and explore the nuances of prompt hacking. We will also observe how this type of communication is remarkably similar to the communication dynamics between humans, making it a familiar concept. By following the principles of prompt engineering, we can improve the quality and reliability of our AI output. So let's start our journey of human-AI communication and prepare for the AI Revolution | proc freq, artificial intelligence | source presentation source |
12 |
| 118 | Keynote Speaker - Maneesh Juneja | Maneesh Juneja | Presentation | clinical programming, clinical trial, clinical trial protocol, data governance, data science, artificial intelligence, independent qc | source | 104 |
| 119 | Lessons Learned From the ADaM Data Development Using ADMIRAL | Katarzyna Golasa, Tomasz Lebitko |
Poster: Lessons learned from the ADaM data development using ADMIRAL Katarzyna Gołasa, Tomasz Lebitko , Late-stage Development, Respiratory and Immunology (R&I), BioPharmaceuticals R&D, AstraZeneca, Warsaw, Poland Workstream - AstraZeneca workstream is a part of the cross - Therapeutic Area R initiative. The project is focused on developing the ADaM datasets using the ADMIRAL package. It is the first time R programming was used to develop ADaM datasets. R language - a free software environment that provides a wide variety of statistical and graphical techniques. R can be extended via packages. Packages are the units of reproducible R code. They include reusable R functions, the documentation that describes how to use them, and sample data. The popular packages are: tidyverse, admiral, lubridate or diffdf | adam dataset, statistical programmer, cdisc | source | 1 |
| 120 | Leveraging 'Ensemble' Machine Learning Models to Drive Automation of Specification Generation and Data Mapping in Clinical Trial Data Submission | Venu Mallarapu | The submission of clinical trial data to regulatory authorities is a complex and time-consuming process, often requiring extensive manual effort to map data to standardized formats. Leveraging an ensemble of machine learning models for automated specification generation and data mapping can streamline this process. By analyzing existing study protocols, statistical analysis plans, and case report forms, these ensemble models can generate data specifications and mappings with high accuracy and minimal human intervention. The ensemble approach combines the strengths of multiple models, reducing errors and improving consistency across studies. This automation accelerates the data submission timeline while ensuring regulatory compliance. Furthermore, automated mapping facilitates the reuse of standardized data elements, promoting data interoperability and enabling cross-study analyses | clinical trial, clinical data, machine learning, response variable, data mapping, proc hpforest, proc hpsplit, data cleaning, support vector machine, regression analysis, logistic regression, raw data, sdtm domain, logistic distribution, neural network, data science, cdisc | source presentation source |
10 |
| 121 | Leveraging Bayesian Posterior Probability for Analysing Treatment Effects in Clinical Trials: A Statistical Programmer's Guide | Vismaya Kondassam Veedu | In clinical trial analysis, the Bayesian posterior probability, particularly with non -informative priors, emerges as a valuable tool for assessing the true underlying mean improvement between treatment and placebo groups. This method enables the evaluation of whether improvements exceed predefined thresholds, such as 0%, 10%, and 20%, leveraging LS means and estimated standard deviations derived from the Mixed Models for Rep eated Measures (MMRM). This approach proves especially beneficial when sample sizes are insufficient for detecting significant differences through traditional MMRM analysis. Utilizing an empirical approach, Bayesian posterior probability is computed using PROC BGLIMM, which generates posterior samples to determine the probabilities of improvement. This paper provides comprehensive insights into this methodology |
posterior probability, bayesian statistics, proc step, likelihood function, sample, clinical trial, prior probability, treatment effect, covariance, change from baseline, covariance structure, proc mixed, statistical programmer, proc genmod | source presentation source |
7 |
| 122 | Leveraging Data Science and Analytics to Optimise Clinical Trial Execution | Melanie Hullings | clinical trial, data science, clinical data, machine learning, clinical research, data governance, data cleaning, clinical data management | source presentation source |
4 | |
| 123 | Leveraging Open Data and Machine Learning Enhanced Data Extraction for Clinical Trials Insight | Hugo Bottois | Open databases such as ClinicalTrials.gov provide valuable information about clinical trials for specific diseases or drugs, including data from protocol and study results. However, some information is not always directly provided in a structured format in protocol descriptions (dosage, frequency, and route of drugs) or in results (type of group). These limitations raise challenges to extract information and fully leverage open data for clinical projects. Machine learning tools like Med7 and BioBERT allow extraction and mapping of this information, making exploitation of data easier over a larger number of clinical trials and reducing human labor. These processes allow multiple use cases such as extracting adverse event data for meta -analytic risk modeling, gathering data to produce tables, or providing a landscape of clinical trials over time with a data visualization dashboard | clinical trial, proc freq, adverse events | source presentation source |
9 |
| 124 | Leveraging Shiny Modules for Modular and Scalable Web Applications for Clinical Trials/Research | Sohan Lal, Pinku Hooda |
R Shiny offers a dynamic platform for developing interactive and customizable tools tailored to specific study needs but shiny modules revolutionize web application development by offering a modular and scalable approach to within the shiny framework. These encapsulated units of code enhance code reusability, maintainability, and collaboration among developers. Modules facilitate the division of complex applications into manageable components and enable easier integration of new features. This abstract presents an overview of Shiny modules, highlighting their benefits in promoting modularity, scalability, and code quality. By embracing modular programming practices in shiny development, developers can streamline their workflows, improve code organization, and deliver high quality applications that meet the needs of users and stakeholders effectively | clinical trial, box plot, histogram, unique key, clinical data, adverse events, clinical research, data science | source presentation source |
10 |
| 125 | Leveraging the USDM Standard in OpenStudyBuilder for Enhanced Data Exchange | Marius Conjeaud | The OpenStudyBuilder (OSB) Metadata Repository (MDR) and Study Metadata Repository (SDR) serve as pivotal components in Novo Nordisk's comprehensive approach to digitizing the entire clinical process. This encompassing digital transformation commences with the Protocol and extends to the Trial Files and Listings (TFL), facilitated by the Schedule of Activity. The OSB leverages these repositories to drive the production of the Case Report Form and specify the Tabulation of Metadata. Central to this process is the Activity concept, which orchestrates the interconnection of metadata elements through a Graph database provided by Neo4j. A recent milestone achieved by the OSB is the capability to extract Protocol metadata within the Unified Study Definition Model (USDM), enabling the generation of a Novo Nordisk protocol but at the same time, an ICH - M11 one, by leveraging shared metadata | clinical research, crf | source presentation source |
7 |
| 126 | MAICing Efficient, Flexible and Maintainable Code with Modular Programming | Michael Allie | The concept of modular programming is a simple one. It involves organising the code into self-contained programs, known as modules, that perform a specific sub-task. These modules can be thought of as building blocks that collectively form a larger structure, or as the individual instruments in an orchestra. They can be arranged, reused, or exchanged as required. While each module is simple in isolation, they can accomplish complex tasks when cleverly combined. A Matching Adjusted Indirect Comparison (MAIC) analysis provides an excellent example of the advantages of a modular approach. The overall process can be broken down into a list of tasks, each of which will vary in predictable ways between different analyses. This paper uses MAIC to demonstrate modular programming design and implementation. Additionally, advanced SAS macro techniques are also highlighted for developing efficient, flexible, and maintainable code | treatment effect, treatment arm | source presentation source |
10 |
| 127 | Making SDTM/ADaM CRT Packages Submission Ready | Phillip Spowart | When submitting a compound, Regulatory bodies require SDTM and ADaM datasets to be an IG version that is accepted as per their standards catalogue. With big compounds we have seen studies where datasets are not in an SDTM/ADaM format or the SDTM/ADaM IG is in a version no longer accepted. This paper aims to cover how to convert Legacy datasets and datasets in an older version of IG into an accepted version of IG. This will cover how the SDTM and ADaM datasets become submission ready CRT packages, which can then be submitted and used to create an integrated database | adam dataset, sdtm domain, clinical trial, raw data, data cleaning, cdisc | source presentation source |
