Jason Binder is the founder and CEO of Ember, where he is building an AI companion to support patients and caregivers navigating serious illness, starting with glioblastoma.

He brings 25 years of experience across oncology drug development, clinical data, and enterprise transformation, with leadership roles at AstraZeneca, Genentech, and Merck & Co.. His work has focused on building clinical data and planning systems at enterprise scale—supporting portfolio decisions, capacity planning, and visibility across the drug development lifecycle.

That perspective changed when his wife was diagnosed with brain cancer. Sitting on the other side of the system he helped build, he experienced how difficult it is for patients and families to find, understand, and act on information when time is limited. That gap now shapes his work—focusing on patient-centered data, quality of life, and extending support beyond the clinic.

Jason writes and speaks on AI in healthcare, the role of lived experience in data, and new approaches to accelerating cancer innovation.

Analysis Data Model (ADaM) documentation has been strewn across many different documents, as the ADaM standard has evolved and expanded since version 2.1 of the ADaM Model and version 1.0 of the ADaM Implementation Guide (IG) were published in 2009.  There have been substantial changes to the ADaMIG in subsequent versions which made some portions of documentation published under the original versions obsolete.  In addition, inconsistencies in variable definitions and attributes have crept in over time. The ADaM team is consolidating all of the ADaM normative documentation, bringing everything up to ADaM version 3.0. This presentation will describe the changes to the structure of the model and IG, while details about examples and new content will be covered in other presentations within the ADaM 3.0 Update presentation series.  This topic is being presented by the ADaM 3.0 Consolidation sub-team lead and the future ADaM team lead.

While consolidating the ADaM Model and Implementation Guides, as outlined in part 1 of this ADaM 3.0 Update presentation series, the team realized that it was not practical to bring over every example from these Guides. An existing ADaM sub-team was already looking at ADaM informative documents, including ADaM Examples documents, Knowledge Base Articles, and Therapeutic Area User Guides (TAUGs), and this sub-team is now also determining where to put these longer, more involved examples from the Guides. This presentation describes how we are ensuring existing Implementation Guide examples are updated as needed, adding new examples, and determining where these examples belong, as part of the larger holistic effort to help users find optimal conformant ADaM implementation solutions. This content is presented by the ADaM Informative Documents sub-team lead, who is also an active ADaM 3.0 Consolidation sub-team member and former ADaM team lead.

With a quick overview of what ADaM 3.0 will contain, which is covered in detail in Part 1 of this ADaM 3.0 Update presentation series, this presentation focuses on new material included in ADaM v3.0. We will walk through new variables not previously part of any normative ADaM documents. We will describe what we have done with PARQUAL, a variable provisionally used in some CDISC example documents and mentioned in the USFDA Pilot OHOP Safety Team Standard Data Requests document. We will cover the new ADaM concept of Analysis Topic. Additionally, since new content comes with new opportunities for conformance rules, we will give a preview of what rules are being added or modified based on this new content. This is presented by two active ADaM 3.0 Consolidation sub-team members, one who is a former ADaM team lead, and the other who is the ADaM Conformance sub-team lead.

CDISC 360i calls for standards implemented as linked metadata with semantic relationships enabling end-to-end automation. The foundational standards are strong. What is missing is a context management layer that connects them at runtime with provenance, temporal versioning, and federated discoverability. NGSI-LD, an open ETSI standard adopted across digital twins and federated data ecosystems worldwide, provides exactly this layer. Its information model represents entities as property graphs with first-class relationships encoded in JSON-LD. Its context broker architecture supports dynamic discovery, subscriptions, and temporal queries, capabilities that map directly onto 360i requirements. This presentation shows how NGSI-LD can serve as connective infrastructure for CDISC standards without modifying them, demonstrates a biomedical concept traversing from USDM through SDTM and ADaM as linked entities, and explores practical alignment with all standards.

Traditional clinical trial protocols often contain ambiguities that can hinder digital data flow. This presentation introduces a framework and provides an in-depth exploration of Objective and Endpoint Standardization. This approach aims to bridge the gap between narrative study design and a structured, automated data flow, serving as the missing link between USDM and analytical concepts.

We will utilize standard codelists from the NCI and CDISC, along with the BEST (Biomarkers, EndpointS, and other Tools) resource developed by the FDA-NIH Biomarker Working Group. Through examples, we will demonstrate how to classify endpoints and establish a semantic layer, and show interconnectivity among activities, which will also help build the foundation for the analytical concepts.

The CDISC Biomedical Concept (BC) framework offers a robust foundation for interoperability. Yet, translating unstructured protocols into these standards manually creates a costly ""Interpretation Gap."" This proposal introduces an AI-driven methodology, leveraging comprehensive ""360i Vision,"" to operationalise the CDISC framework. By designating the BC as the ""Atomic Unit of Truth"" an anchor tying clinical intent directly to technical execution. This approach automates end-to-end study orchestration. The AI engine accurately maps standard BCs while simultaneously capturing sponsor-specific ""Biomedical Variants"" and proprietary legacy nuances. This metadata-driven synchronisation automates USDM enrichment, dynamic ODM-XML eCRF generation, and custom Data Review/ Monitoring Plan validation logic. Shifting the focus from manual programming to expert validation, this automated architecture drastically reduces the time, financial cost, and human effort required for study start-up, achieving database lock-readiness with unprecedented speed and precision.

We talk endlessly about what AI could do with our TMF. We talk far less about what our teams can do with AI. Both however lend themselves to the same question, are we ready? 

Some team members still long for the simplicity of a physical document. Others can't imagine a world without record uploads, workflows and interactive dashboards. Now we're asking both to work with AI, compliantly.
This session asks how we build AI competency across wildly different starting points; what skills do we start with? What does training on this topic even look like? What does this do to career progression? And even, should we be the ones teaching this?

And once we've answered all of that , what does TMF training even look like in a world where AI is just part of the job?

As AI automates TMF processes, the human element becomes even more critical. This session focuses on the intersection of technology and people, exploring how organizations can balance AI-driven efficiency with strong team engagement and collaboration. We will examine how connected systems and standardized processes create transparency, while AI provides actionable insights that empower teams rather than replace them. Key topics include change management, cross-functional alignment, and building trust in AI systems. Attendees will gain insights into creating a sustainable TMF model where human expertise and AI work together to drive quality and innovation.

