Annex 22 / AI Conference 2025

Programm

• General introduction
• (Very) brief introduction to AI & ML
• Drivers (for using AI & ML in pharma)
• Regulations and guidances

• Current situation
• What do you need to watch out for?
• What are the risks?

• Overview of supporting processes: data management, risk management, change management
• Have documentation ready – provide reasoning and justifications
• How to setup mock inspections successfully

• Regulations and the degree of freedom
• What you may want to consider
• Examples and evaluation criteria

• Understand each key step in building an AI model for imaging: data preparation, training, pre-training, testing, and explainability
• Learn through a real-world example: cancer diagnosis and prognosis based on CT scans
• Explore what’s next: opportunities and risks of using large pre-trained multimodal AI models
• Engage actively: think, ask, discuss - your participation shapes the session
• Leave with a clear understanding of how AI can support your own imaging challenges

• Introduction to ChatGPT and current Large Language Models
• Overview of prompt engineering
• Prompt techniques
  • Zero-shot prompting, few-shot prompting
  • Chain-of Thought
  • Tree of Thought (ToT)
  • Reverse Engineering Prompting
  • Example use cases

• Detection of clusters based on Natural Language Processing
• Validated AI tools in GxP Environment
• Human centric approach to ensure quality and Control
• Risk based approach to reduce work-load for investigators

• Traditional image processing and its limitations
• Opportunities through the use of artificial intelligence
• Expectations, risks, and requirements for the use of AI
• Planning, Implementation, Operation & Monitoring – a life-cycle approach following the risk-knowledge-infinity cycle
• Documentation of model development: traceability and risk mitigation

• Overview: Evolution of AI coding agents
• State-of-the-Art: Current capabilities and limitations
• Use Cases: Beyond coding
• Outlook & Transfer: Future vision and imaging agent capabilities

• Deviation management challenges in pharma
• AI-powered solution: Knowledge Graph / LLM RAG
• Key features, benefits, and impact
• Demo and use cases

• Risk-based validation strategies for LLM-based systems
• Defining appropriate performance metrics for mixed-content data
• Human-in-the-loop controls to ensure accuracy and compliance
• Considerations for traceability, auditability, and change management

• How we can learn from other industries: Conducting AI-based compliance audits in the banking & insurance sector (example: DORA regulation)
• Goal: The way to systematic and GDPR-compliant document checks without language barrier
• Solution: Fast-track AI-supported inspections of quality management systems in GMP systems
• Why auditors are still mandatory: Checking the implementation status of written standards
• AI never sleeps: Failure-free creation of risk-based compliance inspection reports
• Verification versus GMP validation: The limitations of AI tools in GxP practice