AI in clinical trial planning and design feasibility

Thank you to members of the Clinical Feasibility Consortium for their contributions to this article.

Human-centered artificial intelligence (AI), which emphasizes harmoniously integrating technology and predictive analyses with a human touch, stands at the forefront of transforming clinical trials.

In 2023, the ZS Clinical Feasibility Consortium explored AI’s role in clinical trial planning and design feasibility. These efforts stem from the industry’s growing recognition of the critical need to address challenges in both trial optimization and planning, as well as diverse site selection, enrollment and using data accumulated through modern technologies. These key considerations play pivotal roles in shaping clinical trials’ efficiency, effectiveness and inclusivity—ultimately ensuring successful trial launches and conduct.

In our previous article, we highlighted AI’s potential in improving clinical feasibility and underscored the array of challenges the industry must address—including questions of data access, integration, skill gaps, scalability issues and trust deficits—to unlock its effective implementation.

Within this context, we now delve into the advantages of implementing AI into clinical development and the obstacles that prevent its widespread adoption. We focus on tackling the challenges that impact the industry’s willingness to embrace AI, a pivotal effort in unlocking AI’s full potential in the clinical feasibility space.

AI’s role and impact, and industry considerations

AI could play a significant role in clinical trial optimization, informing key decision-making processes for enhancing patient enrollment and expediting study timelines. While AI should not serve as a standalone decision-making tool, it empowers human leaders by providing them with data-driven insights and recommendations. This collaborative process represents a fundamental transformation in our approach to clinical development. By integrating AI as a tool to augment existing analytical capabilities, organizations can improve their approaches to clinical trials and address critical challenges in optimizing trial design and planning.

• Trial design: AI is a powerful tool, capable of sifting through volumes of data—including relevant prior protocols and trial results, regulatory precedence and guidance, real-world data (RWD) and patient or site feedback—to refine and enrich predictive analytics models that can inform decision-making processes in trial design. The insights this ability generates can reduce protocol complexity and the burdens on patients and sites, shortening enrollment and improving overall clinical development performance.
• Trial planning: AI-driven strategies—such as utilizing historical clinical trial operational data alongside synchronous data (RWD, for example) to indicate future trends—are already proving to be valuable in trial planning. By harnessing AI to perform sophisticated analysis on frequently updated data, organizations can effectively bridge the gap between expected outcomes and actual results. AI can also be instrumental in enhancing diverse patient enrollment. Analyzing a broader range of patient data than was previously feasible, AI can identify eligible individuals who are currently unaware of their high-risk status or eligibility, as well as investigators and sites best equipped for diverse patient enrollment, enabling broader treatment access and more robust clinical trial outcomes.

AI has the potential to transform clinical trials, providing data-driven insights that empower human decision-makers while fostering a collaborative approach to clinical development. However, its full utility will only be realized when people are comfortable using it.

Trust and scalability

While AI offers vast potential in improving clinical feasibility, resistance to its implementation is common among industry decision-makers—and grounded in human behavioral factors. Factors likely to influence industry estimation of AI, and the foundation of trust necessary to advance its adoption, include the perception of its value, the sense of ownership and control over its functioning and its ultimate impact on human identity and organizational roles. Addressing these factors requires a deep understanding of both human tendencies and AI’s capabilities and limitations. Key strategies include:

• Setting expectations: Promote AI by communicating its evolution and the need for periodic model updates and ensure its interpretability in clinical trials.
• Focus on productivity: Link AI and technological infrastructure investments to cost savings and efficiency, contextualizing their impact for leadership.
• Spotting quick wins: Use AI to enhance existing processes and deliver automation, showcasing its immediate benefits and allowing broader user groups to grasp its benefits quicker.
• Building trust through feedback: Establish feedback mechanisms to fine-tune AI models, ensuring they remain adaptable and responsive.

Consortium members shared their experiences advancing AI and analytics adoption in their sponsor organizations. They acknowledged that articulating AI’s impact on individuals’ day-to-day work (termed the “what’s in it for me?”) had been a key challenge. Members reported that the following strategies had proven successful in overcoming organizational lack of trust and driving adoption:

• Coordinating communications:

  • Share clear messaging on the technology’s benefits, and the need it’s helping to fill, to emphasize how AI can support rather than replace end-users.
  • Continuously share progress on the organization’s adoption efforts, generating buy-in by curating transparency on the “how.”
  • Garner top-down leadership commitment to drive enthusiasm.

• Taking a pilot approach:

  • Share frequent and salient success stories to establish proof-of-concept, demonstrate AI’s value and support return-on-investment projections.
  • Deliver an actionable and valuable output that could not have been achieved without AI augmentation.

• Leveraging expertise:

  • Engage experts fluent in both technology and business to encourage confidence and combat the “black box” impression of AI.

Transparency, communication and the early engagement of stakeholders are critical to steering an organization toward adopting AI. These strategies focus on bringing people along for the entire adoption journey, first creating a perception of value among interested parties and then curating a sense of ownership and control. The goal is to clarify AI’s impact and its role as a support for end-users rather than a replacement. Organizations build trust in AI among their membership by emphasizing the human element often overlooked in these conversations.