Overcoming persistent Trial Master File (TMF) challenges remains critical to ensuring inspection readiness, data integrity, and regulatory confidence. Our presentation focuses on practical, standards-driven approaches sponsors and vendors use to address common TMF pain points, including unclear ownership, delayed and inconsistent filing, quality gaps, and limited cross-functional oversight. Building on these foundational practices, the session explores how artificial intelligence (AI) and machine learning can supplement existing TMF processes to improve efficiency, visibility, and quality,without compromising compliance. Attendees will learn how structured data, standardized metadata, transparent algorithms, and human-in-the-loop controls enable AI to support document classification, quality monitoring, and risk-based TMF management. By emphasizing governance, traceability, and alignment with data standards, this presentation highlights how organizations can responsibly integrate AI to strengthen TMF oversight, improve decision-making, and advance TMF maturity while maintaining regulatory trust

While clinical data and TMF processes have seen increasing levels of standardization, the Investigator Site File remains a largely overlooked component of trial operations. Often managed through fragmented systems and inconsistent processes, ISF practices limit traceability, reduce efficiency, and introduce compliance risks across global studies.

This presentation explores how a more structured, standardized approach to ISF management can enhance data integrity and operational performance. By applying principles aligned with CDISC standards, such as consistent metadata, harmonized document structures, and improved process integration, combined with AI for document classification and automated metadata mapping with sponsor eTMF systems, organizations can strengthen inspection readiness and enable better collaboration between sponsors, CROs, and sites.

Through practical examples, including regional expectations, we illustrate how modernizing ISF practices supports scalability while maintaining local adaptability. Attendees will gain insights into how elevating the ISF with intelligent automation contributes to more efficient, connected, and compliant clinical trial execution.

Building an inspection-ready Trial Master File requires strategic partnership, not reactive remediation. In this joint presentation, Intellia Therapeutics and Phlexglobal share real-world insights on cultivating successful sponsor-provider TMF partnerships across the full study lifecycle.

The session explores how early alignment on governance, accountability, and shared ownership , supported by joint oversight committees, RACI frameworks, and measurable SLAs and KPIs , establishes inspection readiness from Day 1. Grounded in the TMF Reference Model and aligned with ICH E6(R3), FDA 21 CFR Part 11, and EU Annex 11, these frameworks drive transparency and confidence throughout study execution.

Attendees will learn how effective eTMF configuration, study-specific TMF Plans, and continuous completeness monitoring support consistent performance , even amid migrations, system upgrades, and inspection surges. The presentation concludes by demonstrating how true TMF partnerships evolve beyond vendor delivery into strategic alliances that reduce risk and accelerate innovative therapies to patients.

Study Data Tabulation Model 3.0

Version 3.0 of the SDTM is now in the publication process.  We review the changes in the model since version 2.0, which accompanied SDTMIG 3.4.

Following these releases, the primary focus of the SDS team activites will be shifting from revising existing standard documents to supporting the development of more complete digital standards, including biomedical concepts, as part of the larger CDISC 360i project. This presentation will discuss the new activities that the SDS team will be undertaking in the months and years to come.

2: SDTM Implementation Guide 4.0

Version 4.0 of the SDTMIG is now in the publication process. We review the changes since version 3.4, including domains added after the initial review process and the split of clinical classifications from disease response assessments. 

Following these releases, the primary focus of the SDS team activites will be shifting from revising existing standard documents to supporting the development of more complete digital standards, including biomedical concepts, as part of the larger CDISC 360i project. This presentation will discuss the new activities that the SDS team will be undertaking in the months and years to come.

3: SDS: Future Directions

Following these releases, the primary focus of the SDS team activites will be shifting from revising existing standard documents to supporting the development of more complete digital standards, including biomedical concepts, as part of the larger CDISC 360i project. This presentation will discuss the new activities that the SDS team will be undertaking in the months and years to come.

Submitting clinical data across global health authorities requires more than adherence to CDISC standards it demands adaptability to regional expectations and increasing data complexity. This session shares practical lessons learned from FDA, PMDA, and NMPA submissions, enhanced by the use of AI for data validation, anomaly detection, and metadata consistency checks. We will highlight how AI accelerates submission readiness, identifies region-specific gaps, and supports proactive issue resolution. Attendees will learn key and don along with strategies to combine standards and AI to improve submission quality, reduce rework, and streamline global regulatory interactions.

Artificial intelligence is reshaping how organizations approach complex, rule-bound processes. While some functions are shifting rapidly, those with deeply embedded expertise often find AI results underwhelming or counterproductive. The problem is rarely capability, but rather that solutions tend to answer the wrong questions because experts are uncertain how to ask the right ones. It is not so much a failure to communicate as a timidity of imagination.

This paper presents two interconnected use cases. The first automates the generation of the SDTM-annotated Case Report Form.  The second facilitates Define-XML generation. In both, a conversational AI platform serves as a Software as a Solution (SaaS) development partner, enabling rapid prototyping and a more creative approach to day-to-day challenges. The aim is not necessarily to displace enterprise solutions, but to demonstrate, through targeted and immediately practical examples, how domain experts can begin extracting real value from these tools today.

The adoption of Next Generation Submission(NGS) levers is transforming our preparation of global regulatory submissions and enabling accelerated timelines thru standardized, automation-driven, more efficient processes. This paper presents a collaborative submission acceleration of NGS implementation between a CRO and a global sponsor, focusing on pre-DBL planning, Dress Rehearsal approach, and post-DBL strategies. Key NGS levers include alignment with CDISC standards for multi-regional submissions, a shadow team for early restricted unblinded PK/popPKPD data access and analysis readiness, a CSR reduction plan and checklist, and an automation-enhanced process for in-text tables and safety narratives. Challenges like cross-functional coordination, Intercurrent Event(ICE) handling, prompt third-party vendor(TPV) data transfer and issue handling, and regional regulatory variability were addressed through structured governance and iterative approaches. This case study provides practical insights and best practices for accelerated, scalable, future-ready, global submissions leveraging NGS principles. Results demonstrate improved submission process, consistent quality, reduced timelines, and enhanced regulatory compliance

As wearable sensors and digital health technologies (DHTs) become ubiquitous in clinical trials, the industry faces a significant challenge: transforming massive volumes of raw sub-second data into submission-ready SDTM datasets without compromising data integrity or traceability.

