Guidelines & Consensus
AI-enabled clinical trials
Raynaud et alNature Reviews Bioengineering

Randomized controlled trials remain the gold standard for evaluating novel therapeutic interventions, yet conventional trial conduct is frequently impeded by substantial financial costs, stringent eligibility criteria, recruitment delays, long follow-up durations needed to assess hard endpoints, and methodological inconsistencies across medical literature. To address these operational challenges, an artificial intelligence (AI)-enabled clinical trial engineering framework establishes a structured, four-stage protocol adaptable across medical specialties: multimodal data aggregation and harmonization to develop trial-enabling tools; matching validated, fit-for-purpose algorithms to specific research questions and clinical endpoints; conducting AI-assisted trial operations; and generating expedited, data-driven go/no-go developmental decisions to identify non-promising therapies early.\n\nWithin trial execution, AI tools support patient-to-trial recruitment matching via large language models, surrogate endpoint qualification, externally matched synthetic or comparator arms, digital twin modeling, and automated data extraction. Practical implementations highlighted across medical specialties include the iBox surrogate system in organ transplantation, annualized relapse rate tracking in multiple sclerosis, heart digital twins for ventricular tachycardia, and AI-assisted histological evaluation (AIM-MASH) in hepatology. Effective clinical adoption mandates three cross-cutting governance principles: fit-for-purpose validation, continuous human oversight, and ongoing alignment with governing regulatory structures from the U.S. Food and Drug Administration (FDA), the European Medicines Agency (EMA), and the EU AI Act.
Key Takeaways
- Judicious application of an AI-enabled clinical trial framework—incorporating multimodal data harmonization, digital twins, surrogate endpoints, and automated patient matching—enables earlier evidence-based go/no-go decisions to eliminate non-promising drug candidates and accelerate trial timelines.
Read the original sourceNature Reviews Bioengineering
