MHRA announces ten principles for medical devices using AI and machine learning technologies

On October 27, 2021, The U.S. Food and Drug Administration (FDA), Health Canada, and the United Kingdom’s Medicines and Healthcare products Regulatory Agency (MHRA) joined forces in publishing an informative guidance document that outlines “10 guiding principles that inform the development of Good Machine Learning Practice (GMLP)” (MHRA, 2021).[/vc_column_text]

The document aims to “promote safe, effective, and high-quality medical devices that use artificial intelligence and machine learning (AI/ML)” (MHRA, 2021). The development of these self-learning technologies is crucial to improving health care. However, their complexity and swift expansion also pose some distinct issues which would require the advance of GMLP in the regulatory field. Hence, close cooperation between regulators fora and associations, international standards organizations, and others is highly desirable to address them.

The authors “envision that these guiding principles may be used to:

  • Adopt good practices that have been proven in other sectors;
  • Tailor practices from other sectors so they are applicable to medical technology and the health care sector;
  • Create new practices specific for medical technology and the health care sector” (MHRA, 2021).

The the guidance wishes to introduce are the following:

  1. Multi-Disciplinary Expertise Is Leveraged Throughout the Total Product Life Cycle: the aim is to fully understand the risks and benefits, as well as the model’s intended integration into the clinical workflow.
  2. Good Software Engineering and Security Practices Are Implemented: robust practices include risk management, design process, risk management decision and rationale, data authenticity, and integrity.
  3. Clinical Study Participants and Data Sets Are Representative of the Intended Patient Population: ensuring that the collected data can be applied to the population of interest is essential to avoid bias, assess usability, and prevent underperformance. 
  4. Training Data Sets Are Independent of Test Sets: any possible source of dependence needs addressing to assure independence.
  5. Selected Reference Datasets Are Based Upon Best Available Methods to ensure model robustness and generalizability for the intended patient population.
  6. Model Design Is Tailored to the Available Data and Reflects the Intended Use of the Device: clinical benefits and risks are understood and channeled to achieve the device’s intended purpose by deriving clinically significant performance objectives for testing.
  7. Focus Is Placed on the Performance of the Human-AI Team rather than on the model performance on its own.
  8. Testing Demonstrates Device Performance During Clinically Relevant Conditions to generate clinically relevant performance information independently of the training dataset.
  9. Users Are Provided Clear, Essential Information: among others – the device intended use, model performance, indications for use, acceptable inputs, known limitations, user interface interpretation, and means to voice one’s concerns to the developer.
  10. Deployed Models Are Monitored for Performance and Re-training Risks Are Managed to preserve or enhance the model’s safety.

Was this article helpful?

Check Obelis UK News page to keep up to date with the latest Medical Devices and Cosmetics market in the UK.

Tancredi Vergani

RA Department

22 December 2021

Should you like to appoint Obelis UK as your UK RP, or if you simply need guidance on your UK documentation, please do not hesitate to contact us.

Get in Touch

Resources:

MHRA. (2021). Guidance. Good Machine Learning Practice for Medical Device Development: Guiding Principles. Retrieved on December 21, 2021 from https://www.gov.uk/government/publications/good-machine-learning-practice-for-medical-device-development-guiding-principles/good-machine-learning-practice-for-medical-device-development-guiding-principles.

[/vc_column][/vc_row]

Share This

Copy Link to Clipboard

Copy