This presentation explores a robust data flow architecture utilizing the Digital Health Measurement Collaborative Community (DiMe) V3 framework (Verification, Analytical Validation, and Clinical Validation). We detail a specialized pipeline for transitioning from raw sensor output to structured clinical domains. Attendees will understand how to build a scalable, compliant bridge between raw sensor outputs and CDISC standards, ensuring that digital endpoints are as rigorous and transparent as traditional clinical measures.

Transforming real-world data into SDTM introduces data loss that's difficult to characterize and impossible to audit without standardized lineage. RWD collected for clinical care and billing does not map cleanly to SDTM, and the complex processing required creates unquantifiable biases that undermine the reliability required for regulatory review. CDISC's RWD Lineage project addresses this by defining a metadata model that captures data point-level lineage for each RWD-derived data point, from source through submission. We propose Define-XML as the representation format, using a namespace extension (rwdl:lineage) that integrates with existing CDISC submission infrastructure. In this presentation, we will talk about the RWD Lineage data model and walk through worked examples of implementation, demonstrating how lineage metadata is generated and propagated through RWD to SDTM transformations. Standardized lineage enables provenance to be combined across disparate data sources and provides a common foundation for validation tooling, audit workflows, and quantification of transformation performance.

Digital health technologies (DHTs) spanning wearables, eCOA platforms, mobile applications, and remote monitoring sensors are fundamentally transforming clinical trial design and evidence generation. DHT adoption in registered trials grew from under 1% in 2010 to over 11% by 2020, with decentralized trial market projections reaching $38 billion by 2034. The FDA's 2024 final guidance on decentralized clinical trials established validation, reliability, and oversight expectations for DHTs, while ICH E6(R3) formalized eSource's role in GCP-compliant research. Early adopters report 50% faster enrollment and up to 25% cost reductions. However, standardizing DHT-generated continuous
data streams into CDISC CDASH and SDTM-compliant formats remains an unresolved operational challenge, particularly for novel digital endpoints. This presentation examines the current regulatory
framework for DHTs, real-world implementation experience, challenges in digital endpoint harmonization, and the path toward CDISC-aligned data architectures that enable DHT-sourced evidence to meet submission-readiness standards.

Periodic review of clinical trial documentation is essential to ensure that TMF records present a clear, consistent, and inspection ready narrative. However, manual oversight approaches are often tedious, resource intensive, and cost prohibitive. To address these challenges, CSL Behring, in collaboration with Just In Time GCP, conducted a pilot to evaluate an AI driven TMF review and service model with a focus on site personnel records.  Site personnel documentation post site greenlight remains a persistent TMF management challenge and a focal point for regulatory inspection. Discrepancies at the site personnel level also frequently indicate broader risks within the TMF. 

This presentation will describe the pilot's operational model, highlighting key lessons learned and practical considerations for integrating AI into TMF workflows to strengthen oversight and enable sustained inspection readiness. These insights contribute to the broader discussion on the evolving role of AI in TMF management.

While the potential of AI in TMF management is well understood, few organisations have shared concrete lessons from implementation. This presentation addresses that gap. BeOne Medicines, working with an implementation partner, is deploying not a single AI tool but a coordinated set of AI agents across the full TMF lifecycle - classification, document-level quality verification against ALCOA++ principles, and ongoing TMF oversight. Rather than presenting a theoretical framework, this session shares practical experience from the first six months of implementation. We describe our methodology for designing AI agents within GCP-compliant workflows, including risk-based human oversight, confidence-based routing, and inspection-ready audit trails. We discuss project governance, AI decisions and guardrails, what worked, what proved harder than expected, and what we would approach differently. Whether evaluating AI or preparing for implementation, attendees will gain practical insight into the decisions, governance, and trade-offs involved.

Veeva Systems (Jim Horstmann, Senior Product Manager, Veeva eTMF) examines the evolution of large language models (LLMs) and their impact on moving toward an autonomous TMF. While traditional automation handles administrative tasks, agentic AI enables a more intelligent, connected clinical operating model. This session explores how new tools process document data at scale, shifting the TMF from a repository for inspection-readiness to a proactive asset delivering business insights.

Learn how to:

• Evaluate the current state of AI and its application to TMF processes
• Identify considerations for implementing AI-driven workflows to ensure data integrity
• Use AI agents to extract actionable insights and transition from an active to autonomous TMF

In an environment marked by rapid growth, evolving regulations, and advancing technologies, leadership is critical to successful change within Trial Master File (TMF) operations.  This session positions change management as a leadership-driven discipline that enables TMF teams to adopt new processes and technologies while maintaining inspection readiness, compliance, and data integrity.

Using real-world TMF scenarios - such as governance-led eTMF migrations, system and functionality enhancements, and the introduction of AI to support document processing and quality oversight - this session highlights how TMF Change Management Leaders shape culture, clarify ownership, and align cross-functional stakeholders.

Attendees will learn a practical change journey framework that TMF leaders can use to articulate purpose, set clear expectations, proactively manage risk, and demonstrate visible commitment.  Emphasis is placed on how leadership behaviors, decision-making, and communication directly influence user adoption, audit outcomes, and sustained TMF performance.

TMF is often seen as an administrative requirement, but how it is managed affects compliance, oversight, and collaboration. Many smaller sponsors and CROs operate without a dedicated eTMF, relying instead on shared drives, internal platforms, or sponsor-managed systems. This can make maintaining continuity, oversight, and cross-functional coordination challenging.

This presentation will share practical approaches for improving TMF management across teams and projects. Key actions include clearly defining responsibilities, keeping workflows simple, and providing timely guidance to staff. Efforts to shift TMF from a routine task to a recognized part of trial management have helped improve consistency and accountability.

Lessons from both successes and challenges highlight how to maintain oversight, engage stakeholders, and manage TMF effectively,even in organizations without a formal eTMF system. Attendees will leave with practical ideas applicable in real-world trial settings.

