Advarra has launched the Council for Responsible AI in Clinical Trials to ensure the ethical use of AI in clinical trials. Laura Russell, Senior Vice President and Head of Data and AI Product Development at Advarra, shares her insights on the council’s inception, objectives, and challenges. As a leader in the field, Laura discusses the importance of creating a framework that ensures the ethical and effective use of AI in clinical research, aiming to enhance efficiency and patient outcomes.
Moe: When did the idea for this AI council first click at Advarra?
The idea emerged during a moment of realization for our data science and informatics team, who were grappling with questions about regulatory parameters and accuracy thresholds. We recognized a fragmentation in governance and a need for a structured approach to responsible AI. This led us to create a neutral forum to bring together operators, technology developers, and decision-makers to address practical aspects like traceability and replication. It was a serendipitous moment that highlighted the need for collaboration across various stakeholders within our organization. We had data scientists, technology leads, and informatics teams all contributing to this realization. The absence of complex governance made it clear that we needed to take another step in bringing stakeholders together to address these ambiguities.
Moe: Was there any regulatory inspiration that helped with this initiative?
We did consider the draft guidance from the White House and the FDA, which emphasized patient safety and outcomes. However, our focus is more technical, aiming to validate and verify the intelligence within AI models. The initial guidance focused more on immediate patient interactions, but we wanted to delve deeper into how AI models are validated and how their outputs are reproduced. This involves ensuring that the deployment of AI in clinical trials is scientifically sound, ethically aligned, and operationally relevant. Advarra’s deep relationships with sites and sponsors allows us to facilitate these crucial conversations. We aim to create a framework that not only aligns with regulatory expectations but also addresses the technical nuances of AI deployment in clinical trials.
Moe: How has your thinking around AI in clinical research evolved since the council began?
The Council’s discussions have always focused on practical applications of AI that drive value and have an immediate impact. The industry is now more grounded in what is achievable and impactful, taking a thoughtful approach to integrating AI in a stepwise fashion. This reflects a broader industry trend towards practical, measurable improvements rather than chasing an elusive, perfect solution. Our stakeholders are focused on what is useful and can drive value, rather than expecting AI to solve all challenges instantly.
Moe: How have different stakeholders interpreted the concept of responsible AI, and how are these perspectives influencing the council’s work?
Different stakeholders bring varied concerns. Sponsors focus on regulatory risk and traceability, sites on transparency and workflow impact, and technology developers on defining boundaries. For instance, sponsors are concerned about how AI might affect regulatory compliance, while sites are interested in how AI could streamline recruitment and protocol design. The council aims to integrate these perspectives, not just for consensus but to identify pressure points and define responsible behavior in real-world applications. This diversity of thought is crucial in shaping a comprehensive framework that addresses the multifaceted nature of AI in clinical trials. By bringing together these varied perspectives, we can ensure that our approach to responsible AI is well-rounded and addresses the needs of all stakeholders involved.
Moe: What key issue keeps resurfacing that the council is still figuring out?
Traceability and reproducibility are ongoing challenges. We need to strike a balance between transparency and intellectual property concerns. For example, when deploying an e-clinical solution, sponsors are accustomed to detailed access for testing and validation. However, with AI models, the complexity increases, and we must find a way to demonstrate that data is sound and models perform as expected without compromising proprietary information. This tension between transparency and IP protection is a significant hurdle, but as AI becomes more integrated into everyday life, these conversations are becoming easier to navigate. The challenge lies in ensuring that we can provide enough transparency to build trust while protecting the proprietary aspects of AI models.
Moe: If the council is successful, how will it influence the FDA and study sponsors, and what will patients notice first?
Success would mean creating a proactive framework that alleviates regulatory and operational burdens. Sponsors would confidently integrate AI into trial design, knowing the rules of the road are clear. For patients, the shift might be subtle but meaningful, with more inclusive criteria and fewer delays due to protocol issues. Ultimately, it’s about reducing friction and enhancing efficiency and accuracy in bringing novel therapeutics to patients. This collective alignment among stakeholders will lead to a more streamlined process, benefiting everyone involved in the clinical trial ecosystem. Patients may not notice the changes directly, but they will benefit from more efficient and effective clinical trials that bring new treatments to market faster.
Moe: Were there any early hurdles that nearly derailed the council, or is it still too early to talk about challenges?
My primary concern was avoiding the trap of ideation without action within the consortium. We aimed to ensure progress by keeping the group focused and grounded in practical, immediate goals. The emphasis was on neutrality and bringing real operators to the table to ensure meaningful outcomes. We were deliberate about our structure, keeping the group small and focused on actionable steps. This approach helps us avoid becoming an echo chamber and ensures that we are making tangible progress toward our objectives. By maintaining a clear focus and structure, we can ensure that the council remains effective and continues to drive meaningful change in the industry.
Moe: Is there anything you’d like to add?
I’m excited about the industry’s willingness to adopt and explore AI. The complexity of clinical trials demands collaboration, and aligning stakeholders is crucial. Our focus is on governance to ensure trust and accountability while capturing the impact of innovation without adding complexity. There’s immense potential, and we’re committed to a practical and measurable approach to realizing it. This is an exciting time to be in the industry, and I’m optimistic about the positive changes we can achieve through the responsible integration of AI. By working together, we can harness the power of AI to improve clinical trials and ultimately benefit patients worldwide.
Moe Alsumidaie, MBA, MSF, is founder and Chief Editor of Vanguard Publications, which publishes Clinical Trial Vanguard, Pharma Vanguard and BullScope, and Head of Research at CliniBiz. He has two decades in clinical trial operations and data science, with earlier roles at Genentech, Abbott Vascular and Stanford University Medical Center, and is a guest lecturer in clinical trial sciences at Rutgers University.
