AI (Artificial Intelligence) in a GxP Environment + AI in Visual Inspections - Live Online Training

Programm

Gesamtes Programm als PDF herunterladen

• History of AI
• Types of AI
• Real life examples

• Technological basics
• Different learning / training methods
• Example use cases

• Introduction to deep learning models for imaging
• Deep learning for diagnosis and prognosis
• Pre-training of deep learning models
• Explainable Artificial Intelligence

• Challenges for the Life science industry
• The GAMP-perspective on AI/ML
• Use cases / Known scenarios

• Pharmaceutical laws
• EU-GMP Guide Annex 11
• Inspection strategy
• What do inspectors expect from the regulated user?

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

• Overview of model and data types
• Data split: training, validation, testing
• Data quality, representativeness and typical data challenges
• Use of synthetic data

• Introduction to Large Language Models (LLM) and example use cases
• Specialization and tailoring of LLMs in computerized Systems
• Typical risks when using LLMs
• Performance evaluation and validation strategies for LLMs

• AI and law: introduction and overview
• Outlook and update on the EU AI Act
• Copyright and trade Secrets
• Data protection requirements and processing of personal data
• Practical implementation and best practicess

• Maturity: Increasing autonomy and transferring Control
• Governance: Developing and operating AI solutions in GxP-regulated areas

• Risks and Opportunities
• Preparing the Quality Management System for AI
• Where to implement and what to avoid
• Pactical examples and attention Points
• Validation requirements
• Future Developments

• Basics of a ML Risk and Control Framework
• Applying QRM to development and operation of AI applications
• Using hazard clusters to guide the risk process

• Lessons learned from primary and secondary packaging lines and Auto-Injectors production
• When continuously learning and situationally acting tools can help and when not
• Introduction of Use cases for live learning optimization tools in GMP-environment
• Short Intro: AI-based behavior learning on high frequency machine data of whole production lines
• Obstacles during validation
• Learnings from 24/7-operations Integration
• Review of realistic and unrealistic benefits

• Black box or partner: How can transparency of AI be increased?
• Interpretability of AI-based decisions
• Foundations of Explainable AI
• Approaches for the validation of AI systems in GxP environment

• What is Artificial Intelligence?
• Is GenAI a Game Changer?
• Opportunities and risks
• Industrial Use cases

• Legal basis
• GAMP® and AI (ML)
• Validation
• Operation and raw data

• Development of robust, reliable and production-ready models in 4 phases
• Phase 1: Problem identification & description
• Phase 2a: Specification of inspection concept
• Phase 2b: Definition of the sample sets (artificial and production samples), creation of the datasets, clarification of the labelling strategy
• Phase 3: Model design, training and verification - a risk-based approach
• Phase 4: Qualification & validation
• Processes & technologies
• Technologies for efficient image data acquisition, variable model technologies, transfer learning / pre-trained models, labelling application
• Documentation of model development: traceability, risk minimisation and build-up of confidence

• Manual, semi-automated and fully-automated approaches
• Use of Artificial Intelligence
• Single chamber and multi-chamber bags
• Inspection of Blow-Fill-Seal containers
• Inspection of Form-Fill-Seal containers
• General approach
• Training and Machine Learning
• Testing and Validation
• Limitations