Date: September 23, 2025 Location: AstraZeneca, One MedImmune Way, Gaithersburg, MD 20878 Format: Hybrid (In-person: ~45 attendees, Online: 5 attendees)

Event Overview

The 2025 Statistical Innovation Community Summit convened cross-industry heads and experts of statistical innovation/methodology groups to share operating models, impact strategies, and AI/ML adoption approaches. Representatives from nearly all major pharmaceutical companies participated in this collaborative forum.

Program Agenda

Session 1: From Support to Leadership

Moderators: Kristine Broglio, Elena Polverejan

Key Challenges Identified

  • Resource Constraints: Limited resources and high workload pressures
  • Impact Demonstration: Difficulty in clearly demonstrating business value
  • Visibility Issues: Inconsistent visibility and recognition within organizations
  • Engagement Models: Optional engagement leading to missed opportunities
  • Attribution Gaps: Challenges in proper recognition and credit attribution

What Leaders Value

Leaders prioritize measurable outcomes including:

  • Business Impact: Time saved, cost reduction, patients helped, probability of success (PoS) improvements
  • Adoption Outcomes: Concrete evidence of method implementation and uptake
  • Visibility Metrics: Governance presence and leadership recognition

Successful Operating Models

  • TA Liaisons: Embedded workflows with regular touchpoints with therapeutic area heads and product leads
  • Optional Consultancy: Effective where high trust relationships exist
  • Governance Presence: Prioritizing formal representation in decision-making processes

Building Visibility and Culture

  • Co-presentation Strategy: Present jointly with product and clinical teams
  • Regular Reporting: Biannual impact reports delivered to R&D leadership
  • Internal Education: Lunch & Learn sessions and bite-sized educational modules

Recognition and Attribution Strategies

  • Early Charter Roles: Establish clear roles early in complex initiatives
  • Inclusive Governance: Ensure innovation contributors participate in governance structures
  • Awards and Recognition: Make innovation statisticians eligible for patents, publications, and organizational awards

Session 2: Maximizing Impact via Cross-Functional Collaboration

Moderators: Binbing Yu, Qinghua Song

Early Engagement Strategies

  • Protocol Optimization: Focus on study design improvements before operational commitments
  • Fit-for-Purpose Science: Ensure scientific approach matches business objectives
  • Communication: Use visuals and minimize jargon to create shared understanding

Collaboration Structures

  • Department-wide Reviews: Protocol and Statistical Analysis Plan (SAP) review forums
  • TA/Product Liaisons: Dedicated representatives for therapeutic areas and products
  • Working Groups: Specialized teams for payer evidence, biomarkers, and dose optimization with quarterly cross-functional updates

Education and Outreach

  • Leadership Workshops: Design simulations targeted at leadership decision-making
  • Cross-functional Training: Education programs for non-statisticians
  • Office Hours: Regular availability for consultation and support
  • External Engagement: Joint presentations with clinicians and participation in scientific consortia and ASA Biopharm groups

Prioritization Framework

  • Portfolio Alignment: TA-level strategic alignment with business priorities
  • High-Impact Focus: Concentrate on high-impact assets and resource-constrained programs
  • Pilot Approach: Start with receptive stakeholders, then standardize successful approaches

Recognition Management

  • Charter Documentation: Prevent visibility loss through clear roles and responsibilities
  • Credit Balance: Fair attribution between product statisticians and innovation teams
  • Team Culture: Foster collaborative “team win” mentality

Session 3: Embracing AI/ML in Statistical Innovation

Moderators: Demissie Alemayehu, Li Wang

Mindset and Foundations

  • Statistical Grounding: AI/ML approaches rooted in solid statistical theory
  • Human-in-the-Loop: Adopt validation approaches that maintain human oversight
  • Statistical Strengths: Leverage statisticians’ expertise in uncertainty quantification and decision framing

Key Adoption Areas

Design and Analysis

  • Risk scores for randomization optimization
  • Transformer-enhanced Cox models
  • Integrated operating characteristics

Patient Stratification

  • Advanced subgroup identification
  • Personalized treatment approaches

Real-World Evidence (RWE)

  • Digital twins for external control development
  • Enhanced observational study designs

Operational Efficiency

  • Recruitment projections and site feasibility
  • Signal and anomaly detection
  • Process automation including context-aware coding
  • Protocol, SAP, and Clinical Study Report (CSR) drafting under appropriate guardrails

Practical Considerations

  • Data Constraints: Address pharmaceutical industry’s unique data scale limitations
  • Domain Expertise: Leverage domain-structured machine learning approaches
  • Consortium Data: Promising but constrained by anonymization and standardization requirements
  • Skill Development: Python proficiency, interpretability skills, prompt engineering, and multi-model benchmarking

Validation and Governance

  • Uncertainty Quantification: Use confidence intervals, prediction intervals, and resampling methods where feasible
  • Documentation: Maintain clear records of environment and parameters
  • Prototyping: Begin with user-led prototypes before scaling
  • IT Collaboration: Scale with IT support while avoiding architecture-first forcing

Organizational Integration

  • Strategic Positioning: Position innovation teams as statistical data science owners in clinical development
  • Training Expansion: Broaden ML education across teams
  • Cultural Management: Address pushback (e.g., concerns about AI-generated SAPs) by framing AI as freeing statisticians for higher-order design and decision work

Community Development and Next Steps

ASA Biopharm Working Group - Statistical Innovators in Medical Product Development

  • Formalization: Cross-industry working group proposal reportedly approved with formal listing pending
  • Leadership Structure: Proposed chair leadership to be confirmed with flat organizational structure encouraging collective contributions
  • Timeline: Finalization planned before ASA leadership transitions in 2026

Efficiency+ Working Group (ASA Biopharm)

  • Scope: End-to-end optimization across design, feasibility, operations, and supply chain
  • Approach: Iterative modeling across traditional silos
  • Leadership: Statisticians leading quantitative decision frameworks

Action Priorities

The summit identified key action items for the statistical innovation community:

  1. Metrics Definition: Define leadership-valued metrics tied to business impact and organizational visibility

  2. Institutional Models: Institutionalize therapeutic area liaison models and ensure governance presence for innovation statisticians

  3. Education Scale-up: Expand training for non-statisticians and conduct leadership workshops with simulated decision-making scenarios

  4. AI/ML Implementation: Pilot AI/ML use cases with appropriate guardrails and publish internal standards for human-in-the-loop validation, uncertainty quantification, and documentation

  5. Technology Partnership: Partner with IT on digital data flow initiatives (ICH M11/USDM) and user-led prototypes for specification-to-code automation and design operating characteristics

  6. Community Building: Complete ASA Biopharm working group establishment, plan 2026 summit and interim sessions, and share case studies and templates across companies

Key Takeaways

The 2025 summit advanced the statistical innovation community’s understanding of:

  1. Leadership Transition: Moving from support roles to leadership positions within organizations
  2. Cross-functional Integration: Essential strategies for effective collaboration across pharmaceutical development teams
  3. AI/ML Integration: Practical approaches for incorporating artificial intelligence and machine learning while maintaining statistical rigor
  4. Community Formalization: Progress toward formal recognition through ASA Biopharm working groups
  5. Measurable Impact: Importance of defining and tracking metrics that demonstrate business value

Looking Forward

2026 Summit: Planning underway for the next Statistical Innovation Community Summit with interim sessions to maintain momentum and community engagement.

The 2025 summit continued the evolution of the statistical innovation community, focusing on practical implementation strategies, technological advancement, and formal recognition within the broader statistical and pharmaceutical ecosystem.