The European Medicines Agency has recently published a reflection paper on the integration of artificial intelligence (AI) throughout the lifecycle of medicinal products.
The document outlines the potential for AI to transform the development, authorisation and post- authorisation stages, while also addressing the challenges and ethical considerations involved.
The EMA emphasises the importance of collaboration among regulators, industry, and academia as a key factor in ensuring the successful implementation of AI. It is essential that AI systems are designed to continuously learn and adapt to new data and evolving regulatory requirements, ensuring their continued effectiveness and relevance in a rapidly changing environment.
While the EMA guidelines are not legally binding, companies seeking to enter the European market and obtain marketing authorisation for their products are required to adhere to these guidelines. The following section will provide a more detailed examination of the topic.
Drug Discovery
The application of artificial intelligence and machine learning (AI/ML) in drug discovery can have a minimal regulatory impact if suboptimal performance only affects the developer. However, when AI/ML outputs are included in regulatory submissions, they must comply with the established non-clinical development guidelines. This guarantees that the evidence supplied is robust, reliable and compliant with regulatory standards.
Non-Clinical Development
Any AI/ML applications in non-clinical development that aim to improve data analysis and interpretation, potentially replacing, reducing, and refining the use of animals, must comply with existing Standard Operating Procedures (SOPs) and the OECD Series on Principles of Good Laboratory Practice (GLP). This entails adherence to advisory documents on the application of GLP principles to computerised systems and data integrity.
Clinical Trials
Any AI/ML applications used in clinical trials must comply with the relevant Good Clinical Practice (GCP) guidelines. This is to ensure the integrity of the data collected and the safety of the participants involved. This entails furnishing exhaustive documentation pertaining to the model architecture, development logs, validation and testing data, and data processing pipelines. Furthermore, AI/ML systems utilised for clinical management may be classified as medical devices in accordance with the MDR or IVDR. It is imperative that all relevant regulations are adhered to, including the acquisition of the requisite qualifications and classifications. It is possible that even CE-marked devices may require additional qualifications for use in clinical trials. This is to ensure the rights, safety, and well-being of subjects, as well as the integrity of trial data.
Precision Medicine
It is essential that AI/ML applications in precision medicine comply with the relevant regulatory requirements for individualised treatment. AI/ML models must be validated and their outputs should be reliable and actionable.
Compliance
Failure to comply with these provisions may result in regulatory rejection or delays in trial approval, which could have a significant impact on both the pharmaceutical and medical technology sectors.
For more information read the full document