10 |
| 128 | Mapping a Veeva SDS Specification to ODM-XML | Jørgen Mangor Iversen | clinical data, data validation rules, excel, cdisc | source presentation source |
7 | |
| 129 | Mastering Pinnacle 21 for Oncology SDTM | Daryna Khololovych | study day, char., missing value, controlled terminology, sdtm programming, cdisc | source presentation source |
7 | |
| 130 | Mastering Production-Grade Shiny Apps with Rhino: Lessons and Insights | André Veríssimo, Vedha Viyash |
Buildingashinyappisagreatexperienceandaneasywaytoquicklyshareyourworkwithothers.However,whenitistimetodeployittoproductionitisneitherfastnoreasy. AtAppsilon,wefocusonbuildingproduction-readyshinyappsandhavelearnedalotoflessonsalongtheway.Wecreated{rhino}tostandardizetheprocessofbuildingproduction-gradeshinyappsanddecidedtomakeitopensource. Fromtheget-go,itintroducesthebestsoftwaredevelopmentpracticeslikemanagingthedependenciesandsettingupdifferentlevelsoftestingforyourapplication. Inthistalk,wewilldiscussthelessonswelearnedandhowtothinkaboutappsbuilttolastandeasytomaintain | source presentation source |
7 | |
| 131 | Maximising GitHub APIs: A Journey to Open-Source Metrics | Alanah Jonas | Over recent years, many pharma companies have been adopting open -source technologies. But how can we effectively measure our progress on this journey? At GSK we have been using GitHub for code development in our study reporting. This p aper will explore the innovative utilisation of GitHub APIs as a powerful tool for extracting information from GitHub to enable us to track and analyse open -source efforts. By harnessing the capabilities of GitHub APIs, we can gain invaluable insights into ope n-source language usage in Biostatistics as a whole and by disease area. Attendees will discover how GSK extracts pertinent information from G itHub repositories to generate comprehensive language usage metrics. By leveraging GitHub APIs, we empower our decision -makers with actionable insights to make changes to drive open - source adoption more effectively | rest api, pharmaceutical industry | source presentation source |
10 |
| 132 | Medidata RAVE and Veeva CDMS Data Extraction Macros | Craig Mistal, Nicola Tambascia |
Web-based study data extractions allow for the immediate availability of clinical data for review, transformation, and analysis by Data Managers, Programmers, and Biostatisticians, respectively. The paper shows a technical overview and background of the SAS macros we created and validated to extract live EDC data from Medidata RAVE via the Rave Web Services (RWS) and Veeva CDMS via the Veeva CDMS API. The process around the APIs (e.g. access needed in the systems), and the advantages of using these methods of data extraction over procedures which utilize other scheduling methods, and requiring subsequent retrieval of extracted data from FTP sites is also detailed | stored macro, dataset, data cleaning, macro catalog, audit trail, clinical data | source presentation source |
4 |
| 133 | Meta-Analysis on Aggregate Data and Individual Patient Data Using R | Yen Phan | Poster: Heart failure (HF) is a global health concern characterized by symptoms like fatigue, breathlessness, and fluid retention, placing significant strain on individuals and healthcare systems. Various diuretics, such as loop diuretics (e.g., furosemide, torsemide, bumetanide) and thiazides, are used to manage fluid overload, but their impact on weight loss varies widely across patients. This study combines aggregated data (AD) and individual participant data (IPD) from five clinical trials to explore whether loop diuretics differ significantly from thiazide diuretics in promoting weight loss in HF patients. Using R, a meta-analysis was performed to account for differences in age, dosage, and sex, aiming to uncover insights into the effectiveness and variability of diuretic treatments. This poster is based on research using data from data contributors Bayer, Boehringer Ingelheim, and GlaxoSmithKline that has been made available through Vivli, Inc | source | 1 | |
| 134 | Multilingual Programming Language (MPL): A Paradigm Shift in Clinical Programming | Pavan Kumar KP | clinical programming | source presentation source |
2 | |
| 135 | Navigate Clinical Study Report Tables and Listings - The iTable & iListing Project | Katja Glass | clinical trial, csr, missing value, adverse events | source presentation source |
13 | |
| 136 | Navigating and Building Bridges: A Framework for Clinical Programming Using Open-Source Tools | Yash Ankurbhai Shah, Kavitha Allala |
The rapid evolution of open-source tools and packages presents both opportunities and challenges for clinical programmers. While these resources offer potential for many efficient and interactive outputs, navigating this landscape can be overwhelming, particularly for those transitioning to R programming and open source. To address this, we developed a generalized framework using the well-supported open-source packages Tplyr, Reactable, and Shiny. This framework allows programmers to create summary tables and listings and pass them as arguments and generate clickable tables which offers traceability back to the source data with some added features, without further additions. This enables programmers to focus on their core tasks which is to build and define tables rather than the intricacies of software engineering. Designed to be loosely coupled, the framework is intended to be adaptable to various outputs and environments and provide long-term value | clinical trial, clinical data, raw data, adverse events | source presentation source |
8 |
| 137 | Navigating First Data Submission: Pierre Fabre's Journey and Insights into Data Submission Challenges with the NMPA | Romain Raymond, Benoit Sansas |
Pierre Fabre conducted clinical trials in mainland China to support submission dossiers in 2024. These were the company's first National Medical Products Administration (NMPA) data submissions. The first data submission to the NMPA is not an easy and straightforward journey as guidance is only available in Chinese, and finding how to best fit to NMPA requirements can be challenging as it requires: getting the database translated into Chinese, submitting executable programming codes, and choosing the best submission strategy approach. Communication with the sponsor Chinese team and the authorities, and support from Contract Research Organization (CRO) local teams, notably for the regulatory and programming aspects, were very important in preparing this data submission. This paper outlines challenges met during the data submission preparation, the options we considered, and shares feedback on the solution we ended up choosing | clinical data, clinical trial, label statement, crf, data science, adam dataset, cdisc | source presentation source |
7 |
| 138 | Navigating Personal and Continuing Professional Development - A Journey into Data Science | Jawan Thapa | data science, statistical programmer | source poster source |
2 | |
| 139 | Navigating the Clinical Research Landscape: The Role of Regulatory Intelligence and LLMs | Federica Citterio, Soundarya Palanisamy |
clinical research, clinical trial | source presentation source |
9 | |
| 140 | Navigating the Data Storm: Strategic Mastery for Sustaining Technical Excellence Amid Industry Shifts | Mahdi About | data science, clinical trial, pharmaceutical industry, proc sql, machine learning | source presentation source |
8 | |
| 141 | OAK Productivity Experiment - How We Made SDTM Programming 5x Faster | Amy Franklin, Magdalena Krochmal |
The OAK Garden initiative aims to revolutionize the process of creating SDTM (Study Data Tabulation Model) deliverables by automating up to 80% of the tasks. This paper presents the results of a productivity experiment comparing our innovative R-based approach - OAK with the traditional SAS-based approach - DTS (Data Transformation Specification). The experiment focuses on how the use of the MINT+ Web Interface, the {roak} R package and adherence to global data standards impact productivity and efficiency within study teams. This paper presents findings on the current state of OAK, productivity gains, and recommendations for future resource allocation and skill development. The findings indicate that OAK can make SDTM programming up to 3-5 times faster, significantly enhancing efficiency and reducing manual effort | sdtm programmer, sdtm programming, sdtm domain, data standard, clinical trial, clinical data, cdisc | source presentation source |
5 |
| 142 | Optimising Workflow and Collaboration in a Pooled Coding Environment for Clinical Development | Louise Noergaard | Managing multiple phase 3 programs in parallel is challenging and requires considerable time and dedicated resources. The conventional approach of working within study "silos" often requires substantial post-conduct effort to align data packages for submissions. We present an innovative approach implemented by Novo Nordisk to manage multiple phase 3 programs by working in a pooled coding environment. We describe how using R and Git allows us to efficiently handle both generic and study-specific coding in one place during development. The code is then later "sliced" into individual study containers, ensuring that technical programming decisions are consistent across studies. We also present our implementation of Azure DevOps to ease collaboration, distribute tasks, track progress and much more. Finally, we describe our implementation of G it for continuous version control through pull -requests and automated code-functionality checks | phase iii, adam dataset, cdisc | source presentation source |
