Prospective In-silico Simulation of the VESALIUS-CV Trial Using Biomedical Knowledge Graph and Real-World Data-Driven AI Modeling

A locked, semi-mechanistic in silico simulation of the VESALIUS-CV trial predicted evolocumab vs placebo reduces three-point major adverse cardiovascular events (HR 0.78; 95%CI 0.70–0.87) at 54 months.

• Why it matters: Cardiovascular outcomes trials are slow and expensive, and decision-making before results are disclosed needs credible evidence from real-world data–based simulations.
• What they did: The team integrated patient-level real-world time-to-event modeling with knowledge graph-derived drug representations, then aggregated predictions into virtual trial arms using a prespecified design aligned to the protocol.
• The result: Retrospective checks showed time-dependent AUCs of 0.80–0.90 and, across 22 trials, primary-outcome prediction accuracy 0.91 (95%CI 0.77–1) with F1 0.95 (95%CI 0.86–1); the prospective HR matched direction and timing, aligning with the reported HR 0.75 (95%CI 0.65–0.86) at 55 months.

AI summary of the abstract · may contain errors · not medical advice · read the paper.