AI and drugs regulation, the UK example

Un report dell’MHRA esplora l’utilizzo dell’intelligenza artificiale per la regolamentazione dei farmaci nel Regno Unito (e non solo)

0
18

In April this year, the UK Medicines and Healthcare products Regulatory Agency (Mhra) published a paper exploring the influence of artificial intelligence (AI) on the regulation of medical products. The paper, entitled “Impact of AI on the Regulation of Medical Products”, is structured in three parts, each of which reflects one of the different roles that the agency can take on in the context of AI.

1. AI Product Regulator

Currently, these devices are regulated under the Medical Devices Regulations 2002, which covers the entire product life cycle, from pre-market clinical trials to post-market surveillance. With the evolution of AI,the Mhra intends to update the regulations to guarantee a regulation capable of balancing the needs in terms of safety and effectiveness and risk management without stifling innovation. The agency has published several guidelines for AI device manufacturers and expects additional documents by 2025.

2. Public service organisation

In this role, the MHRA is leveraging artificial intelligence to improve the efficiency of its regulatory processes. In particular, it wants to use advanced technologies such as supervised machine learning to optimize the initial evaluation of authorization applications for new medicines, in order to significantly reduce the time needed to approve products while maintaining safety requirements and effectiveness.

This innovation allows human reviewers to focus on critical activities that require advanced skills, such as innovation enablement (the set of strategies implemented to facilitate and promote the development and implementation of new technologies in the healthcare sector) and engagement of patients. Furthermore, the English Agency is developinga comprehensive data strategy, which includes the adoption of large language models (LLM) and generative AIto support various business functions, from c >communication to customer service, up to post-market surveillance.

3. Organisations making evidence-based decisions that impact public and patient safety

This section emphasises the importance of effective regulation in the age of artificial intelligence. This is crucial as AI begins to play an increasingly significant role in the generation and management of the evidence on which regulatory decisions are based. For example, AI is being used to collect and analyse data from different sources, improving the agency’s ability to evaluate medical products.

This includes the use of real data, advanced analytics and machine learning models to better understand the efficacy and safety of products. By integrating AI into decision-making processes, Mhra aims to ensure that decisions are based on robust and verifiable evidence.

The agency is also working with international regulatory bodies to develop best practices and move towards global alignment of procedures. It is also exploring the use of AI and advanced analytical models to improve vigilance and post-market safety, ensuring that products remain safe and effective throughout their life cycle.

The idea is to apply Bayesian models in vigilance activities: these models allow for the continuous integration of new information from real-world data (such as data from electronic health records, disease registries and other health data sources), improving the system’s ability to identify and assess safety signals of medical products.

 

Predefined Change Control Plans (Pccp)

To manage changes to the artificial intelligence systems that will be used in the healthcare environment in a standard and controlled way, the Agency has introduced PCCPs (Predetermined change control plans), plans that define how the AI-based medical device manufacturers can manage and document changes made to their modelsafter they are released to the market.
These plans, developed in collaboration with international partners such as the US Food and Drug Administration (FDA) and Health Canada, allow manufacturers to implement changes to their models in a controlled and compliant manner regulators, ensuring that such changes do not compromise the safety or effectiveness of the device.
PCCPs guarantee complete traceability of changes made to artificial intelligence (AI) models in medical devices, from the development phase through to distribution and use. This control system ensures that each change is documented in detail, allowing regulatory bodies to verify compliance with current regulations and ensuring that changes do not introduce new risks for patients.

An integral part of PCCPs is continuous risk assessment, which helps maintain the safety and effectiveness of the device. Additionally, these procedures facilitate the adoption of a standardized approach to change management by defining clear criteria for evaluating and implementing changes. In the future, the MHRAintends to introduce PCCPs into mainstream regulations, initially on a non-mandatory basis,to progressively improve lifecycle management of medical devices with AI.