In this interview, Michael Torok, Head of Quality Assurance Programs at Roche, delves into the operationalization of RBQM (Risk Based Quality Management) and how AI in clinical trials can enhance RBQM practices. This discussion uncovers the nuanced complexities, challenges, and significant advancements in integrating AI into quality assurance to protect trial participant safety (“safety”) and trial data reliability (“reliability”).
Moe Alsumidaie: Could you elaborate on the current state of Risk-Based Quality Management (RBQM) in clinical trials?
Michael Torok: The clinical trial ecosystem in which RBQM practices are executed is increasingly complex and multifaceted. At its core, RBQM protects the most critical requirements during trials: safety and reliability. When thinking about how to apply AI to RBQM, I find it easiest to consider the four-pillar RBQM approach we use at Roche. Many companies use similar approaches; each pillar comprises fundamental RBQM requirements and opportunities to execute those requirements by leveraging AI in clinical trials. The four pillars of RBQM are: 1) Critical to Quality factors (CtQs), 2) Quality by Design (QbD), 3) Quality Risk Management (QRM), and 4) Quality Briefs (QBs).
There is a large amount of information available on the first three pillars, provided through ICH E6(R2), draft ICH E6(R3), and ICH E8(R1) guidance, and risk-based principles from health authorities and groups like the Clinical Trials Transformation Initiative. The fourth pillar, Quality Briefs, is a work product under development with the IMPALA (Inter-company Quality Analytics) Consortium. At its core, Quality Briefs essentially tell the end-to-end story of RBQM for a study or program with a focus on transparency aimed to increase health authority confidence in study execution.
Moe Alsumidaie: How is AI in clinical trials strategically integrated into RBQM practices?
Michael Torok: Incorporating AI into Risk-Based Quality Management (RBQM) is a strategic move. AI refines study protocols, pinpointing and mitigating high-risk areas, notably through applying large language models (LLMs) in clinical trials. These models scrutinize extensive data, bolstering RBQM and trial efficiency.
For example, AI can enhance RBQM pillars by streamlining study protocol design. Simplifying a study reduces risks, and AI, particularly LLMs, can evaluate draft protocols for feasibility and risk. This evaluation identifies likely risks,
prompting revisions to minimize them. Iteratively refining protocols with AI leads to efficient, lower-risk versions. AI’s data analysis helps spot risk patterns, aiding in the design of safer, more effective trials. AI’s role extends to drafting protocols with fewer risks, suggesting controls within trial plans, and identifying effective risk mitigation strategies. LLMs support decision-making around critical quality points (CtQs) and their risks, improving trial protocols and plans.
The ultimate aim of RBQM’s Quality Risk Management (QRM) is early risk identification and resolution, which is crucial for participant safety and data reliability. Advances in risk-based monitoring, quality analytics, and AI-enabled content generation for CAPAs and training materials mark significant progress. AI facilitates proactive monitoring and issue resolution in trials, ensuring high-quality assurance standards.
Moe Alsumidaie: How does AI in clinical trials facilitate a culture shift?
Michael Torok: AI is fundamentally reshaping the culture of clinical trials by introducing a new paradigm of risk management and quality assurance. It’s not just about implementing new technologies; it’s about fostering a mindset shift towards proactive, data-driven decision-making. AI in clinical trials enables us to anticipate potential issues, identify risks early, and take swift, informed actions. This shift towards a more proactive, anticipatory approach requires a cultural change, moving away from traditional reactive models.
In addition, it’s very exciting how this RBQM-focused cultural shift is pushing the boundaries beyond the confines of individual organizational structures, reaching out to encompass vendors, CROs, and clinical sites. This paradigm shift is not just about adopting new technologies; it’s about cultivating a mindset that encourages collaboration and critical thinking and transcends traditional operational silos. We advocate for a holistic approach where everyone involved in a clinical trial operates with a unified vision, focusing on quality and efficiency.
AI facilitates greater collaboration and transparency across various clinical trial functions. By providing a common platform of data and insights, AI in clinical trials breaks down the traditional operational silos, fostering a more integrated and cohesive approach. This ultimately leads to a culture that is more agile, responsive, and focused on continuous improvement.
Moe Alsumidaie: What are the primary challenges in implementing RBQM, particularly regarding patient safety and trial integrity?
Michael Torok: Implementing RBQM effectively comes with its challenges, categorized primarily into three areas: cultural, technological, and knowledge management. Culturally, the biggest hurdle is breaking down the silos within organizations and creating new ways of working. This requires a shift in mindset, comprehensive training, and leadership that champions a culture of open communication and collaboration.
From a technological standpoint, challenges include data harmonization and literacy. Teams must understand and utilize complex data sets and related insights. In terms of knowledge management, the focus is on the consistency and accuracy of documentation, data interpretation, and learning, which is crucial for maintaining trial integrity and ensuring patient safety.
Moe Alsumidaie: Can you detail the role of the Impala Consortium in enhancing RBQM at Roche?
Michael Torok: The Impala Consortium is not-for-profit and comprised of many companies. Founded in 2022, the consortium is working to engage health authorities to define guiding principles for using advanced analytics to complement, enhance, and accelerate current QA practices. In addition to the QB work package, IMPALA’s quality analytics work packages provide an objective quality lens on RBQM practices like risk-based monitoring. Upon further development, work packages will be open source.
Roche is a member of IMPALA, and data analytics and QBs are central to our RBQM approach at Roche. For instance, we use analytics to focus on areas like informed consent and adverse event underreporting. These packages are not just tools for compliance; they represent a proactive stance in safeguarding the safety and integrity of trials.
Moe Alsumidaie: Finally, what recommendations do you have for organizations looking to implement RBQM?
Michael Torok: For organizations embedding and enhancing RBQM, I recommend starting with a clear alignment on the CtQs that need protection in clinical trials. This alignment forms the bedrock upon which RBQM strategies are built and expanded. A focused approach streamlines the RBQM process and ensures critical trial aspects are prioritized and safeguarded.
Furthermore, organizations should invest in training and developing their teams, ensuring they possess the requisite skills and knowledge to navigate the complexities of RBQM and AI. Embracing technology is key, but it must be coupled with a strong cultural foundation that values quality, collaboration, and continuous learning. Ultimately, it’s about fostering a culture that embraces innovation and is committed to advancing patient safety and trial integrity through these sophisticated methodologies.