Validating Intelligent Automation Systems in Pharmacovigilance: Insights from Good Manufacturing Practice

Intelligent automation of pharmacovigilance (PV) tasks holds significant potential to automate manual work associated with the processing and evaluation of reported adverse reactions, thereby facilitating higher quality and more efficient risk mitigation. However, while these technologies provide great hope for vastly improving pharmacovigilance processes for both sponsors and patients, existing ways of validating technology will need to be adjusted to cover the more novel AI-based systems.

Rules-based software has been widely deployed in pharmacovigilance for years based on existing guidance from regulators and other standards setting bodies, such as GAMP. However, the emergence of novel technologies such as artificial intelligence (AI), machine learning (ML), and natural language processing (NLP) have surpassed the boundaries of traditional validation frameworks. As adjustments are made to these AI-based systems, pharmacovigilance professionals will play an increasingly active role in bridging the gap between business operations and technical advancements to ensure inspection readiness and compliance with global regulatory authorities.

Our recent paper, “Validating Intelligent Automation Systems in Pharmacovigilance: Insights from Good Manufacturing Practices,” focuses on validation considerations for these novel emerging technologies, with a framework based on areas where AI-based systems can introduce new risks.

The paper proposed three categories of intelligent automation systems that will require distinct validation approaches:

1. Rules-Based Static System: Automation is achieved via static rules designed to obtain the desired outcome
2. AI-Based Static Systems: System configuration includes components that are AI informed but subsequently “frozen,” i.e., systems based on AI or ML that do not adapt in production (after “go-live”). These are also called “Locked” models
3. AI-Based Dynamic Systems: System configuration includes components that are AI informed and can continually adjust their behavior based on incoming data after initial implementation in production, using a defined learning process

Our paper provides a framework of considerations for implementing novel technologies and is intended to facilitate dialogue and collaboration with Health Authorities, industry, and other stakeholders on this important topic. For more information on our Intelligent Automation Opportunities in Pharmacovigilance Initiative or to use our interactive ICSR & Automation Technologies Tool, check out our solutions page.

Kristof has been with UCB since 2009 and has been heading safety systems since 2014. He was previously head of safety surveillance and head of safety data management. Kristof currently serves a dual role within patient safety and clinical operations. Since 2018, Kristof has led a diverse TransCelerate sub-team of PV business owners and other SMEs in areas such as computer system validation and machine learning. They have recently published “Validating Intelligent Automation Systems in Pharmacovigilance: Insights from Good Manufacturing Practices” in the journal Drug Safety.