Ever since the advent of SDTM standards, non-standard data has been mapped using the vertical structure outlined in Supplemental Qualifier (SuppQua or Supp) datasets. This has caused issues sometimes when the data needs to be merged back with parent domains and used for analysis further down the data reporting process. That is going to change majorly with the release of the new version of SDTM: Model v3.0 and IG v4.0. This paper describes the transformation from vertical to horizontal structure for mapping non-standard datasets to the newer, more efficient Non-Standard Variables (NSV) datasets in SDTM. This would enable the direct integration of non-standard data into General Observation class domains for easier viewing and metadata application, thereby streamlining submissions by reducing structural limitations and promoting consistency within SDTM standards.

In the realms of pediatric, oncology, and diabetic trials, a challenge in capturing treatment and dose levels arises when protocol-planned dose levels are adjusted during the treatment period. How can data analysts choreograph the Analysis Data Model (ADaM) data sets to capture these nuanced dose levels? The treatment variables TRTxxP/N in the Subject-Level Analysis Dataset (ADSL) and their partners TRTP/N in Basic Data Structure (BDS) and Occurrence Data Structure (OCCDS) are designed to group subjects by treatments as columns in the summary tables. But we also need to preserve the dose level adjustments on a subject- and record-level basis. DOSExxP and DOSExxA are used in ADSL, while their counterparts, DOSEP and DOSEA, lead in the BDS and OCCDS data sets. Together, these harmonious variables pirouette across the ADaM data sets, capturing the very essence of the adjusted dose levels in a dance that seamlessly unfolds.

Biomarker data from external vendors are often delivered in proprietary or non-standard formats, creating challenges for analysis, traceability, and regulatory submission. This presentation describes a pragmatic approach for transforming vendor biomarker data into CDISC compliant SDTM datasets using internal mapping tables developed. Early standards engagement ensures that only biomarker data relevant to study objectives and Clinical Study Report requirements are mapped, minimizing unnecessary transformations and downstream rework. The approach includes identifying raw vendor data elements, applying CDISC controlled terminology, aligning data to appropriate SDTM structures, and capturing mapping metadata to support transparency and reproducibility. This methodology enables consistent and analysis ready biomarker datasets across studies, improves data quality, supports traceable derivations, and enhances regulatory readiness. Attendees will gain practical insights into integrating diverse vendor biomarker data into submission ready CDISC standards across programs while supporting reuse and long-term governance models for biomarkers.

Clinical study builds still rely on manual interpretation of protocol PDFs, creating inefficiencies and inconsistencies across study design. We present a multi-agent AI system that transforms protocol PDFs into USDM-compliant, machine-actionable study definitions. Specialized agents handle document parsing, semantic extraction, terminology alignment (NCIt, MedDRA, CDISC CT), and validation. Every extracted element is traceable to its exact source location, ensuring transparency and auditability, while a human review layer maintains quality and control.

Beyond extraction, a pluggable adapter layer generates EDC-ready study builds, leveraging sponsor and therapeutic area-specific form libraries to produce tailored eCRF designs aligned with platform-specific requirements. New EDCs need only a new adapter.

Recognized as a runner-up in the CDISC AI Innovation Challenge, our solution demonstrates a scalable approach to protocol digitalization. We share implementation insights, lessons learned, and key challenges, including interpreting complex study schedules, resolving footnote-driven logic, and handling legacy documents not originally designed for machine consumption.

Clinical protocol authoring remains slow, manual, and inconsistently integrated with downstream data systems. This abstract reveals the results of our experiments in building an AI-assisted digital-protocol authoring system aligned with CDISC USDM 4.0 and the ICH M11 Technical Specification (2025).

We implemented multi-agent architecture covering study planning, section authoring, data architecture, and quality validation using the Anthropic Python SDK. The system generated all ICH M11 sections from CT.gov and structured study-input, producing USDM-compliant JSON through an end-to-end automated workflow. A 21 CFR Part 11-compliant audit trail with SHA-256 hash chaining was validated across all generated sections. Practical challenges included mapping free-text clinical intent to USDM entity classes, containing agent hallucinations in regulatory contexts, and maintaining ALCOA+ data integrity throughout AI-assisted generation. Our experiments revealed that agentic-AI has real potential to support DDF adoption at a pharma organization; however human review of all AI-generated content remains a firm requirement.

The promise of CDISC Digital Data Flow is compelling: if study intent is captured early in structured, standards-aware form, downstream processes can become faster, more consistent, and less dependent on manual reinterpretation, with greater reliability and less rework. In practice, many organizations still struggle to turn digital protocols into tangible value across study build, monitoring, biometrics, and submissions.

This presentation shares EDETEK experience applying neuro-symbolic intelligence to bridge that gap. Using AXIS.AI, structured protocol representations are aligned with CDISC standards and transformed into executable logic that supports CRF alignment, SDTM mapping, monitoring readiness, compliance checks, and biometrics deliverable preparation. We will discuss how AI assists with interpretation and authoring, while deterministic, standards-based execution ensures consistency, traceability, and inspection readiness across downstream workflows. Attendees will gain a grounded view of how Digital Data Flow can evolve from structured documents into live, standards-executable intelligence that supports both operational execution and regulatory delivery.

The pharmaceutical industry faces increasing complexity and volume in regulatory submissions, driving the need for innovative documentation solutions. This paper will outline a framework utilizing Generative Artificial Intelligence (GenAI) to automate the creation of Data Reviewers Guides (DRGs), which provide crucial context for CDISC-compliant datasets submitted to regulatory agencies. GenAI employs Large Language Models (LLMs) with Retrieval-Augmented Generation (RAG) to generate content from reference documents, including study protocols, annotated case report forms, and statistical analysis plans. The paper demonstrates that GenAI streamlines the creation and maintenance of DRGs, significantly reducing manual workload while improving efficiency, consistency, and reliability for regulatory submissions such as FDA, PMDA, and NMPA.

Agentic AI systems observe, reason, and act autonomously toward a goal. What if we applied this framework to clinical trial data quality? Instead of manually configuring simulation parameters and hoping synthetic data satisfies hundreds of edit checks, an agentic system could analyze the rules, generate test data, evaluate violations, diagnose root causes, and adjust parameters iterating until convergence.

