Intelligent Clinical Trials 2024

Feasibility Medium Impact Medium Time horizon Medium (2–4 years) Barriers Recruitment methods that are slow, inefficient and not designed for trials for which inclusion criteria are narrow and eligible patients are hard to recruit. High site burden leads to inconsistent site performance, low motivation and poor patient experiences. Biggest unlock Streamlining the recruitment process by automating recruitment and consent form completion. Alleviating site burden via automation. Increased site performance transparency. Personalizing communications using Gen AI to tailor messaging and predictive AI to identify the ideal channel. Clinical operations today Recruitment efforts rely largely on traditional methods such as physician referrals, patient registries and site-driven outreach, all of which are time-consuming and suboptimal for reaching diverse patient populations. Retention strategies involve periodic check-ins, reminders and incentives, but they often fail to engage participants throughout trials. Although digital end points are helping to address participant burden (e.g. time commitments, disruption to daily life, financial impact), dropouts remain commonplace, driven not only by participant burden but also by poor communication and logistical hurdles that complicate site visits. Participant engagement is reactive and one-size- fits-all, making limited use of predictive tools to identify patients at risk of dropout or tailoring engagement strategies to mitigate risk. Site burden remains a major pain point. Clinical operations enhanced with generative AI In the future, clinical trial teams will use Gen AI- driven tools to analyse diverse data sources – such as electronic health records (EHRs), insurance claims and even data from patient advocacy groups – to reach a more diverse participant pool. This unstructured data will be used to feed predictive models that assess the likelihood of participant dropout and suggest tailored next-best actions to improve participant experience – in essence creating personalized “marketing plans” to keep participants engaged and enrolled.Evolution, not revolution: Clinical operations, specifically patient recruitment and retention3 Sometimes people cite how only 3% of patients participate in clinical trials. But why? Because their doctors don’t want to participate. Administrative overhead is quite high, and the regulated nature of clinical trials can be an order of magnitude more burdensome to sites than care delivery. The more we can simplify what physicians and their teams need to do, the more we can augment the supply and footprint of potential trial sites. Henry Wei, Head of Development Innovation, Regeneron Researchers at Mass General Brigham have demonstrated16 how Gen AI can be used to significantly accelerate patient screening for clinical trials without sacrificing accuracy. In the COPILOT- HF study, a tailored Gen AI application used EHRs to screen for heart failure patients eligible for a trial. The application identified patients with 100% accuracy, outperforming traditional manual methods in both speed and accuracy. The tool reduced the patient screening cost to just $0.11 per patient. While the potential for streamlining patient identification and recruitment is clear, researchers stress the need for safeguards to prevent bias, protect privacy and ensure accuracy.USE CASE Mass General Brigham Intelligent Clinical Trials: Using Generative AI to Fast-Track Therapeutic Innovations 9