8 |
| 143 | Our Road to Adopting CORE: an Implementation Journey in SDTM Dataset Validation | Marisa Wyckmans | The Secure Data team, an unblinded team within SGS Data Management, daily handles numerous data transfers. Particularly during critical periods, delivery of SDTM compliant datasets is defined by a tight turnaround time. Consequently, the team relies on tools that can convert and validate high quality datasets efficiently. We are enhancing our dataset validation tool and combine this with our experience and knowledge from the CDISC Open Rules project. This presentation illustrates the requirements for our ne w dataset validation tool, including the transition to the cloud, and aligning them with the capabilities of CDISC Open Rules Engine ( CORE). Moreover, we showcase that we can create our own rules using the CORE Rule Developer and discuss the challenges we faced. We are also outlining and finetuning our own SGS pathway for future requirements/implementations | data cleaning, sdtm domain, cdisc | source presentation source |
7 |
| 144 | P_MACRO, Parameters Extraction From Macros to SAS Dataset | Eswara Satyanarayana Gunisetti | Macro parameters are variables whose values initialized when we invoke the macro. Parameters give macros flexibility, and including PARMBUFF option gives more flexibility by allowing a variable number of parameters. In this paper w e learn how to extract macro parameters, along with their default values from a group of macros in a folder, to a SAS dataset, to easily view all parameters in one location. With the P_MACRO resultant dataset, it is easy to automate the macro call by including all defined macro parameters. We will cover the importance of standard programming practices to ease the extraction of macro parameters, and approaches to follow if nested macros are present or if no parameters are defined. We will discuss how we retain the macro parameter order as it was defined within the macro while saving to a dataset. We will also discuss use cases to see how this could help our programmers | macro call, macro parameter, dataset, macro variable, macro definition | source presentation source |
6 |
| 145 | Perl Regular Expressions in SDTM and ADaM Programming: Live Study Examples | Pavol Valovic | In clinical research, free text fields are often used when the clinical site staff wants to include important information regarding adverse events, concomitant medications, or any other result deemed important for the investigator. Extracting useful data from free text fields is often difficult, however, Perl Regular Expressions (PRX) present a unique solution to this problem. PRX are a powerful tool for determining patterns of text within a string. SAS supports the usage of PRX and provides a great opportunity to tackle day-to-day tasks in determining data patterns in free text fields. Many papers have already been published on the general topics of PRX. In our submission, we present actual examples of PRX in live studies and discuss cases in which the usage of very complex PRX should be used cautiously and other programming techniques might be more suitable | adverse events, protocol deviations, concomitant medications, clinical trial, medical history | source presentation source |
7 |
| 146 | Pre ADSL - Revolutionise Your ADaM Workflow | Korak Datta, Taniya Chatterjee |
PRE-ADSL, an ADaM OTHER class dataset introduced in ADaMIG v1.2, is a convenient location for Ana lysis Results or Flags that may be used in multiple derivations across ADaMs. We plan to provide real-life examples of PRE-ADSL from our real-life experiences. While the ADaM ADSL dataset is usually the first ADaM to be created, it often contains several variables that are dependent on complex derivations driven by values or flags from other ADaM datasets. This creates a Circular Dependency problem, tricky to manage in the programming run -cycles as well as challenging to document in Specification documents. Also in certain situations, derivations in ADSL may need row-level data-point traceability from SDTM to ADaM which does not align with the ADSL structure. We hope to convince you that using PRE-ADSL can help to de-duplicate effort, derivations, and streamline ADaM programming workflow by centralising programming logic | adam dataset, treatment arm, adverse events, medical history, cdisc | source presentation source |
8 |
| 147 | Principal Component Analysis (PCA) - An Efficient Way of Variable Reduction Method and its Application in Oncology Trials | Pradeep Acharya, Ephicacy Lifescience |
In clinical studies, it is not feasible to use all variables in final analysis model even though they have significant influence on the outcome variable. The independent variables are usually correlated to each other which can introduce multicollinearity when the greater number of variables are included in the model. Application of Principal Component Analysis (PCA) on these variables can help solve such problems. PCA is a statistical method used to reduce the dimensionality of large data sets, by transforming a large set of variables into a smaller one that still contains most of the information in the large set which can then be used for final analysis. This paper illustrates how to apply PCA using SAS to reduce the dimension of oncology data without information loss. A clinical dataset of higher dimension has been used for the demonstration and the efficiency of PCA in reducing the variable dimension has been discussed | proc princomp, covariance | source presentation source |
7 |
| 148 | Py R You SASing Me: Converting SAS Dataset Filters into R and Python | Matt Metherell | If looking to code SDTMs in R or python, we may still receive sponsor standards that use SAS as their example code for filtering the input datasets, so in this paper I present R code to convert these filters into R, and python code to convert them into python. This can also be used as a ready reckoner for a SAS programmer taking their first steps in R or Python | missing value, missing function | source poster source |
5 |
| 149 | Quality Tolerance Limits: Understanding the added value and statistical need of implementing QTLs to assure subject safety and data quality | Nathalie van Borrendam | secondary endpoint, data cleaning, data science | source presentation source |
3 | |
| 150 | R and SAS: How They Can Co-Exist. Approaches and Considerations in Adopting Multiple Languages in Clinical Trial Analysis and Reporting | Bill Qubeck, Rakesh Gotiwale |
Poster: Training and Expertise Adapting to R requires retraining existing staff or hiring R-proficient talent. Upskilling legacy staff is key for a smooth transition. Offering internal training and/or working with external experts can help ease resistance and ensure effective integration. Migration costs and library maintenance Fully replacing SAS with R involves significant labor and time, especially when duplicating SAS libraries in R. This leads to higher costs as both libraries must be managed, requiring duplicate efforts for code changes, validation, and maintenance. A more cost-effective approach is to adopt a coexistence model, where SAS is retained for standardized tasks while R is gradually introduced for specific tasks and analyses. This minimizes disruption, reduces duplicative efforts, and allows R to enhance visualization and complex modeling alongside SAS | source | 1 | |
| 151 | R Evolution: A Guide for Training SAS Programmers | Michael Walshe | In the changing landscape of pharmaceutical programming, many companies are exploring R alongside SAS for its open-source flexibility and broad community. However, some SAS programmers find it difficult to transition to using R, and it often falls to the next generation of statistical programmers to foster and encourage the use of R in their company. Based on practical experience designing and delivering R training to SAS programmers, this presentation dives into how to train SAS programmers in R, highlighting effective methodologies, resources, and best practices tailored for those familiar with SAS. It outlines common pitfalls and challenges that SAS programmers may encounter when learning R, such as: different modes of thinking about data; dealing with missing values; changes in numerical methods; and dealing with larger -than-memory data | sas programmer, missing value, proc step, pharmaceutical industry, data science, proc sql | source presentation source |
10 |
| 152 | R Package Validation Maintenance - We Didn't Know What We Didn't Know | Mark Sellors | data cleaning, clinical programming, r package | source presentation source |
4 | |
| 153 | R Scripts for Streamlining Non-Compartmental Analysis (NCA): A Data Programmer's Dream | Valentin Legras | Non-compartmental analysis (NCA) plays a critical role in early-stage drug development, offering streamlined pharmacokinetic (PK) parameter estimation without requiring complex assumptions. The traditional workflow, although widely accepted by regulatory agencies for bioequivalence studies, is often time -consuming and reliant on manual processes. This paper presents an automated, end -to-end solution to streamline the NCA process through the development and implementation of R scripts. The solution, named automated NCA (aNCA), seeks to significantly reduce the time and resources required for analysis while maintaining quality and compliance. By joining the Pharmaverse, an open-source collaborative network, aNCA aims to improve efficiency, reduce dependency on external vendors, and enhance data accessibility and anal ysis control. This paper outlines the current landscape of NCA, the proposed automated workflow, and the future outlook for widespread implementation | clinical research, compartmental analysis | source presentation source |
