Note: This blog was initially published as “What is BS/AAMI 34971?”, but was renamed to reflect international recognition of BS/AAMI 34971 by ISO as ISO/TS 24971-2 in June of 2026.
What is ISO/TS 24971
ISO/TS 24971-2:2026 is an internationally-recognised copy of BS/AAMI 34971:2023 a joint standard developed by the British Standards Institution (BSI) and the Association for the Advancement of Medical Instrumentation (AAMI). Its full title is “Application of ISO 14971 to Machine Learning in Artificial Intelligence”, and it offers guidance for applying the risk management principles of ISO 14971 to medical devices that incorporate Artificial Intelligence (AI) and Machine Learning (ML) technologies.
In summary, SO/TS 24971-2 complements ISO 14971 by offering a focused framework for managing the particular risks associated with AI/ML in medical devices, ensuring that best practices evolve with technological advancements.
Who should apply ISO/TS 24971-2 Standards?
The ISO/TS 24971-2 standard is essential for developers of medical devices that incorporate AI or ML components.
While it hasn’t been formally adopted by medical device regulators in the UK or EU (yet), it is a Recognized Consensus Standard in the US (under AAMI CR34971:2022). This standard offers insights and guidance on effectively managing risks related to AI and ML in medical devices. Its adoption into an ISO standard in 2026 demonstrates its importance to medical device governance.
What are the Main Objectives of ISO/TS 24971-2?
Rather than introducing a new risk management framework, ISO/TS 24971-21 extends ISO 14971 specifically for AI and ML technologies, which introduce unique challenges such as data bias, overtrust in automated systems, and adaptive algorithms that may change after deployment.
The main objectives of ISO/TS 24971-2 are:
- To provide specific guidance on applying ISO 14971 to medical devices using AI/ML technologies.
- To address unique risks and challenges associated with these medical devices.
- Help manufacturers identify and mitigate AI/ML-specific hazards during a product’s lifecycle.
- To provide a framework for integrating AI/MLrisks into the overall risk management process for medical devices.
- To support manufacturers in developing safe and effective AI/ML medical devices while meeting regulatory requirements.
What are the Key Components of a ISO/TS 24971-2-Compliant Risk Assessment?
A ISO/TS 24971-2-compliant risk assessment for AI/ML-enabled medical devices should include the following key components:
- Alignment with ISO 14971: Adapting the established risk management process to address AI/ML-specific challenges.
- AI/ML-specific Hazard Identification: Recognising unique risks associated with AI/ML such as data quality issues, algorithm biases, and performance degradation.
- Data Management: Highlighting the importance of data quality, completeness, and consistency in AI/ML systems.
- Adaptive Systems: Tackling change control challenges in evolving AI systems.
- Bias and Fairness: Guidings manufacturers in identifying and mitigating potential biases in AI/ML systems.
- Performance Monitoring: Emphasises the need for ongoing performance assessments post-deployment.
- Risk Evaluation: Offering guidance on evaluating overall residual risk, considering algorithmic outcomes and decision thresholds.
- Explainability: Documenting AI decision-making processes for stakeholders.
What are the Benefits of Implementing ISO/TS 24971-2?
Implementing ISO/TS 24971-2 offers several key benefits for medical device manufacturers developing AI/ML-enabled products:
Enhanced Risk Management
The standard offers targeted guidance for identifying and mitigating risks specific to AI/ML systems, enhancing the safety of medical devices. It builds on the ISO 14971 framework, providing tailored guidelines to address unique risks such as data handling, potential bias, privacy issues, and the evolving nature of AI/ML algorithms, which traditional frameworks may overlook.
Regulatory Alignment
Compliance with ISO/TS 24971-2 streamlines regulatory processes by aiding manufacturers in preparing comprehensive submissions for AI/ML-enabled medical devices. The standard aligns with global regulations, including FDA guidelines, which recognize ISO/TS 24971-2 (as AAMI CR34971:2022) as appropriate guidance. It also aligns with emerging international frameworks like the EU AI Act, helping manufacturers meet regulatory expectations across various regions.
Improved Product Quality
ISO/TS 24971-2 promotes stringent design controls for AI/ML systems, tackling unique challenges throughout the development lifecycle. This emphasis on continuous validation, effective monitoring, and performance management results in higher quality and more reliable devices. Manufacturers following this standard are better positioned to deliver consistent, high-performing products that meet rigorous healthcare quality standards.
What are the Common Challenges in Implementing ISO/TS 24971-2?
Complexity
The complexity of ISO/TS 24971-2 can make it difficult to implement without specialist knowledge, with some struggling to adapt it effectively without explicit guidance available.
Dynamic Security Threats
Balancing Safety and Innovation
Striking a balance that allows flexibility across different AI applications while upholding robust safety protocols remains a key challenge for any AI-focused standard like ISO/TS 24971-2 (as well as the healthcare industry in general).
Overly rigid risk management requirements could stifle innovation, delaying the deployment of promising AI technologies in healthcare, while deploying AI systems too hastily, without sufficient safeguards, poses substantial risks to patient safety.
Addressing these challenges requires collaboration between manufacturers, healthcare providers, and regulators to ensure that ISO/TS 24971-2 can be both comprehensive and adaptable in guiding the safe deployment of AI in healthcare.
How do DCB0129, ISO 14971 and ISO/TS 24971-2 connect?
DCB0129, ISO 14971, and ISO/TS 24971-2 form a comprehensive framework for managing risks in medical devices and health IT systems.
ISO 14971 establishes a standardised risk management process essential for medical devices, providing the foundational principles and structured approach to identify, assess, and mitigate risks associated with device use. ISO/TS 24971-2 complements this framework by offering specific guidance on implementing ISO 14971, making it easier for organisations to apply these risk management principles effectively in practice.
DCB0129 tailors these ISO 14971 principles specifically to health IT systems in the UK, ensuring consistent patient safety standards across medical devices and software. Together, these standards provide a harmonised approach to risk management in healthcare, linking international standards (ISO 14971), practical guidance (ISO/TS 24971-2), and region-specific requirements (DCB0129) to create a robust system for protecting patient safety across medical devices and health IT solutions.
Summary
Overall, ISO/TS 24971-2 offers essential, tailored guidance for managing the unique risks of AI and ML in medical devices, building upon the widely adopted ISO 14971 framework. It addresses the particular challenges posed by adaptive algorithms, data quality, and evolving security vulnerabilities, helping manufacturers integrate robust risk management practices for AI/ML components across the product lifecycle.
While implementation requires specialised knowledge and ongoing vigilance, adhering to ISO/TS 24971-2 enhances regulatory alignment, product quality, and safety, ensuring manufacturers can navigate the complexities of AI in healthcare responsibly.
Building an AI-powered Medical Device?
Let us help you find a route from from where you are, to where you need to go. We’re here to help.