We present a working implementation of this concept: a self-teaching simulation engine that reverse-engineers CDISC-CORE and other quality rules into field-level generation constraints, runs optimization loops with learning rate decay and oscillation detection, and produces digital twins that satisfy conformance requirements with measurable convergence. The system maintains a complete audit trail of every parameter change, enabling human-in-the-loop oversight of autonomous decisions.

How can AI assist in data quality planning? By transforming quality rules from passive validators into active generation guides closing the loop between specification, simulation, and data quality.

Lilly is piloting a process to split Analysis Results DataSets (ARDS) and TFLs into separate deliverables.  Our new system treats creation of ARDS files in the same vein as CDISC datasets.  This system creates, stores, and maintains standards for SDTM, ADAM, and ARDS datasets.  Code is directly produced from these standards to create datasets that maintain consistency and traceability from data collection through analysis results.  The code will facilitate transparency since the configured standards will be automatically reflected in the generated code.  A separate TFL system consumes ARDS files and generates the displays.  TFL standards were re-written to use ARDS as input rather than ADAM making the TFLs a cosmetic step.  This enables a user-friendly GUI TFL tool to be created which focuses on displaying ARDS results as TFLs without concern for deriving the analysis results.  This presentation will report on the progress and findings of the pilot studies.

The clinical study data flow diagram might be described as a visual representation of how data (which includes data, documents, metadata, and other electronic content which is created and collected) moves through a clinical study's lifetime, from its initial capture to final archiving. It may include the processes, systems, and stakeholders involved in that journey.  The clinical study data flow diagram is increasingly being requested during audit and inspections to provide clarity on how study data is generated, processed, and managed across the clinical trial lifecycle. 

This presentation will include 2 speakers, the 1st speaker who will provide an overview of the clinical study data flow diagram, its purpose, scope, stakeholders, and contents, and the 2nd speaker, a sponsor, who will provide case study details of the clinical study data flow diagram incorporation into their clinical study management processes and audit/inspection related preparation and conduct activities.

Preparing TMF for regulatory inspection under aggressive timelines is often viewed as an impossible mission,particularly when documentation gaps, quality inconsistencies, and resource constraints collide. This session explores how a risk based approach combined with practical applications of Artificial Intelligence (AI) can transform TMF readiness from a reactive scramble into a focused, inspection ready outcome. 

This session is designed for inspection readiness teams seeking measurable, sustainable improvements in TMF readiness. Attendees will learn how AI can be leveraged to rapidly assess TMF completeness, identify high risk content, prioritize remediation efforts, and support quality oversight without compromising quality expectations. The presentation will demonstrate how AI enabled tools can assist with completeness assessments, trend detection, and inspection narrative preparation,complimenting human efforts rather than replacing it.

With the right strategy and controls in place, Mission Impossible is no longer impossible ,it is achievable, in part, through thoughtful and strategic use of AI. 

Strong sponsor oversight is essential to achieving an inspection-ready Trial Master File (TMF), yet many organizations struggle to move from intention to execution. This case study highlights how a sponsor's TMF evolved from limited oversight to inspection readiness within 12 months, supported by a successful audit and mock inspection.

The session will share a high-level view of the approach used, key challenges encountered, and practical lessons learned along the way. Attendees will gain insight into how focused oversight, collaboration, and process improvements can strengthen TMF quality and support sustainable inspection readiness.

Clinical platforms have evolved. The eTMF no longer operates in isolation; it now sits within an interconnected ecosystem that includes CTMS, EDC, ePRO, and other clinical systems generating data, milestones, and evidence directly tied to TMF expectations.

This convergence creates a governance challenge. These systems follow different workflows, data models, and ownership structures, while traditional TMF governance, focused on document filing and periodic review,was not designed to manage cross-system dependencies or upstream accountability.

This session examines how TMF governance must evolve in a connected clinical environment. We explore how the TMF Governance Committee's scope expands from document oversight to cross-application data flows, process alignment, and system interactions. We discuss shifting roles and accountability when upstream system activity drives TMF expectations. Finally, we redefine TMF completeness, moving beyond document counts to confirming that expected evidence exists, is progressing, or is justified using metrics, metadata, and system signals,enabling continuous oversight across an ecosystem.

Real-world evidence studies expose the limits of trial-centric eTMF models: manual filing, inconsistent metadata, weak traceability, and fragmented system handoffs. This presentation explores how a connected, product-driven approach can modernize TMF management for RWE. We will demonstrate how TMF standards, RWE-specific document frameworks, interoperable APIs, and FHIR-aligned data exchange can enable smarter configuration of expected content from study startup through inspection readiness. We will also examine practical AI use cases, including automated artifact classification, metadata recommendation, completeness forecasting, anomaly detection, and continuous quality monitoring, with human oversight built in. Attendees will leave with a blueprint for transforming eTMF from a document repository into an intelligent, interoperable platform that reduces operational burden, improves compliance, and scales with the future of digital clinical research.

The FDA, EMA and MHRA all publish and share inspection findings.

In this panel session hosted by Arkivum's Chris Sigley, Genmab's Traci Wendler and Hedian's Sarah Hitching will discuss common end of study inspection finding themes and what we can learn from them.

Pulling together an analysis of warning letters, inspection finding publications, freedom of information requests and first hand experience, the panel will discuss topics such as:

• End of study quality checks and missing records
• Gaps identified in end of study and retention related SOPs and processes (including their ongoing maintenance)
• TMF oversight and post trial data sprawl
• End of study TMF mismanagement including poor archiving processes
• Access to closed study records and data

We will also open out the panel for questions (and stories from the floor). This interactive session will share lessons learnt with the aim of helping improve future inspection outcomes.

Regulatory approval remains the ultimate goal for the majority of clinical trials, yet what appears complete at first glance may not fully withstand regulatory scrutiny. While implementation of CDISC standards has increased yearly and standards compliance has significantly improved, approval cannot be guaranteed based solely on standards and regulatory conformance.

Validation plays a critical role in confirming datasets meet predefined structural requirements. However, while rule-based checks ensure compliance, they may not reveal flaws in other aspects critical for regulatory review.

