Masterclass: Market Authorisation of AI-enabled Medical Devices Training Course ★★★★★

Introduction to Artificial Intelligence

This session introduces the fundamentals of artificial intelligence (AI), including how AI systems work, where they are used in healthcare, and the opportunities and challenges they present. Participants will explore the core principles of AI systems, from data-driven learning to reasoning and decision-making, and understand the algorithms that enable machines to perceive, learn, and support clinical and operational tasks.

The session also examines AI risks and the importance of responsible development and deployment. Participants will learn how AI has evolved, the disciplines that underpin modern AI technologies, and why regulatory and ethical considerations are essential to ensuring patient safety.

Topics include:

• What artificial intelligence is and how it is used in healthcare
• Key AI concepts, terminology and scope
• Machine learning, deep learning and natural language processing
• Common AI risks, including privacy, safety, security, reliability, interpretability, transparency, accountability, sustainability and human oversight
• Principles of responsible AI
• Historical development of AI and emerging trends

AI legislation as a global power play

Artificial intelligence is becoming a strategic driver of economic competitiveness, technological leadership and geopolitical influence. As governments develop regulatory frameworks, they also shape international standards, market access, and innovation ecosystems.

This session examines the geopolitical forces influencing AI legislation and explores how different regulatory approaches affect the healthcare sector. Participants will gain insight into the role of governments, standards organizations, and major technology companies in shaping global AI policy, as well as the implications for manufacturers of AI-enabled medical devices.

Topics include:

• AI regulation as a strategic policy instrument
• The Brussels Effect and the Beijing Effect
• Horizontal and sector-specific approaches to AI legislation
• Digital sovereignty and international competitiveness
• The role of standards and open-source software
• The influence of major technology companies on AI regulation
• Case study: AI supply chain resilience under export controls and cloud access restrictions
• Future developments in global AI governance

Best practices for developing AI-enabled medical devices

This session translates regulatory expectations into practical development activities. Drawing on international regulatory frameworks and real-world submission experience, participants will examine the full development life cycle of AI-enabled medical devices, from defining the intended use and algorithm through post-market monitoring.

The session emphasizes robust data governance, clinical evidence, statistical validity, and quality management throughout the product life cycle. Participants will also discuss common regulatory deficiencies and practical approaches to avoiding them.

Topics include:

• Regulatory foundations
  •    Regulatory obligations for AI-enabled medical devices 
  •    Applying AI legislation within existing medical device frameworks 
  •    Lessons learned from regulatory submissions 
  •    Common regulatory deficiencies and how to avoid them 
• Defining the AI solution
  •    Defining the intended use and clinical purpose 
  •    Describing AI use cases and algorithms 
  •    Assessing clinical impact 
• Data governance
  •    Data origin and acquisition 
  •    Data identity and traceability 
  •    Data access and collection 
  •    Data quality and integrity 
  •    Data annotation 
  •    Data protection 
  •    Dataset relevance and independence 
  •    Bias and fairness 
  •    Dataset shifts over time 
  •    Data limitations 
  •    Reference standards 
  •    Data augmentation 
  •    Testing datasets 
• Model development
  •    Training AI models 
  •    Model tuning 
  •    Performance testing 
  •    Validation strategies 
• Clinical evidence
  •    Clinical evaluation 
  •    Statistical analysis 
  •    Generalizability 
  •    Subgroup analyses 
  •    Statistical power 
• Human factors
  •    Usability engineering 
  •    Interpretability 
  •    Explainability 
• Post-market activities
  •    Performance monitoring 
  •    Model updates 
  •    Complaint handling 
  •    Post-market surveillance 
• Clinical applications
  •    Diagnostic AI 
  •    Prognostic AI 
  •    Predictive AI

Including practical examples, regulatory case studies, and discussion of implementation challenges.

Participants will work through the development life cycle, applying best practices to realistic regulatory scenarios and discussing practical considerations for preparing AI-enabled medical devices for market authorization.

EU Artificial Intelligence Act

The European Union Artificial Intelligence Act (EU AI Act) is the world's first comprehensive regulatory framework for artificial intelligence. It introduces new obligations for organizations that develop, deploy, and manage AI systems, including manufacturers of AI-enabled medical devices.

This session provides a practical introduction to the EU AI Act and its implications for the medical device sector. Participants will understand the Act's risk-based approach, economic operators' responsibilities, and the relationship between the AI Act and existing sector-specific legislation. The session translates legal requirements into practical regulatory, technical, and organizational considerations that support implementation.

Topics include:

• Overview of the EU AI Act
  •    Objectives and scope of the EU AI Act 
  •    Risk-based regulatory framework 
  •    Impact on software development processes 
  •    Implications for AI-enabled products and portfolios 
• AI system classification
  •    Prohibited AI systems 
  •    High-risk AI systems 
  •    General-purpose AI (GPAI) models 
  •    Limited-risk and minimal-risk AI systems 
  •    Practical examples and classification exercises 
• Roles and responsibilities
  •    Obligations of providers 
  •    Obligations of deployers 
  •    Importers and distributors 
  •    Authorized representatives 
  •    Responsibilities of subcontractors and research partners 
  •    Transparency requirements 
  •    Human oversight 
  •    Logging and record-keeping 
  •    Post-market monitoring 
• Interaction with other legislation
• General-purpose AI models
  •    Integration of GPAI models into medical devices 
  •    Responsibilities of GPAI providers 
  •    Responsibilities of medical device manufacturers 
  •    Conformity assessment considerations 
  •    Technical documentation requirements under the AI Act and MDR/IVDR 
• Standards and implementation
  •    Harmonized standards 
  •    Codes of practice 
  •    Technical documentation expectations 
  •    Preparing for conformity assessment 
• Organizational readiness
  •    Governance structures 
  •    Process adaptation 
  •    Resource planning 
  •    Collaboration with notified bodies 
  •    Preparing quality management systems 
• Notified bodies
  •    Role of notified bodies under the AI Act 
  •    Conformity assessment of high-risk AI systems 
  •    Expectations for technical documentation 
  •    Quality management systems 
  •    Post-market obligations 
• Governance and implementation timeline
  •    Key milestones 
  •    Transition periods 
  •    Future regulatory developments

Including practical interpretation of the legislation, regulatory examples, and discussion of implementation challenges.

