Artificial Intelligence in GxP Regulated Environments: How to Harness its Power While Mitigating Risks
Introduction
Recent years have seen the Life Science industry undergo significant advancements through the integration of cutting-edge technologies such as artificial intelligence (AI) and machine learning (ML) . These technologies have greatly enhanced process efficiency and enable the analysis of large-scale datasets. Nevertheless, these advancements have presented new challenges, particularly in ensuring regulatory compliance and suitability of IT systems, applications, and solutions in validated environments.
Whether you are a researcher, a software developer, or a healthcare professional, this article provides valuable insights into the latest developments in life sciences technology and their impact on healthcare delivery.
Use of AI in Life Science industry
Drug discovery, Design, and Development:
According to recent reports, it now costs an average of US$2.6 billion[1] to bring a new drug to market. Unfortunately, even promising drug candidates that show potential in laboratory testing can still fail during clinical trials, with less than 10% [2] making it to market after Phase I trials. As a result, experts in the field are turning to the incredible data processing capabilities of AI systems to accelerate the drug discovery process while keeping costs in check. This, in turn, leads to efficiently identifying new drug candidates and optimizing their properties. In fact, some pharmaceutical companies have utilized AI systems at the core of their research and drug discovery efforts, emphasizing the importance of these advanced tools in the industry.[3]
Pre-clinical and Clinical trials:
AI-based systems are making the drug development process more efficient and cost-effective, particularly when it comes to patient enrollment and site selection for clinical trials. By analyzing large amounts of data, AI can identify suitable patients for clinical trials and match them with appropriate trial sites, streamlining the process and reducing costs.[4]
In clinical trials, AI-based systems can play a crucial role in patient selection and monitoring. AI algorithms have the potential to analyze patient data, such as electronic health records, to identify individuals who are most likely to benefit from a specific treatment, which can speed up the clinical trial process. Additionally, AI can monitor patient safety and drug efficacy during clinical trials, providing real-time data and insights to help improve decision-making.[5]
AI in pharmaceutical manufacturing:
The implementation of AI in the pharmaceutical industry offers opportunities to optimize manufacturing processes such as process design, control, smart monitoring, and maintenance.[6] Predicative maintenance is for example an emerging use of AI, to warn when equipment such as reactors, centrifuges, and packaging machines is likely to fail and to pre-emptively schedule maintenance before the failure occurs. For instance, AI based software can analyze data from bioreactor processes, providing real-time insights and predictions that aid in decision-making and production optimization. [7]
Compliance Strategies for AI Systems in GxP-Regulated Areas
Several regulatory bodies have issued guidance on the use of AI in GxP regulated environments, including the FDA (U.S. Food Drug Administration), EMA (European Medicines Agency), and EC (European Commission). These guidelines provide a framework for the development and use of AI systems in these industries, and they emphasize the need for risk management, transparency, and accountability.
FDA: Artificial Intelligence/Machine Learning (AI/ML)-Based Software as a Medical Device (SaMD) Action Plan
The FDA’s “Artificial Intelligence/Machine Learning (AI/ML)-Based Software as a Medical Device (SaMD) Action Plan” is one such guideline. The plan emphasizes the need for manufacturers to ensure continued safety and effectiveness of these devices by continuously monitoring and assessing their performance. “The highly iterative, autonomous, and adaptive nature of these tools requires a new, total product lifecycle (TPLC) regulatory approach that facilitates a rapid cycle of product improvement and allows these devices to continually improve while providing effective safeguards.” [8]
In addition to recommending a TPLC approach to validation, the FDA suggests that manufacturers of AI/ML-based SaMD include additional modules that can help explain the results of the software to human users known as explainable AI. [9]
The guidance also encourages stakeholders to contribute to the development of Good Machine Learning Practice (GMLP) and work towards establishing standards that are based on consensus in this field.
FDA Discussion Paper: Artificial Intelligence in Drug Manufacturing
FDA provides an overview of expected advancements in using AI in pharmaceutical manufacturing (Advanced Manufacturing), like advanced and real-time process control and process monitoring. This discussion paper invites the industry to provide input on important questions such as the development ofcommon practices for validation and maintenance of self-learning AI models. This promises to be a very interesting conversation between the agency and the industry to be followed up on.[10]
Good Machine Learning Practice (GMLP)
GMLP is a concept that is still in development and has not yet been fully established as a formal set of standards or guidelines. It is being developed by several organizations and stakeholders in the medical device industry, including the FDA, the International Medical Device Regulators Forum (IMDRF), and industry groups such as AdvaMed and MITA.