This presentation will examine common challenges encountered on the path to regulatory approval and consider how these challenges extend beyond traditional validation. The benefits of standardization in enabling consistency and interoperability will be highlighted, alongside areas where additional focus is needed. Finally, potential solutions will be explored, moving us beyond the illusion that CDISC and regulatory conformance alone ensure approval and closer to revealing workflows to truly vet our submissions.

Rare disease programs often progress through multiple studies built on differing CRF architectures, metadata assumptions, and derivation rules. The INH5, INH8, and INH7 hemophilia studies exemplify this challenge, as each was developed using distinct data collection frameworks that produced fragmented metadata, heterogeneous derivation logic, and inconsistent controlled terminology. The most impactful divergences involved non uniform ABR algorithms, differences in dosing schema, heterogeneous laboratory schedules and units, age calculations, and distinct eDiary workflows, all of which propagated into the extension study and resulted in mixed derivation rules and inconsistent assumptions across CSR packages.

These foundational setup issues also created substantial complications during ISS/ISE preparation, where cross study inconsistencies became more pronounced and required extensive remediation.
This paper provides a technical assessment of these architectural gaps and proposes a standards driven framework emphasizing unified CRF design, centralized metadata governance, harmonized derivation specifications, and proactive SDTM/ADaM alignment to ensure CSR ready, compliant outputs.

While the ICH E9(R1) estimands framework has reshaped clinical trial planning and analysis, practical implementations at the data‚ standards level remains limited, particularly within ADaM. The final PHUSE WP-092 white paper (2025), building on the draft published in 2023, provides consolidated recommendations to support implementation. Drawing on real‚ world study experience, this presentation describes our practical applications of the estimands framework in ADaM, by leveraging the existing CDISC standards in alignment with the latest PHUSE WP‚ 092 guidance. We demonstrate how key estimand attributes, including treatment, population, variable, intercurrent‚ event handling, and population‚ level summary, are systematically embedded within ADaM datasets to enable traceability. Particular emphasis is placed on implementing different intercurrent‚ event strategies across multiple estimands in ADICE, and on the impact on related ADaM datasets, including subject‚ level population derivations in ADSL. This work provides a concrete, submission‚ ready template for operationalizing PHUSE WP‚092 recommendations.

A discussion around how Evinova has used USDM to support designing studies and authoring protocols together in a single system. It will start with how the system is aligned to USDM. It will transition to discussing explorations for adapting USDM to support conceptual studies, study costs, and operational study data. It will close by discussing the end user challenges which are important to address for DDF to reach its potential.

The Digital Data Flow vision promises end-to-end automation grounded in USDM, but the path from unstructured protocols to conformant, operationally useful structured data is far more complex than it appears. Drawing on real-world production USDM use, this presentation maps the engineering and domain challenges organizations should anticipate: semantic ambiguity in protocol documents, the gap between valid USDM and useful USDM, the vocabulary alignment problems that emerge at scale, and the validation rigor required before downstream systems can reliably consume structured protocol data. We share what the ecosystem still needs (from tooling gaps to standard extensions) and illustrate the downstream workflows that become possible when these challenges are solved: automated consent generation, screening derivation, and cross-system interoperability.

Honestly, we thought structured protocols would be the easy part. The technology was ready USDM, ICH M11 alignment, API driven study builds. What caught us off guard was everything else. This presentation walks through what actually happened when we moved from Word-based protocols to DDF-enabled digital authoring. Some things worked immediately: syntax templates caught ambiguities our teams had been reconciling manually for years. Semantic search surfaced prior studies we had forgotten existed. Automated EDC configuration cut weeks from our timelines. Other things required iteration‚ getting authors comfortable with structured fields, convincing stakeholders that upstream investment pays downstream dividends, integrating with systems that weren't quite ready. We share real metrics alongside real friction points. No theoretical frameworks here; just practical lessons from an implementation still evolving. If you're considering DDF adoption or stuck midway through, this session offers a candid look at the path forward.

Clinical research sites are increasingly adopting their own technology, EHRs, eSource tools, ambient voice transcription, creating both opportunity and friction when integrating data into sponsor systems. While CDISC standards provide a foundation, variations in data models, terminologies, and business practices across therapeutic areas and geographies expose real gaps, particularly as new capture modalities emerge that current standards don't fully address.

This session explores the digitization of site operations and what it means for data interoperability. Key topics include:

1. Mapping disparate site data models to CDISC standards
2. Harmonizing terminology across distributed networks
3. Extending CDISC to support new data sources like ambient voice
4. Real-world use cases from inside pharma

The goal is open dialogue, identifying gaps, accelerating standards development, and inspiring collaboration toward faster, higher-quality data with less site burden.

CDISC Biomedical Concepts(BCs) offer a powerful foundation for connecting clinical research metadata across the data lifecycle. This presentation describes an implementation that positions BCs as the foundational semantic layer of an eSource platform, enabling automated EHR-to-EDC data transfer, AI-Grounded clinical data extraction, standards-driven data visualization and metadata-driven study configuration from a single concept library.

Our approach organizes BC metadata as a knowledge graph, capturing relationships between data element concepts across domains and built on a standards-agnostic canonical layer inspired by triangulating across FHIR, CDASH, OMOP and other reference models. This graph structure, combined with crosswalks to healthcare and research data standards, makes all captured data semantically coded and machine-readable, regardless of study-specific CRF structure or sponsor naming conventions. This allows us to automate transfer and normalization of EHR data into CRF data. That combined with our study builder tool, this also enables us to map EHR data to any eCRF.

As clinical research moves toward interoperability and automation, aligning healthcare data standards with regulatory submission requirements remains a challenge. HL7 FHIR is increasingly adopted across EHR systems to enable standardized access to real world clinical data, while CDISC standards (SDTM and ADaM) continue to underpin regulatory submissions. Bridging these paradigms is essential for enabling end-to-end digital clinical trials.

This presentation outlines a standards-based framework for implementing FHIR to CDISC mappings aligned with emerging CDISC and HL7 guidance. It demonstrates the transformation of core FHIR resources to SDTM domains and emphasizes reuse of CDISC metadata, alignment with CDASH, and integration with existing sponsor pipelines.

The framework addresses key challenges including semantic alignment, variability in FHIR implementations, and governance, with mitigation strategies aligned to evolving regulatory expectations. It further highlights benefits such as reduced site burden, minimized redundant data entry, accelerated study startup, and support for decentralized and real-world data enabled trials.