Participants will examine practical scenarios, review regulatory expectations and discuss how organizations can prepare for compliance while integrating AI into existing medical device development processes.

Implementing the EU Artificial Intelligence Act

This session presents a practical implementation roadmap for medical device manufacturers and other economic operators. Participants will work through the activities, timelines, and organizational considerations needed to comply with the EU AI Act and integrate its requirements into existing quality and regulatory systems.

Topics include:

• Implementation roadmap
• Organizational responsibilities
• Gap assessment
• Compliance planning
• Documentation requirements
• Quality management integration
• Implementation timelines
• Practical implementation checklist

AI standards

Standards are becoming increasingly important in demonstrating compliance with AI and medical device regulations. However, not all published standards carry the same regulatory significance or provide equal value during product development and conformity assessment.

This session examines a selected set of AI standards that have attracted particular attention from regulators and notified bodies. Participants will learn how to distinguish between standards essential for regulatory compliance and those that provide supporting guidance, enabling them to make proportionate, well-informed implementation decisions.

Topics include:

• Regulatory significance
  •    Standards recognized by regulators 
  •    Standards commonly referenced by notified bodies 
  •    Standards supporting conformity assessment
• Practical application
  •    Selecting standards appropriate to the intended use 
  •    Understanding overlaps and complementary requirements 
  •    Identifying limitations of individual standards 
  •    Balancing regulatory expectations with practical implementation
• Implementation strategy
  •    Prioritizing standards 
  •    Developing a proportionate standards strategy 
  •    Supporting technical documentation 
  •    Preparing for regulatory review 

EU Support Structures for AI Medical Devices

Europe is developing an extensive ecosystem of initiatives that support the development, validation, and deployment of AI-enabled medical devices. These initiatives provide access to datasets, testing environments, regulatory expertise, technical infrastructure and collaborative networks.

This session explains how manufacturers can use these support structures throughout product development while recognizing that participation does not replace regulatory requirements for market authorization.

Topics include:

• Data laboratories
• Data intermediation services
• Data altruism organizations
• Testing and Experimentation Facilities (TEF Health)
• AI Factories
• European Digital Innovation Hubs
• Union AI testing support structures
• Regulatory sandboxes
• Public-private partnerships
• Access requirements
• Practical applications during development and validation
• Limitations of publicly funded support initiatives

Real-world performance testing for AI-enabled medical devices

AI is challenging the traditional regulatory model for medical devices. Historically, manufacturers have been expected to demonstrate safety and performance before market access. But increasingly versatile GenAI, adaptive AI, and agentic systems raise a harder question: what if an AI system can perform tasks that were not fully specified when it was developed?

This session explores how regulators are responding by developing adaptive staged gateways that combine controlled real-world deployment with evidence generation, continuous monitoring, and life cycle oversight.

We will compare developments in the UK, EU, US, China, and Singapore, including regulatory sandboxes and supervised real-world testing.

Attendees will gain practical insight into questions such as:

• When and where do these pathways become available?
• What evidence is still needed before deployment?
• How should human oversight, monitoring, updates, and rollback work?
• How might responsibilities shift between manufacturers, healthcare providers, and regulators?

The session concludes with what this means for manufacturers: where these pathways may create opportunities for earlier evidence generation and which capabilities product teams may need to build.

Topics include:

• What are staged adaptive real-world testing pathways?
• What needs to they address?
• How do they work?
• When and where are they operational?
• What regulators plan routine monitoring of real-world performance?
• How might real-world monitoring impact you, and how can you prepare?

Natural Language Processing Systems

Natural language processing (NLP) is transforming the way healthcare organizations capture, interpret and generate clinical information. From AI scribes and clinical documentation to text classification, retrieval-augmented generation (RAG) and large language models (LLMs), these technologies are creating new opportunities while introducing new regulatory and validation challenges.

This session examines the application of NLP and generative AI in healthcare, with a particular focus on evaluating the performance, reliability and clinical suitability of LLMs. Participants will explore common challenges, including hallucinations, partial correctness and model variability, together with approaches for evaluating AI systems using automated methods, expert review, synthetic data and real-world data.

Topics include:

• Regulatory landscape
  •    Regulatory considerations for generative AI 
  •    Emerging regulatory expectations for NLP systems 
  •    Regulatory implications for healthcare applications
• Clinical validation
  •    Clinical validation of generative AI for healthcare 
  •    Defining appropriate validation objectives 
  •    Performance evaluation in clinical settings
• Evaluation methodologies
  •    Assessing large language models for healthcare 
  •    Evaluation metrics and benchmarking approaches 
  •    Measuring accuracy, robustness and consistency 
  •    Hallucinations and partial correctness 
  •    Model variability and reproducibility
• Validation strategies
  •    Human-in-the-loop evaluation 
  •    Automated evaluation methods 
  •    Real-world datasets 
  •    Synthetic datasets 
  •    Strengths and limitations of different evaluation approaches
• Case study
  •    Clinical validation using synthetic data 
  •    LLM-as-a-judge methodologies 
  •    Practical considerations for implementing evaluation frameworks