EMA: Concept Paper on the revision of Annex 11
While the current version of Annex 11 does not specifically mention AI, the concept paper for the revision of Annex 11 recognizes the increasing relevance and prevalence of AI in GMP processes and calls for updated guidance that addresses the validation, qualification, and documentation requirements for AI-based computerized systems.[11]
GAMP Guide: Risk based, and data-driven approach
Annex of GAMP guide second edition provides guidance on how to adapt the GAMP 5 framework to the specific challenges of validating ML/AI systems.
By following this guidance, organizations can help ensure that their ML/AI systems are developed and validated in a rigorous and systematic way, which helps to ensure their safety, reliability, and effectiveness. The most important points to consider when validating AI-based systems according to GAMP 5 are risk management and data integrity. The quality and integrity of the data used to train and test the AI system are critical to ensuring its accuracy and reliability not only the amount. Additionally, the selection of algorithms and models should be carefully considered based on the intended use and level of complexity required, and the performance of the system should be evaluated to ensure that it meets user requirements. [12]
The European Commission’s “Ethics Guidelines for Trustworthy AI”
Although not directly linked to the Life Science industry, these guidelines provide a framework for developing AI systems that are ethical, transparent, and accountable. The guidelines cover a range of topics, including human oversight, privacy, data protection, and the responsible use of AI in decision-making. By aligning with risk management strategies often employed in Life Science, these guidelines can foster greater trust and acceptance of AI technologies, benefiting both industry and society.[13]
Conclusion
AI technology holds great promise for advancing the fields of drug development and medical device innovation. However, it’s crucial to implement effective management strategies to ensure safety, efficacy, and compliance with regulatory requirements. To achieve this, organizations working with AI must stay informed of the latest developments and consult with regulatory authorities as needed to stay on track.
At GxP-CC, we offer expert services to support you in leveraging the power of AI technology, while ensuring compliance with GxP regulations. Contact us to learn how we can assist you in navigating this complex landscape and achieving the highest standards of quality, reliability, and patient safety .
[1] Innovation in the pharmaceutical industry: new estimates of R&D costs: https://www.sciencedirect.com/science/article/abs/pii/S0167629616000291?via%3Dihub
[2] Clinical Development Success Rates 2006-2015: https://www.bio.org/sites/default/files/legacy/bioorg/docs/Clinical%20Development%20Success%20Rates%202006-2015%20-%20BIO,%20Biomedtracker,%20Amplion%202016.pdf
[3] Insilico: end-to-end, artificial intelligence-driven pharma-technology company using Generative AI to Design New Drugs for rare diseases.
Existensia: an AI-driven pharmatech company committed to progress AI-designed small molecules into the clinical setting.
BenovolentAI: Second novel target for idiopathic pulmonary fibrosis was discovered using BenevolentAI’s drug discovery Platform and selected for AstraZeneca’s drug development portfolio.
[4]Thermo Fisher Scientific Partners with Medidata to Optimize Clinical Research Site Selection and Speed Patient Enrollment in Clinical Trials: https://corporate.thermofisher.com/content/tfcorpsite/us/en/index/newsroom/press-releases/2022/Feb/07-Thermo-Fisher-Scientific-Partners-with-Medidata-to-Optimize-Clinical-Research-Site-Selection-and-Speed-Patient-Enrollment-in-Clinical-Trials.html
[5] https://www.ncbi.nlm.nih.gov/pmc/articles/PMC8011519/
[6]Artificial Intelligence in Drug Manufacturing: https://www.fda.gov/media/165743/download
[7] Aizon Launches GxP AI Bioreactor Application for the Pharma Industry: Aizon Launches GxP AI Bioreactor Application for the Pharma Industry to Scale Manufacturing & Quality – aizon. See: https://www.aizon.ai/aizon-launches-bioreactor-application
[8] The FDA guidance on the use of AI in GxP regulated environments is called “Artificial Intelligence/Machine Learning (AI/ML)-Based Software as a Medical Device (SaMD) Action Plan. See: https://www.fda.gov/media/122535/download
[9] Pharmaceutical engineering: The official Magazine of ISPE January-February 2023 | Volume 43, Number 1
[10] Artificial Intelligence in Drug Manufacturing: https://www.fda.gov/media/165743/download
[11] Concept Paper on the revision of Annex 11 of the guidelines on Good Manufacturing Practice for medicinal products – Computerized Systems, section 31: https://www.ema.europa.eu/en/documents/regulatory-procedural-guideline/concept-paper-revision-annex-11-guidelines-good-manufacturing-practice-medicinal-products_en.pdf%20
[12] GAMP 5: A Risk-Based Approach to Compliant GxP Computerized Systems, Second Edition
[13] Ethics Guidelines for Trustworthy AI, see: https://digital-strategy.ec.europa.eu/en/library/ethics-guidelines-trustworthy-ai
Definitions:
ML = Machine Learning
AI = Artificial Intelligence