2 |
| 154 | R Shiny: From Code to Cloud | Abu Sami | R Shiny is a commonly used tool for transforming clinical trial analysis from static to dynamic, which is nowadays of high interest to stakeholders. However, the development and deployment of a complex R Shiny app can be challenging. Following good software development practices in a collaborative effort is essential. Additionally, cloud technologies are popular for managing the app lifecycle. The advantages include improved collaboration, excellent reproducibility, accessibility, and enhanced security. In the presentation, we will explore the design of a Shiny app with modules that can flexibly be added or updated and demonstrate how to create comprehensive documentation. Furthermore, we will present a collaborative process for developing an app in a cloud environment, using a CI/CD (Continuous Integration and Continuous Deployment) pipeline with Azure DevOps which facilitates automatic linting, testing, R Markdown knitting, and deploying | clinical trial, data visualization | source presentation source |
12 |
| 155 | Resolving algorithmic SMQs - using AI chatbot and other techniques | Corentin Le Scanff | This paper focuses on how to process algorithmic Standardized MedDRA Queries (SMQs). It examines how algorithms can generally be resolved, the issues specific to SMQ algorithms, programming techniques and efficiency. Using the example of SMQs, methods and platforms for resolving algorithms are examined as well as AI chatbo ts capacity to understand the problem. SMQ algorithms need events to happen together and determining "together" is potentially complex. How do we consider events as overlapping? Do we consider only start dates or start to end intervals? Are adjacent events allowable, if so, what is the allowed gap? We present a solution that reverses the problem. Instead of looking at adverse events in relation to each other, we examine patients to see if the SMQ is satisfied. We show other techniques including using hash tables and custom functions | hash table, create table, adverse events, proc sql, data step, proc step, sas programmer, if-then-else, proc sort | source presentation source |
21 |
| 156 | Responsible AI in Protocol Design for Use in Digital Data Flow | Alistair Dootson, Poorya Amini |
model deployment, protocol amendment | source presentation source |
4 | |
| 157 | Return On Investment as a Financial Feng-Shui in Data Science. Practical Strategies to Get Projects Accepted and Evaluated | Iwona Matyjaszek | data science, likelihood function | source presentation source |
2 | |
| 158 | Revolutionising Literature Reviews with a GPT-Assisted Tool: Enhanced Search, Extraction and Visualisation | Yingjia Chen, Martin Schaerer |
Conducting literature reviews in drug safety and pharmacovigilance is resource -intensive, requiring manual search, selection, and data extraction, which can lead to inefficiencies and subjective variations. This study presents a GPT - assisted tool that auto mates key aspects of the process, improving efficiency and reliability through automated data extraction, summarization, and visualization. The tool, implemented in Python as a Streamlit application using the GPT-4o model from OpenAI, enables customized literature queries tailored to specific research questions in any language. It leverages GPT to categorize papers into "included" and "excluded" groups based on predefined research criteria. In this study, we focus on papers discussing drug-event associations for immune checkpoint inhibitors (ICIs) and immune -related hepatitis. Papers positively identified for inclusion undergo further analysis to extract key measures of association (e.g., ROR, RR, HR) | forest plot, adverse events, extract, exposure, odds ratio, web service | source presentation source |
12 |
| 159 | Revolutionising Statistical Programming Activities - StatProg CoPilot Consortium Leveraging on Open-Source World and Generative AI | Sébastien Tourlet | Capgemini invites pharmaceutical organizations to join a consortium aimed at advancing the StatProg CoPilot. An open-source solution designed to revolutionize statistical programming activities. Building on both Capgemini's R&D work and experience from projects with pharma companies, and on the success of open -source tools (e.g. OpenStudyBuilder, R). The StatProg CoPilot utilizes Generative AI to automate and enhance SAS, R, and Python code generation for ADAM and TFL transformations, accelerating clinical trial analysis and ensuring regulatory compliance. By sharing expertise and resources, members will co-create a tool which ensures broad accessibility and adaptation. The consortium will develop a standardized framework promoting data consistency, increasing code quality, reducing costs, and speeds drug development. We seek for partners c ommitted to transforming clinical data analysis through advanced technologies and collaborative innovation | clinical programming, clinical trial, clinical research, data science, unstructured data, statistical programmer, data cleaning, cdisc | source presentation source |
4 |
| 160 | SAIL.R: A Journey Towards R Software at Johnson & Johnson - A Training Perspective | Yannick Vandendijck | clinical data, statistical programmer, data visualization, line chart, data science, adam dataset, clinical trial, cdisc | source presentation source |
10 | |
| 161 | SAS Custom Step & R-Runner: Develop any application you can dream about | Olivier Bouchard | For a long time, people has been thinking SAS as a "One-code" shop; trusting the code, its flexibility and its reliability. With newest technology, this is time for NoCode/LowCode approach, OR time for "Code when you want" type of thinking. SAS VIYA now offers not only to coders, but to any data citizen person, unlimited ways of creating outcomes within SAS and building business application. If you combine the flexibility of Custom Steps and the SAS VIYA native integration with OpenSource (R, Python) your team can now build any kind of End To End application to deliver quick value solutions. We will show in this presentation how to build "application" that you can easy parametrize to get expect behaviors | sas viya, sas studio, data science, machine learning | source presentation source |
12 |
| 162 | SCE Alchemy: The Return of the QC Process, and Rise of Training via LLM | Steve Griffiths, Eleanor Sparling |
independent qc, quality control, staging area, data science, csr, proof of concept, data integrity | source presentation source |
6 | |
| 163 | Scheduling on FHIR - An Update on the PHUSE/Vulcan Schedule of Activities Working Group | Geoff Low | Poster | source | 1 | |
| 164 | Serverless Shiny - Evolving Landscape of Clinical Analytics | Philip Bowsher, George Stagg |
data science | source presentation source |
5 | |
| 165 | Shiny, FAIR and Compliant? | Shivoy Pandita, Jörn Bilow |
A unified platform for clinical data analytics facilitates creating regulatory-mandated submission datasets and performing exploratory analysis using access-controlled workflows. Such a platform should also foster collaboration, strengthen data protection and accelerate insights. This presentation highlights the tight integration of R and Posit with the Statistical Computing Environment entimICE FastTrack and shows, how stakeholders can benefit from running an R Shiny app directly from a regulatory-compliant Statistical Computing Environment, while maintaining data integrity. The integration capabilities of entimICE FastTrack facilitate exploratory analysis, and streamline workflows to get insights on clinical data, while keeping both FAIR principles and regulatory compliance intact. Programmers can create and run R Shiny apps within the advanced computing environment and share their code and the results without violating any compliance rules | clinical data, clinical data management | source presentation source |
2 |
| 166 | SLURM + Open-Source Clinical Workflows: Current State | Philip Bowsher, Michael Mayer |
clinical trial, data cleaning | source presentation source |
4 | |
| 167 | Standards Implementation with a Knowledge-Driven, AI-Powered Platform | Roger Gagali Angoh | data cleaning, clinical data, clinical research, data standard, artificial intelligence | source presentation source |
3 | |
| 168 | Starting a new Functional Service Provider relationship may be daunting and presents many challenges: Where do we start?! | Nicky Lawrence | Many pharmaceutical companies are now working with CROs in a Functional Service Provider (FSP) relationship - this may be across several services, e.g. Clinical, Biometrics, Safety, Medical Writing, or perhaps only Statistical Programming. We will concentrate on FSPs where study work is assigned to and managed by the CRO. Once the agreement is in place and first study(s) assigned - how do we operationalise the plans? We will look at the importance of strong communication and clarity of responsibility and roles. Both teams will have their own standard processes - how do we find one process going forward? What are the expectations in terms of study delivery requirements, overall performance and metrics? How do we keep learning and improving as we go? These are just some of the considerations in a process that needs to be set-up for short term delivery and long-term partnership to ensure maximum performance for the sponsor | clinical programming, service provider | source presentation source |
