Overview
IEC PAS 63621:2026 focuses on establishing a high-level, robust framework for the management of data used in artificial intelligence (AI) enabled medical devices. Published by the International Electrotechnical Commission (IEC), this standard guides organizations in effectively overseeing the data lifecycle-covering stages from data planning and acquisition to usage and decommissioning-when developing, validating, or maintaining AI models within medical devices. Emphasis is placed on ensuring data quality, traceability, security, and compliance throughout these processes, helping manufacturers and suppliers demonstrate their competence in managing critical healthcare data in accordance with relevant international guidelines.
Key Topics
- Data Lifecycle Management: Outlines processes for each phase in the data lifecycle, including planning, acquisition, development, provisioning, and decommissioning.
- Data Suitability and Quality: Defines requirements for data suitability, accuracy, completeness, representativeness, and dataset classification-ensuring data meets its intended medical purpose.
- Integrity, Traceability, and Security: Stresses maintaining data integrity and traceability, with detailed controls for data privacy, security, and robust governance.
- Data Annotation and Metadata: Highlights the need for comprehensive metadata and accurate data annotations to facilitate effective AI model training, testing, and validation.
- Bias Mitigation and Versioning: Encourages practices to minimize data bias, track dataset versions, and ensure equitable model performance.
- Risk Management: Incorporates risk identification and mitigation when data could contribute to hazardous situations or impact patient safety.
- Quality Management System (QMS) Integration: Specifies that data management processes should align with organizational quality management systems, supporting regulatory compliance and continuous improvement.
Applications
IEC PAS 63621:2026 is applicable across the entire ecosystem of AI-enabled medical devices, including:
- Medical Device Manufacturers: Supporting organizations involved in design, development, production, installation, or servicing of AI-based medical devices to ensure systematic, quality-driven data management.
- Data Suppliers and External Partners: Offering guidance for third parties providing data or QMS-related services to device manufacturers.
- Healthcare IT and Systems Engineers: Enabling effective data stewardship practices in healthcare environments where AI-driven diagnostic or therapeutic devices are deployed.
- Quality and Regulatory Compliance Teams: Assisting in building processes that align with medical device regulations and international standards for AI, ensuring patient safety and trustworthiness of AI models.
- AI Model Developers: Informing developers of good data management practices, covering data sampling, cleaning, augmentation, and validation techniques relevant for safe and effective medical AI solutions.
Related Standards
Organizations implementing IEC PAS 63621:2026 may reference the following standards for enhanced data management and regulatory alignment:
- ISO 13485: Medical devices - Quality management systems - Requirements for regulatory purposes.
- ISO/IEC 22989: Artificial intelligence concepts and terminology.
- ISO/IEC 25012: Data quality model requirements.
- ISO 8000-2: Data quality management processes and definitions.
- ISO/IEC 2382: Information technology - Vocabulary.
- ISO/IEC Guide 63: Guidance on managing risks from medical devices.
Practical Value
Adopting IEC PAS 63621:2026 enables organizations to:
- Demonstrate control and oversight of data critical to AI-enabled medical device performance.
- Facilitate compliance with global regulatory requirements through an auditable, standardized approach to the data lifecycle.
- Improve the safety, reliability, and efficacy of AI medical devices by managing risks associated with data quality, integrity, and bias.
- Support market access, customer confidence, and clinical adoption for innovative AI healthcare solutions.
By implementing the principles and processes outlined in this standard, manufacturers and stakeholders can confidently build, manage, and maintain data assets pivotal to the future of AI in healthcare.