Unlearn.AI

Unlearn.AI — A model asset rather than a raw corpus: trained generative digital-twin models that forecast an individual participant's clinical outcome trajectory, built per therapeutic area from sponsor clinical trial data and natural-history data. Published evidence areas include ALS, Alzheimer's and a pan-cancer foundation model. The licensable artefacts are the fitted patient-progression generators, the synthetic control/cohort populations they emit, and the validation evidence set that was used to demonstrate twin accuracy against real trial arms. Small, science-heavy and well-funded: ~73…

Unlearn.AI offers (Alternative, Reference) — Not published. Bounded read: volume is measured in modelled participants per engagement rather than a stored corpus — a single Phase III digital-twin comparison typically simulates a control arm in the hundreds to low thousands of participants, and the evidence library spans multiple disease programmes across ALS, Alzheimer's and oncology. The generator, not a dataset, is the accumulated asset.

Synthetic control arms and comparators for randomised and single-arm studies; trial design pressure-testing before protocol lock (sample size, eligibility, endpoint and comparator choices); interim and post-readout re-analysis to extract more information from existing data; reduction of enrolment burden and control-arm size in rare-disease and neurodegenerative programmes; regulatory-simulation evidence; and for a model buyer, a rare case where a fitted foundation model of human disease progression — or its outputs — can be licensed rather than raw records

Coverage spans US; Health Care Technology; alternative, reference; equities.