5 |
| 169 | Status and News in OpenStudyBuilder | Mikkel Traun, Martin Gram |
Poster | source | 1 | |
| 170 | Stop Chatting, Start Scanning: AI's Visual Takeover in Pharma for Data Extraction with the Power of LLMs | Ercan Suekuer | Therapidadvancementofartificialintelligence(AI)andlargelanguagemodels(LLMs)hasusheredinaneweraofefficiencyin pharmaceuticalandclinicaldatascience.AI-driventechnologies,particularlylargemultimodalmodels(LMMs),arenowcapableofinterpretingandextractingvaluableinsightsfromcomplexdatasets,includingimages,withunprecedentedaccuracy. Thispaperexploresthetransformativeimpactof thesemodelsondataextraction,pharmacometricmodeling,anddrugdevelopmentprocesses.Demosof AIbotapplicationsshowcasehowthesemodelscanaccelerateworkflowsandsupportinformeddecision-making.As thesetechnologiesevolve,theircapabilitiestomanagesophisticatedtaskswillexpand,makingthemindispensableinclinicalsettings | clinical trial, clinical data, extract, clinical research, pharmaceutical industry, data science | source presentation source |
3 |
| 171 | Strategies to Improve Understanding of Estimands and Analysis Section of Statistical Analysis Plan | Inderjeet Arora | Typically, a Statistical Analysis Plan (SAP) comprehensively delineates data analysis and data handling strategies for a clinical study. The Biometrics team relies on details and instructions provided in SAP to generate essential outputs such as tables, listings, and figures (TLFs). Hence, it's imperative to draft SAP in a manner that facilitates understanding and adherence, ensuring seamless data handling and analysis. Misinterpretation of details from SAP could result in misapplication of rules, necessitating rework leading to missed deadlines and impacting overall timeline of the study deliverables. This paper will focus on the estimands and analysis sections (includes four attributes, the population, variable, intercurrent events, and population-level summary) of SAP, addressing strategies to prevent confusion and conflicts in statements | data science, secondary endpoint, adverse events, intercurrent event, treatment effect, proc means, missing value, analysis set, clinical trial, primary endpoint, imputation, randomization | source presentation source |
8 |
| 172 | Streamlining Complex Analysis: An SAS Macro for Marginal Structural Modelling | Amine Khemiri | In the intricate world of clinical trials, understanding the true impact of treatments on outcomes is crucial. Unlike conventional approaches, Marginal Structural Model (MSM) doesn't shy away from the challenge of time-varying confounders, offering a nuanced understanding of causal relationships. To streamline MSM analyses, we developed a SAS macro "Model_MSM.sas" tailored to handle complex data structures commonly encountered in clinical trials. It adeptly navigates through the maze of data structures inherent to longitudinal studies, ensuring seamless manipulation and interpretation of data. This macro facilitates the implementation of MSM methodology, offering researchers user-friendly way for data manipulation, model estimation, and result visualization. Through this macro, researchers can efficiently conduct MSM analyses, enhancing the interpretability and reliability of findings in clinical research | missing value, clinical trial, repeated measures, treatment effect, macro definition, model training, repeated measures anova, longitudinal data, response variable, regression analysis | source presentation source |
6 |
| 173 | Streamlining MHRA Inspection Preparation and Execution in CROs | Nicolas Rouillé | This paper delves into the meticulous process of readying for and conducting a Medicines and Healthcare products Regulatory Agency (MHRA) inspection at Cytel, focusing on logistics, personnel management, and document handling. Core elements include strategies for preparation and execution, such as preparing interviewees, staff training, managing concurrent sessions, and efficiently handling internal and inspector information flows. A robust framework with defined roles, live demonstrations, and thorough explanations fosters productive dialogue during inspections, aiming to cultivate collaborative engagement with MHRA inspectors to demonstrate regulatory readiness and integrity. Post-inspection discussions have spurred innovative solutions for ongoing enhancement. This paper offers pragmatic insights for navigating MHRA inspections adeptly, stressing the importance of thorough preparation strategies and execution precision | clinical trial, liquidity gap | source presentation source |
6 |
| 174 | Study Process Optimisation | Andrea Parodi | Poster | source | 1 | |
| 175 | Synthesised Real-world Data: JSON and the Argonauts | Anil Sekhari | When it comes to developing products and tools for life sciences organisations, it is difficult for tech-based companies to acquire clinical data. An alternative is to use synthesized real world data (RWD). This paper demonstrates how to create CDISC SDTM domains from synthesized RWD. The journey starts with individual unstructured data-JSON files (in HL7 FHIR data exchange format) and how this data is first transformed to the OHDSI OMOP domains (common data model v5.4 data standard). The final stage in the journey is a second transformation from OMOP domains into SDTM domains. All the transformations are performed using SAS and this paper highlights the obstacles encountered and how to overcome them. The hope is that this will serve as a useful paper to begin your own journey into using synthesized RWD in the development of products and tools | sdtm domain, proc sql, data standard, life sciences, cdisc | source presentation source |
10 |
| 176 | Tailoring Git for SAS Projects: Overcoming Challenges, Achieving Excellence | Rajwanur Rahman, Spencer Renyard |
Implementing version control in SAS projects presents unique challenges compared to other development projects, like web. However, our organization has successfully adopted Git, a powerful distributed version control system, by embracing its flexibility and adapting it to the specific requirements of SAS projects. Through an innovative approach, we have tailored our workflow to leverage Git's capabilities, ensuring regulatory compliance, validation, and an auditable trail of changes. Git's branching and mer ging features have revolutionized our code review processes, fostering seamless collaboration among teams. Furthermore, the ability to track commit history has significantly improved productivity by enabling efficient issue identification and resolution. By overcoming the challenges of adapting Git to SAS projects, we have achieved excellence in project management, delivering high -quality, reproducible SAS solutions to our pharmaceutical clients | sas programmer, data cleaning, quality control, sas studio, sas programming | source poster source |
4 |
| 177 | The Challenges of Using Parent Study Data in a Long-Term Extension Study | Kirsty Wall | Imagine you need to set up the programming to report a Long-Term Extension (LTE) study that was designed to recruit subjects from multiple different parent studies. How would you go about the complicated task of pulling all the parent study data (e.g. labs, AEs, baselines) into the LTE study for reporting? We will discuss different options along with our chosen solution, considering the assignments of SUBJID and USUBJID in this situation, use of the DQ domain, what variables to include in ADSL to reference the parent studies and how to merge the parent study data, whilst being mindful of potential treatment unblinding issues and the need for CDISC compliance | adam dataset, sdtm domain, sdtm programmer, adverse events, baseline, cdisc | source presentation source |
9 |
| 178 | The Development of the CDISC eTFL Portal to Support the Adoption and Implementation of the Analysis Results Standard | Bess LeRoy, Bhavin Busa |
In April 2024, CDISC published the Analysis Results Standard (ARS) to support automation, consistency, traceability, and reuse of results data. In order to promote the adoption and implementation of ARS, CDISC is developing an eTFL portal that will reside within the CDISC Knowledge Base on the CDISC website. The eTFL portal will be comprised of packages containing informative examples showing how to implement ARS for a specific analysis concept. Each package will contain a definition of an analysis concept and examples of display shell representation, ADaM dataset and metadata, analysis results metadata, and an analysis results dataset. This presentation will provide an overview of the eTFL portal, associated assets, and future development | adam dataset, clinical trial, regulatory submission, change from baseline, clinical data, system organ class, adverse events, data science, cdisc | source presentation source |
8 |
| 179 | The Environmental Impact of Healthcare Data Science | Louise Hazeldine | We are striving to learn more about treatments to improve people's lives but what impact is the industry having on planet health? Some categories could be surprising! A lot of important work is being done, but there are some simple changes that can be made to reduce our carbon footprint individually and collectively. What is realistic for one person might not be for someone else. It's not realistic to walk 20 km to the office and not everyone can afford solar panels, but we could use the train and switch off the radio when leaving the room. It is important to assess the benefit versus impact of choices we make for ourselves or as part of a company-wide decision. Whether that is travelling to the office, buying new technology, printing the document to review, or even sending a simple 'Thank you' email | source poster source |