ICH E6(R3) guideline introduces a modernized, risk-based framework for quality management that aligns nicely with the forthcoming TMF SM V1. Together, these both create unique opportunities to transform TMF management into a more data-driven, risk-informed discipline.

This session examines how risk-based principles from ICH E6 can be systematically integrated with TMF standardization efforts to support inspection steadiness. The presentation will explore how standardized TMF metadata, taxonomy, and controlled terminology can be leveraged to identify critical-to-quality  record types, quantify risk, and enable proactive quality oversight.

Attendees will gain practical perspectives on aligning Quality Risk Management (QRM) methodologies with standardized TMF structures, including the use of metrics that matter within a harmonized data framework. The session will also address how structured TMF data can enhance transparency, facilitate analytics, and support regulatory expectations for traceability and inspection readiness in a risk-proportionate environment.

Digital transformation of TMF is as much about mindset as it is about technology. This session examines how organizations can embed risk-based thinking into everyday TMF activities while scaling engagement across global teams. Leveraging AI and standards, we will show how connected systems provide real-time visibility into TMF health and risk, enabling smarter decisions at every level. The discussion will also highlight the critical role of leadership, emotional intelligence, and change management in driving adoption and sustained performance. Attendees will leave with a practical framework for building a connected TMF culture where risk awareness, collaboration, and continuous improvement are deeply ingrained.

Is your eTMF truly inspection-ready, or are you waiting for issues to surface before taking action? Much like routine health checkups, eTMF Health Assessments offer a proactive approach to maintaining quality, identifying risks early, and ensuring ongoing compliance throughout the study lifecycle.  

In this session, we'll explore how regular, risk-based reviews can drive deeper, targeted assessments, using data and insights to prioritize high-risk areas and focus resources where they matter most. Through structured reviews and trend-based insights, attendees will learn how Health Assessments provide meaningful metrics on timeliness, quality, and completeness, enabling teams to identify early warning signs before they impact study milestones or inspection readiness.

By shifting from reactive remediation to continuous oversight, organizations can strengthen inspection readiness, reduce rework, and embed sustainable quality practices into their processes. Moving beyond a reactive mindset makes compliance a built-in foundation that supports long-term success.

AI has significant potential to transform TMF management , but confidence and compliance don't come from the technology. They come from the governance foundation built around it.

Two regulatory developments are raising the bar for what TMF teams must demonstrate. ICH E6(R3) expands essential records to include communications, decision rationale, and electronic data regardless of where they live. FDA's Remote Regulatory Assessment guidance creates mandatory response windows as short as 15 calendar days. Together, they mean TMF teams are accountable for surfacing evidence scattered across systems no single eTMF can see.

Enterprise AI, deployed as a governed knowledge synthesis layer across an eTMF and connected systems, enables teams to meet these demands , replacing days of manual searching with real-time, sourced, traceable answers that free TMF professionals for higher-value work. Real-life use cases will demonstrate governance methodology in practice, showing how AI-assisted TMF activities remain compliant, traceable, and inspection-ready.

As AI becomes embedded in eClinical platforms, organizations must balance innovation with strict regulatory requirements. Unlike deterministic software, AI introduces probabilistic behavior, creating challenges in validation, traceability, and oversight.

This session examines practical approaches to deploying AI in a validated state within regulated environments. Drawing on frameworks such as ICH E6(R3), FDA AI/ML guidance, ISO 42001, and the EU AI Act, it outlines strategies for ensuring compliance across the AI lifecycle.

Topics include validation of non-deterministic models, human-in-the-loop controls, model monitoring and drift prevention, and ensuring transparency and reproducibility. Real-world use cases, including eTMF auto-classification and clinical data automation, demonstrate how organizations can maintain inspection readiness while scaling AI adoption.

Attendees will gain insight into governance models, risk-based validation approaches, and methods to operationalize trustworthy AI without compromising compliance.

At post-export TMF custody boundaries (e.g., submission, archival, or CRO handoff), integrity is typically demonstrated through vendor attestations, re-exports, or manual reconciliation. These methods indicate what should have happened rather than what occurred at the byte level.

We evaluated a system-independent, zero-custody verification model for post-export TMF evidence. In a blinded, multi-operator study of 230,253 evidence files and 21,966 evidence-set fingerprints, tamper detection achieved 100% sensitivity (879/879 altered items detected) with no false positives (229,374/229,374 unaltered items confirmed). Mean verification time was 0.076 seconds per file and scaled with tamper burden (r=0.91, p<0.0001), but not corpus size for proof-of-unchanged verification (r=-0.01, p=0.93).

The model runs locally within the custodian’s environment, with no data or identifiers leaving the institution. Hash-only export packets enable independent offline re-verification without vendor access, supporting ICH E6(R3) inspection readiness.

Derivation logic for CDISC SDTM is increasingly delivered as reusable functions within standardization engines. While this modular strategy enhances consistency, authoring high-quality functions remains a significant investment due to variability in clinical data structures, conventions, therapeutic areas, and protocols.

This paper introduces a human-in-the-loop, multi-agent AI workflow that accelerates SDTM Engine function development and validation. Specialized agents handle requirements interpretation, code generation, test-case design, and documentation, each under developer-led review. We demonstrate the approach using functions covering reference-data lookups, conditional transformations, and data-integrity validation, verified through regression suites spanning functional, negative, and edge-case scenarios.

To support adoption, a complementary RAG-based assistant provides context-aware guidance on function selection, configuration, and troubleshooting from a Git-based knowledgebase.

A tiered governance model master-prompt constraints, developer review, and independent validation ensures AI accelerates delivery without compromising quality or traceability. Initial results show meaningful reductions in iteration cycles and improved throughput.

SDTM implementation relies on human-readable mapping specifications that are valuable for collaboration but difficult to verify consistently before execution. Downstream standards such as Define-XML and CORE validate final deliverables, yet many organizations lack a formal, machine-verifiable layer for the transformation logic itself.

We describe a specification-driven approach grounded in production experience at Merck processing approximately 4,000 SDTM updates annually. Transformation intent is captured as structured (machine-readable) specifications declaring source fields, target fields, approved versioned capabilities, and dependencies. A validation layer checks technical executability, governance compliance, and domain correctness before any pipeline runs. AI assists with authoring candidate specifications but remains subject to the same validation boundary as human-authored work.

