SandboxAQ（CryptoSense） vs Unlearn.AI

Model-based forecasts of each participant's clinical outcomes, generated from baseline data already collected, that act as a virtual comparator or as additional information per enrolled patient.

A twin-based covariate adjustment approach that Unlearn describes as EMA-qualified and FDA-supported, used to reduce sample size in randomized controlled trials while maintaining power.

Build and compare design scenarios — endpoints, eligibility criteria, sample size, assumptions — so governance decisions are made with quantified tradeoffs instead of positions.

Pre-specified twin-based analyses added to the statistical analysis plan before unblinding, plus twin-adjusted analysis delivered at interim and at readout.