3 | |
| 180 | The Journey to Open Source Submission Readiness and Scalability | Casey Higgins | Presentation | funding risk | source | 14 |
| 181 | The Jungle in the Cloud | Sven Prasse | Poster: The transition of clinical data into the cloud is gaining speed with the adoption of open-source and AI technologies. Technical challenges have been solved by service providers like Entimo who operate cloud-native solutions for the pharmaceutical and biotech industry via browsers from wherever the users are located. This new world comes with pitfalls in legal restrictions, regulatory compliance and data protection. These management aspects remain the responsibility of pharma companies and their leadership teams, and they must be aware of these from the beginning of each project. Regulations such as GxP, GAMP and CFR 21 Part 11 have global relevance for clinical platforms. Other laws and regulations - GDPR, HIPAA, CRA, CSDDD - must be considered for globally operating companies | clinical data, data protection | source | 1 |
| 182 | The Magical Trinity of TLF Automation: Digital Documents, Metadata, and GenAI | Alastair Scarlett | Are we entering a new era for the analysis of clinical studies? An era in which massive time savings, quality improvements, and cost savings can be achieved by automation of the production of statistical results. The holy grail of every Stats Programming team, surely! The fast-evolving field of generative AI (GenAI) is capable of amazing feats of code generation. Inform those models with structured digital documents (not digitised documents) to define the target outputs; and inform those models with guidelines and constraints provided by standards metadata & models. The result will be high quality outputs, simply requiring QA by a human (or perhaps an independent AI model?), for at least 80pct of planned outputs | training set, adam dataset, clinical trial, data science, cdisc | source presentation source |
4 |
| 183 | The Open SCE: A Call to Action One Year Later | Caroline Phares, Pascal Bouquet |
data cleaning, pharmaceutical industry, clinical trial, statistical programmer, clinical programming, data integrity, data science | source presentation source |
6 | |
| 184 | The Oz Principle: Unlocking the Path to Accountability and High-Performance Teams - An Interactive Journey | Lynn Clipstone | Calling those who want to improve results through individual and organizational accountability! Get ready for an interactive and transformative journey through The Oz Principle by Roger Connors, Tom Smith, and Craig Hickman. Join me as we break free from conventional presentations and dive into an engaging experience that brings the concepts of accountability and high-performance teams to life. We will explore how the Oz Principle can revolutionize yours and your teams accountability, through the See It, Own It, Solve It and Do It method. This interactive presentation is your opportunity to gain practical insights and leave with actionable strategies to elevate your people management skills. I will also share some real life examples of how this has been used within PPD, to inspire our staff to greater ownership. Get ready to click your heels and unlock the power of The Oz Principle! | source presentation source |
3 | |
| 185 | The Prentice-Wilcoxon Test to Study Time to Healing for Split Face, Double Blind, Dermatology Clinical Trials | Shrutimita Pokhariyal, Craig Wesselman |
Dermatology clinical trials often have randomized, blinded studies that have a split face comparison for two treatments, usually investigational medicinal product (IMP) and a vehicle or active comparator. In indications such as rosacea, acne vulgaris, healing after cosmetic treatment etc., a participant is enrolled in the study and is randomly allocated the study treatment to each side of the face. The objectives of such clinical trials are to assess the safety and efficacy of the IMP and one of the clinical endpoints can be time to healing. While time to event is a common endpoint in drug development, tests such as Log-Rank or Gehan test prove to be inadequate when within subject correlation is involved. This p aper will talk about the Prentice-Wilcoxon Test used to test for differences between the time to first healing between two treatments | wilcoxon test, correlation, clinical trial, sample size, sample, clinical research | source presentation source |
10 |
| 186 | The R Journey to Submission | Hinal Patel, Welwyn Garden |
In the rapidly evolving landscape of pharmaceutical development, Roche embraces the use of open-source R in the regulatory submission process. In this paper we will share a pioneer experience of end -to-end R submission for a new drug application to the FDA, EMA and the NMPA. We will unfold the journey of overcoming the initial challenges to achieving the filing goal and emphasise the practicalities of preparing clinical trial data in R leading to a comprehensive eSubmission package. We will also share our strategic communication and collaboration with regulatory agencies, highlighting how R -generated data and documentation were effectively presented and validated within the submission framework | data cleaning, clinical trial, data science, adam dataset, regulatory submission, cdisc | source presentation source |
6 |
| 187 | The Significance of Generic Data Transfer Specification for Non-CRF Data in Clinical Trials | Samaneh Sabiani, Kristina Zweier |
Non-CRF data in clinical trials encompasses information not captured in the CRF, including data from medical devices, lab equipment, imaging systems, patient diaries, and other patient-reported outcomes. This data provides vital context, support, and evidence for trial outcomes but is often less structured, posing standardization and integration challenges. Standardizing non-CRF data involves mapping diverse data types to a common format aligned with CDISC standards. A generic data transfer specification is crucial for ensuring data consi stency, accuracy, and establishing a mapping roadmap for SDTM compliance. It also provides a structured approach for mapping data to SDTM when direct compliance is unfeasible, enhancing interoperability and harmonization of clinical trial data | data cleaning, sdtm domain, clinical trial, excel, data integration, data standard, sdtm programming, cdisc | source presentation source |
9 |
| 188 | The TransCelerate/CDISC Digital Data Flow Project: Practical Electronic Study Designs | David Iberson-Hurst | Presentation | cdisc controlled terminology, controlled terminology, adam dataset, clinical trial, protocol amendment, regulatory submission, data science, cdisc | source | 8 |
| 189 | Theory to Practice: Harnessing R to Enhance Workflows | Joana Marques Barros, Matthew Brierley |
Experience has shown that there is often a gap between introductory classroom training in R and the skills needed for independent work. In many instances, SAS programmers may receive training in R but then continue to work on studies or projects with infrequent opportunities to consolidate these skills. Here we outline a few techniques insights and references attempting to motivate the eager SAS programmer to continue on the R journey, focusing on ideas that could be used outside of standard ADaM and TLF generation | excel, clinical programming, sas programmer, clinical data, cdisc | source presentation source |
11 |
| 190 | Tips and Tricks: From Specification to Define.xml v2.1 using SAS | Caroline Pelckmans | Developing a Define.xml that is compliant with CDISC standards and ready for regulatory submission can be time consuming. Therefore, it is of interest to have a validated tool within your company that can be easily used by any programmer. In this presentat ion, we will go deeper into the process flow of how to use SAS software to generate a Define.xml, going from a filled ADaM specification to a Define.xml (a machine -readable document). We will share practical guidance on incorporating new elements and attributes from Define v2.1, such as subclass, versioning of standards and controlled terminology and origin enhancement. Lessons learned will be given on how to improve your SAS programs based on Pinnacle 21 and/or visual checks, to make your Define.xml compliant with the required standards | proc template, controlled terminology, proc step, macro definition, dataset, macro variable, proc copy, excel, data science, adam dataset, cdisc | source presentation source |
18 |
| 191 | Topic and Sentiment Analysis of Anti-Vaccine Activism Using Natural Language Processing and Large Language Models | Tamas Bosznay | Anti-Vax activism and vaccine hesitancy have always been existing phenomena in modern society; however, they became more prevalent in recent years, mostly related to the COVID-19 pandemic. To understand the factors around and to improve vaccine uptake, researching the public sentiment and opinion on COVID-19 vaccines is a useful first step. This study uses Social Media data (mainly datasets from X - formerly known as Twitter) to perform topic analysis around vaccine hesitancy. The relevant English language tweets are labelled against topics using various NLP techniques such as LDA and NMF. As an alternative method, various generic (GPT-4, Claude3, Llama3) and task-specific Large Language Models will be used. Following the visualisation of topics, sentiment analysis will also be performed both using more traditional sentiment classifiers and LLMs (BERT, VADER, TextBlob). The results specifically for the use case of anti-vaccine activism will be presented at the end of the paper | feature extraction, proc freq | source presentation source |