The approach complements Define-XML and CORE by formalizing upstream transformation intent. We discuss the validation model, governed function registries achieving 60‚ 80% cross-study reuse, controlled AI integration, and implications for CDISC standards alignment.

Many organizations invest heavily in building a CDISC standards library with expectations of compliance, improved quality, and efficiency. While compliance and quality are often achieved, true efficiency remains elusive when standards function merely as static references rather than reusable, governed building blocks.

Alcon addressed this challenge by pairing strong governance with automation to drive high standards reuse across Pharma and Medical Devices. Automated comparison of study CRFs against the standards library and even against completed studies enables intuitive reuse recommendations, minimizes deviations, and reduces manual effort. Embedded quality checks within the standards library further accelerate SDTM and ADaM development and enhance cross-study consistency.

This presentation will describe how Alcon cultivated an enterprise-wide standards mindset and built a standards-powered ecosystem. Attendees will gain practical insights into how governance, automation, and actionable reporting can dramatically increase standards reuse, improve data quality, and unlock the full value of CDISC standards implementation.

As clinical research advances toward connected digital ecosystems, organizations must modernize data management through automation, artificial intelligence, and CDISC standards. Aligned with the theme The Future is Connected, this presentation outlines an end to end framework that integrates Digital Data Flow principles with metadata driven study builds and AI enabled validation workflows. By linking protocol design, case report form development, SDTM mapping, and ADaM generation within a unified standards based architecture, teams can significantly reduce manual effort and improve data quality.

The session highlights practical use cases where AI powered rule engines and reusable mapping libraries improved submission readiness and reduced validation findings. Integration of decentralized trial data and real world sources further demonstrates how interoperability strengthens regulatory compliance and accelerates database lock timelines. This approach positions CDISC standards not only as compliance tools but as strategic enablers of scalable digital transformation.

Clinical trials are at an inflection point, where static standards must evolve into intelligent, executable frameworks. Emerging CDISC standards, including USDM, Biomedical Concepts, Analysis Concepts (AC/DC), and Analysis Results Standards (ARS), provide the foundation for this transformation.

This paper demonstrates how these standards move beyond compliance artifacts to become the backbone of agentic AI systems. Biomedical Concepts and USDM enable computable study definitions, while AC/DC and ARS formalize analysis intent and outputs, collectively powering AI-driven automation of SDTM, ADaM, TFL, and CSR generation, along with protocol optimization and real-time insights.

Through real-world implementations, we showcase software solutions that generate SAS and R code from standards-driven metadata, reducing manual effort, improving traceability, and enhancing quality. Aligned with the CDISC 360i vision, we demonstrate purpose-built solutions connecting study design through to submission.

With the FDA signaling potential adoption of CDISC Dataset-JSON v1.1, sponsors, standards leaders, and technology teams need to understand both the strategic importance and the implementation realities of this emerging submission format. Designed as a modern alternative to SAS XPT, Dataset-JSON enables more structured, machine readable data exchange, richer metadata representation, and improved compatibility with contemporary analytics and review workflows.

This presentation provides a practical overview of why Dataset-JSON matters, what capabilities it introduces, and what organizations should be doing now to prepare. Based on experience from the PHUSE and CDISC joint pilot and informed by William Qubeck's leadership of the original Define-XML initiative, the session explores implications for standards governance, metadata alignment, legacy process integration, automation, traceability, and GxP validation. Attendees will gain a pragmatic view of the road ahead and a framework they can use to support internal planning, cross functional alignment, and modernization of submission ecosystems.

The SAS Version 5 transport file has been the industry standard for two decades, but its limitations 8-character variable names, two data types, no embedded metadata, no compression, and software dependency are increasingly incompatible with current and future clinical trials involving wearables, genomics, and EHR-linked real-world data. The evolving AI regulatory environment demands data exchange formats that are queryable, semantically rich, and metadata complete.

Rather than a single replacement, next-generation submissions in 2041 will involve a use-case-driven strategy (Dataset-JSON for SDTM/ADaM, Parquet for sensor data and Fast Healthcare Interoperability Resources for real-world data) for accelerated sponsor preparation and regulatory review. Open-source validators and converters can reduce adoption barriers. We explore advanced data exchange formats against next-generation submission requirements, including data type richness, metadata integration, compression, and queryability. We also outline a phased, backwards-compatible adoption path toward continuous, API-driven regulatory data exchange.

Safety data pipelines have historically tolerated semantic drift — safety intent defined in protocols is silently reinterpreted at each downstream handoff, eroding consistency by submission time. This paper describes an enterprise safety standards initiative that addressed this challenge by anchoring core safety objectives to CDISC Biomedical Concepts (BCs), enabling meaning to travel intact from protocol through SDTM.

Enterprise-level safety endpoints, linked to standard study activities and BCs, drove eCRF design and SDTM specification without study-specific reinterpretation. BC identifiers embedded across artifacts enabled automated semantic validation, surfacing misalignments before first patient. The initiative progressed through three BC maturity levels, consistent definition, semantic reuse, and executable infrastructure with cultural adoption proving the most significant barrier.

This pattern establishes a replicable model for Digital Data Flow, where safety intent becomes machine-readable, reconciliation moves upstream, and consistency is achieved by design rather than manual effort.

We need standardization for the Investigator Site File. In this session, hear one of the co-leads for the CDISC ISF Reference Model initiative talk about this pivotal project and how it's transforming the industry.

From aligning global stakeholders to overcoming implementation challenges, this session offers a look at how the ISF Reference Model V1 Provisional is bringing consistency, clarity, and compliance to site-level documentation.

What attendees will learn:

• What the ISF Reference Model V1 Provisional is and how it was developed
• How it aligns with the Trial Master File Reference Model
• Overcoming common challenges with implementation

Panelists Invited

This presentation will give an overview of the different activities that are being undertaken or that are planned as part of the TMF Reference Model Standardization. A presentation of the Standardization roadmap will be followed by presentations of the different workstreams including DDF/ICH M11 Alignment, metadata model development, tools development and controlled terminology.