In this video interview, Claire Riches, VP of clinical solutions at Citeline, walks through the categories of hidden protocol risk that AI-driven simulations can identify—from overly restrictive eligibility criteria to patient dropout patterns and structural trial assumptions.
In a recent video interview with Applied Clinical Trials, Claire Riches, VP of clinical solutions at Citeline, discussed how AI is transforming the way sponsors approach trial design—from endpoint selection and protocol pressure-testing to risk management and operational planning—and what clinical operations professionals just beginning to use these tools should keep in mind. She opened by reframing endpoint selection as something that can now draw on thousands of comparable trials rather than the bounded experience of a single sponsor's internal team, using AI to surface not just what has worked but what caused past trials to fail at the endpoint level, including endpoints from adjacent therapeutic areas that might not have entered standard design thinking.
On trial simulations, Riches identified three categories of design flaws that AI is particularly well-suited to surface: overly restrictive inclusion-exclusion criteria that look reasonable on paper but dramatically narrow the real-world patient funnel; patient behavior patterns including dropout rates and visit compliance that make the difference between a trial with enough data to submit and one that stalls; and structural assumptions around comparator arms and wash-out periods that can be stress-tested before a single amendment is needed.
She described the most valuable data inputs for protocol pressure-testing as the combination of analog trial performance—what was planned versus what actually happened in comparable studies—and real-world patient data that grounds enrollment assumptions in what is genuinely achievable rather than aspirational. Site-level recruitment history, she added, matters as much as site participation history.
Throughout, Riches was consistent in placing humans at the center of final decision-making, framing AI as a sophisticated advisor that generates evidence while the clinical trial team makes the call. She closed with a practical message for ClinOps professionals just starting their AI journey: the tools are accessible at every scale, the cost barrier is lower than many assume, and waiting for competitors to move first is a mistake.