12 |
| 192 | Transcriptomics - A Demanding Data Analysis that Expands the Limits of Regulated Clinical Data Handling | Lea Yoshida Hildebrandt | clinical trial, quality control, csr, sample, clinical data, data wrangling, data standard, primary endpoint, proc princomp, data science | source presentation source |
6 | |
| 193 | Transformative Leadership: Innovating for Sponsor-Vendor Partnership Success | Nalin Gupta, Satish Shetty |
source presentation source |
5 | ||
| 194 | Transparent ggplot2: Enhancing Reproducibility with ggcall | Maciej Nasinski | Transparency and reproducibility are fundamental principles in data analysis across various fields, from academic research to industry applications. The ggplot2 R package is widely used for data visualization in R, but complex plots generated within functions can hinder transparency due to inaccessible underlying code. This paper introduces the ggcall, an open-source R package designed to extract the construction code of ggplot2 objects, enhancing transparency and reproducibility. I discuss the benefits of ggcall for improving practices, illustrate its use with examples, and explore its integration with other tools promoting code transparency. By providing tools for accessing hidden code, ggcall represents a significant step toward better practices in data analysis and promotes transparency and reproducibility. Future directions include the development of an extension to the magrittr package and extend existing pipes with gathering code functionality | data visualization, data science | source presentation source |
6 |
| 195 | Trustworthy and responsible AI in health care and life sciences from a leadership perspective | Mark Lambrecht | The adoption of AI is booming. AI is revolutionising human life and opening up a whole new world of innovation - from helping doctors diagnose medical conditions earlier to enabling researchers to develop new drugs to treat them and enabling more effective conservation of endangered species and habitats. As AI becomes more prevalent, it will affect almost every aspect of society, from our professional to our personal lives. AI also carries significant risks if it is developed, implemented and managed without intentionality and discipline. For example, AI systems trained on biased data can learn historical patterns of discrimination against women, people of colour or vulnerable populations. If biases in AI systems are not identified and mitigated before implementation and use, those implemented systems can adopt social biases and cause unintended consequences on a large scale | clinical trial, clinical research, life sciences | source presentation source |
3 |
| 196 | Two Hats, One Noggin: Working as a Developer and as a User of the Admiral R Package for Creating ADaM | Edoardo Mancini, Benjamin Straub |
r package | source presentation source |
12 | |
| 197 | Unlocking the Enigma: Exploring SAS and R Strategies for Handling Non-Printable and Special Characters | Eldho Alias, Vishal V Nair, Pooja Pradeep Pillai |
Quality is one of the key requirements in clinical trial deliverables. The existence of non-printable and special characters in data can cause serious problems in the reports which includes incorrect counts/statistics in the deliverables, or the minor issues such as incorrect line breaks, page brakes or appearance of strange symbols in the reports. Here comes the importance of this paper to identify and handle the non -printable and special characters using SAS 9.4 as well as R. This paper will portrait the concept of SAS macro/R functions which have the flexibility to either remove those characters or replace with specified characters in a single run for all the datasets in a library. Additionally, a report will be created with information including the identified/replaced characters with the associated dataset and variable name that will help the user for easy understanding. SAS/R codes will be shared along with the paper |
clinical trial, excel, dataset, clinical data, macro definition, clinical research, macro variable, data cleaning | source presentation source |
11 |
| 198 | Unlocking the Potential of Unstructured Data in Real-World Evidence: Leveraging Text for Enhanced Patient Classification and Regulatory Validation | Georgios Panagiotakis, Maria Puri |
The utilization of unstructured data in Real-World Evidence (RWE) has emerged as a promising avenue for extracting deeper insights into patient health, diagnosis, and treatment outcomes. [1]. While structured data from Electronic Medical Records (EMR) and claims provide a foundation for analysis, the richness of unstructured, free text data remains largely untapped. By developing in house capabilities such as Natural Language Processing (NLP) techniques, researchers can extract meaningful information from diverse textual sources such as clinical notes, free-text reports or physician narratives, and patient-reported outcomes | unstructured data, unsupervised learning, machine learning, clinical data, insurance claims | source poster source |
10 |
| 199 | Unlocking the Potential of Your Data with Retrieval-Augmented Generation (RAG) While Upholding Privacy: Use Cases with Business Impact at Roche | Mathieu Cayssol, Adam Forys |
ThelaunchofChatGPTinNovember2022igniteda transformativerevolutionacrossalltheindustries,includingthepharmaceuticalindustry, to enhanceinnovationandproductivity. Oneof thecommonLargeLanguageModelusecasesisthedevelopmentofRetrievalAugmentedGeneration(RAG)solutionsbuiltonspecificknowledgebases.Itconsistsof a ChatGPT-likeapplicationthatutilizesenterpriseknowledge,upholdingdataprivacy, andoperatescost-effectively. Throughthisrevolution,manyindividualsoftenfindthemselvesoverwhelmedbythecomplexitiesofintegratingadvancedAItechnologiesintotheirworkflows. Inourpresentation,wedelveintotheworldofLargeLanguageModels(LLMs),focusingonRAGarchitectures.Webreakdownthetechstacksandinfrastructureneededtocreatetheseapplicationsandemphasizetheimportanceofdataprivacyconsiderationsthroughoutthedevelopmentprocess.Inaddition,wearesharingsomeexcitingprojectsdevelopedinternallyatRochethatareboostingefficiency | privacy, access control | source presentation source |
6 |
| 200 | Unveiling R Package Risk Assessment: A Comparative Analysis of Metadata Extraction | Malte Stein, Sebastià Barceló |
In this paper, we delve into our innovative approach for assessing risk associated with R public packages from CRAN, focusing on the extraction of metadata crucial for informed decision-making. We elucidate our methodology for accessing R packages metadata, shedding light on the intricacies of metadata extraction. Through a comparative study, we showcase the disparities in metadata extracted using our approach versus conventional methods like the R riskmetric package. By highlighting these differences, we provide valuable insights into the limitations of existing techniques and the advantages of our approach in comprehensively assessing package risk. Join us to gain a deeper understanding of R package risk assessment and discover new avenues for enhancing package evaluation practices | data cleaning, api | source presentation source |
8 |
| 201 | Unveiling the Tipping Point: An In-depth Analysis | Varsha Mithun Patil, Ephicacy Lifescience |
Tipping point analysis in clinical trials is a statistical technique used to assess the robustness of primary and secondary analysis results when dealing with missing data. Despite efforts to minimize the amount of missing data through proper study design and conduct, some missing data is inevitable. When analyses for respective endpoints are performed based on the Missing at Random (MAR) assumptions, sensitivity analyses are necessary to assess the impact of deviations from these assumptions. This is particularly important when a significant difference in treatment effect is observed, but the number of patient discontinuations is imbalanced between treatment groups. In such situations, the validity of the results could be questioned. Regulatory agencies often request tipping point analysis as a sensitivity analysis under the Missing Not at Random (MNAR) assumption | missing value, imputation, sensitivity analysis, clinical trial, proc mi, treatment effect, proc step, multiple imputation, missing at random, data science | source presentation source |
8 |
| 202 | Use of SAS Packages in the Pharma Industry - Opportunities, Possibilities and Benefits | Bartosz Jabłoński | When working with SAS○Rcode, especially with a complex one, there is a moment when you decide to break it into small pieces. You create separate files for macros, formats/informats, for functions or data too. Eventually the code is ready and tested, and sooner or later you will want to share it with another SAS programmer . You have developed a program using local PC SAS, but the deployment is on a server with a different OS. Your code is complex (with dependencies such as multiple macros, formats, datasets, etc.) and is difficult to share. Often when you try to share code, the receiver will quickly encounter an error because of a missing macro, missing format, or whatever ...small challenge, isn't it? In this article we discuss a solution to the problem - the idea of SAS Packages, what they are, how to use and develop them, and how to make code-sharing a piece of cake | sas packages, sas enterprise guide project | source presentation source |
