The Medicines and Healthcare Products Regulatory Agency (MHRA), a UK authority in the sphere of healthcare products, has published a roadmap describing the changes to be implemented to the existing framework for software- and artificial intelligence(AI)- based products subject to regulation as medical devices. By virtue of the document, the authority highlights the key issues identified and also describes the amendments to be implemented in order to address these issues properly and ensure the safety and effectiveness of the products placed on the market. At the same time, the document itself is not intended to introduce new rules or impose new obligations but rather to provide additional information and clarifications regarding the upcoming changes to the respective legal framework.
The scope of the document covers, inter alia, the aspects related to the regulatory status of the products utilizing artificial intelligence technology. In this respect, the authority acknowledges the lack of clarity with respect to the specific requirements these products are subject to in terms of safety and effectiveness when used for their intended purpose.
First of all, the authority outlines the main objectives to be achieved, namely:
- Utilize existing regulatory framework to ensure AIaMD placed on the market is supported by robust assurance that it is safe and effective.
- Develop supplementary guidance to better ensure AIaMD placed on the market is supported by robust assurance with respect to safety and effectiveness.
- Outline technical methods to test AIaMD to ensure the device is safe and effective.
When describing the way the above goals are expected to be achieved, the authority pays special attention to guiding principles to be followed by all the parties involved in operations with the respective devices.
In this respect, the MHRA mentions the principles of good machine learning practice (GMLP) initially issued in collaboration with other authorities (namely, US FDA and Health Canada) earlier in October 2021. It is explained that these principles are basic, so their application could be complicated. The initial purpose of the said basic principles is to address properly the unique nature of the products with a machine learning function that differs significantly from general software-based products. These principles are also expected to encourage further development of such innovative products while ensuring their safety and proper performance. As it is stated by the authority, the 10 guiding principles identify areas where the International Medical Device Regulators Forum (IMDRF), international standard organizations, and other collaborative bodies could work to advance GMPL; areas of collaboration include research, creating educational tools and resources, international harmonization, and consensus standards, which may help inform regulatory policies and regulatory guidelines.
The purpose of the present document issued by the MHRA is also to describe the way the said principles could be applied in the context of the respective regulatory requirements AIaMD products are subject to, including the ones established under the Medical Devices Regulations 2002 (with changes and amendments).
Apart from the basic principles, the authority pays attention to the applicable standards medical device manufacturers (software developers) may refer to. In this respect, the authority emphasizes the important role of standards in the harmonization of regulatory requirements at the international level. The MHRA further intends to collaborate with various national and international bodies involved in the development of such standards.
The MHRA will also contribute to the further development of the applicable standards addressing various aspects related to the safety and effectiveness of AIaMD.
Best Practice Guidance
The authority also intends to publish guidance documents describing in detail the basic practice for the development and deployment of AIaMD, expanding the recommendations provided in the guidance dedicated to best practices for SaMD in order to reflect the specific nature of AIaMD products. Apart from that, the document will also describe the approach to be applied when assessing the actual performance of AI-based products.
According to the document, the authority identifies the three major challenges to be addressed in the new guidance, namely:
- Performance of AIaMD across populations and different real-world conditions;
- Ensuring data are properly contextualized to avoid AIaMD perpetuating inequalities or leading to poorer performance in subpopulations;
- Working to ensure that AIaMD meets the needs of the communities in which it is deployed in terms of verification and validation.
The new guidance will also provide additional recommendations to be taken into consideration by the parties involved in order to mitigate potential bias. These recommendations will be based on the information available to the authority with respect to different methods to be used to detect and measure bias appearing when it takes to the use of datasets. According to the guidance, an approach will be developed to identify under-represented features in data and then use synthetic data to oversample the under-represented features and achieve a better overall distribution of features. However, the authority also mentions that the aforementioned approach will be subject to rigorous testing in order to ensure the goal is achieved.
In summary, the present MHRA guidance outlines the key issues associated with the existing regulatory framework for AI-based products and also provides additional clarifications regarding its further development. The document pays special attention to the applicable standards and basic principles to be followed by the parties involved in order to ensure the safety and proper performance of medical devices.
How Can RegDesk Help?
RegDesk is a holistic Regulatory Information Management System that provides medical device and pharma companies with regulatory intelligence for over 120 markets worldwide. It can help you prepare and publish global applications, manage standards, run change assessments, and obtain real-time alerts on regulatory changes through a centralized platform. Our clients also have access to our network of over 4000 compliance experts worldwide to obtain verification on critical questions. Global expansion has never been this simple.