10 |
| 203 | Using Git and GitLab to Empower QC Workflow and Project Management in Clinical Reporting | Kieran Martin, Welwyn Garden |
Git has been an extremely popular tool in software engineering over the last 2 decades. In this talk I will look at how we are taking advantage of this tool and Gitlab in the context of QC workflow. I will explore the advantages of the integrated project management tools, the clear audit trail, and the ability to collaborate effectively when using Git. I will discuss some of the difficulties of using this new tool, both in terms of the specific needs of our workflow, and the learning curve of Git, and how we have overcome these | data science, clinical data | source presentation source |
5 |
| 204 | Using RWD to Inform the Next Study: A Case Study | Jean-Marc Ferran | Presentation | logistic regression, analysis set, regression analysis, concomitant medications, logistic distribution, change from baseline, confidence interval | source | 22 |
| 205 | Validation and Open-Sourcing: Can We Tame the Two Headed Beast? | Ben Barnaby-Pass | Whenever we speak about open sourcing it always comes with the rebuttal of 'How do you do validation?' . While the validation process completely differs from SAS, the process in R can be as 'simple' as SAS too. Of course, we externally source packages to help make R function, tidyverse and Shiny are just two examples, these packages must have a certain level of integrity to them. We are unable to run a package that hasn't had an official release, and we are unable to use these packages in a continuous update state, this prolongs the validation process. To get past this issue there's the opportunity to freeze the packages at a certain version, however, it puts the pressure on companies to keep kicking the can down the road. Even from this abstract you can see the troubles that open source has on Validation | data cleaning, stage 2 | source presentation source |
4 |
| 206 | Verify: Facilitating End-to-End Automation of Clinical Validation Processes Using AI Algorithms and Dynamic Collaboration | Ilan Carmeli | The current gold standard for statistical output validation is double programming of datasets and TLFs. However, this approach is limited: it also relies on time-consuming processes such as visual review, manual processes to validate formats, and often leaves cross-validation to the last stage. In addition, output review and communication between stakeholders is often conducted across multiple disconnected platforms. Verify streamlines validation workflows by leveraging ML/NLP technologies and providing an interactive platform for real-time communication. Both features lead to a significant reduction in validation timelines. Verify performs validation checks to identify discrepancies between outputs and reference documentation (such as mock shells), consistency checks between outputs and ADaM, checks to validate formats, and in-table and cross-table checks to validate logical and numerical consistency | data cleaning, independent qc, mock shell, cdisc | source presentation source |
3 |
| 207 | Visual Monitoring of Intercurrent Events with Heatmap | Alona Bulana | Poster: Problem Identification, analysis, and visualisation of intercurrent events (IEs) are becoming an integral part of clinical studies implementing the estimands framework . Heatmaps are an effective visualisation for displaying many data points on a 2D canvas. Events such as concomitant medications, adverse events, study milestones or other events of interest are overlaid on the heatmap with different shapes, allowing for clear visual exploration. In order to enlarge the image without losing quality, it is essential to produce it in a vector format like SVG | intercurrent event, adverse events, concomitant medications | source | 1 |
| 208 | When the Answer to Your Problems Is Just One Call (execute) Away | Luis Osorio, Lara Suckut |
Recently, many pharmaceutical companies have reported a shift from traditional to metadata -based data structures and programming. This calls for a dynamic, data -driven programming style, which is uniquely represented in SAS programming by the function "call execute". However, to date, this function has not received the recognition or utilization it deserves. We set the stage for this versatile tool, as learning its framework opens up a wide range of applications using pre-existing coding skills. We highlight the power of data -driven code generation and delayed execution from a DATA STEP and place it in the context of macro programming. We also advise on common pitfalls and support each application with explicit examples. We hope to leave the audience with a firm grasp of the opportunities and the ability to make use of this neglected function to support the current shift in the industry |
call execute, macro call, data step, macro quoting function, proc sql, proc step, dictionary tables, macro definition | source presentation source |
6 |
| 209 | Where Does Your SCE Fit in Your Data Strategy? | Phil Loucks | A data strategy is akin to a blueprint for managing, interpreting, and utilising data within an organisation. It encompasses plans for data collection, storage, analytics, and alignment with business objectives. Imagine it as a roadmap guiding how data - regardless of its source - should be organised, stored, and accessed. Your statistical computing environment (SCE) is a key step in your data value chain and thus it should feature highly in your organisation's clinical data strategy | clinical trial, clinical data, data integrity, data governance, data science, audit trail, edc, data cleaning, key performance indicator | source presentation source |
7 |
| 210 | Whose Job is it Anyway? A Foolproof Way to Create a Culture of Engagement | Jennifer Dootson | Presentation | source | 19 | |
| 211 | Why Are Old Wines Sometimes Tastier Than New Ones? | Lars Andersson, Henning Pontoppidan Föh |
adam dataset, phase iii, statistical programmer, sas programming, imputation, adam programmer, clinical trial, pharmaceutical industry, phase ii, temporary dataset, data cleaning, cdisc | source presentation source |
10 | |
| 212 | Why Writing Tests Does Save Time in Development | Magnus Mengelbier | The Life Science industry spends an unproportionate effort to create unofficial re-usable utilities, such as the SAS macros copied from study-to-study that never reaches the global library. As R is used more broadly across an organization, bespoke packages are soon to follow. An interesting lesson learned is how tests for an R package is managed in the overall development cycle and what roles they can play. We consider different development approaches, from the common develop first and test last to incorporating adaptations of Test Driven Development. An interesting reflection we explore is how writing the tests early in the development cycle can both be used to significantly decrease the development time for utilities and tools written in SAS, R, Python, etc., but also how that can make that SAS macro transition from study to the more beneficial study to global library | data cleaning, macro definition | source presentation source |
6 |
| 213 | Working with Omics data - A Statistical Programmers perspective | Adrian Czaban | sample, clinical trial, sdtm domain, data science, clinical research, unique key, data cleaning, cdisc | source presentation source |
10 | |
| 214 | {admiralvaccine} Project JouRney | Federico Baratin, Michael S Rimler |
{admiralvaccine} is an open -source R package which enables creation of ADaM compliant domains specific to Vaccine clinical development. The package is an extension of the {admiral} family of packages from the pharmaverse and is a collaborative effort of three Pharmaceutical Companies (GSK, Pfizer and J&J). In this paper, we aim to provide the audience a complete overview of the project's strategy to deliver the initial release in less than 6 months, including the approaches for development, testing, and release. We will also describe key features of the package, planned future enhancements, and challenges experienced along the way | adverse events, pharmaceutical industry, cdisc | source presentation source |
8 |
| 215 | {r}amnog: Towards a Common, Open-Source Solution for Programming AMNOG-Style EU-HTA Analyses | Matthew Phelps, Christian Olsen |
The upcoming EU HTA requirements of the EU Joint Clinical Assessment (JCA) represents both a challenge and an opportunity. It's a challenge because draft guidelines indicate significant overlap with the analysis requirements for Germany reimbursement (AMNOG), well known for extensive workloads under challenging timelines. What infrastructure exists to handle the AMNOG analyses? Is it EU HTA proof? At Novo Nordisk, we have open-sourced an ecosystem of R packages to substantially reduce the programming time required for AMNOG style submissions. We aim to engage the wider community in further development, so they work across the industry. Therin lies the opportunity - to move towards a common way of programming AMNOG style submissions, so each company doesn't have to re-invent the wheel | regulatory submission, csr | source presentation source |
6 |
| 216 | {SAS to R-evolution}: A Statistical Programming Team Journey | Charlotte Cheinin | Poster | statistical programmer, clinical research | source | 1 |
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.