Who we are
An independent lab that builds biomedical data out of mathematics instead of collecting it from people.
We make training data for the cases where real patient records are off the table. Sometimes that is a legal wall. More often the records exist but nobody will hand them over inside a year, or there were never enough of them to begin with, which is the usual story in rare disease. Trial simulation, dosing algorithms, oncology imaging: if the biology has been written down as equations somewhere, a cohort can be built from it.
Nothing here was made by training a model on an existing dataset. Each cohort comes out of published equations with their constants pinned to the literature, and no real person passes through the pipeline at any stage. The privacy question is settled by how the data is built, not by a promise on a web page.
Experiments that never had to happen
We also run trials on synthetic populations, human and animal, and sell the outcomes. Dose-response curves, virtual cohorts, animal models that no animal ever stood in for. The mathematics is the same mathematics you would use on the real thing. What is missing is the cost, the waiting and the ethics committee.
Most people who buy from us are researchers and ML teams who ran into the same wall: the data they need does exist, inside a hospital that will take eighteen months to say no.
It starts with an equation
Every dataset begins as a differential equation, a stochastic process or a published epidemiological model. Templates do not come into it.
Nobody real is in here
No patients and no animals. There are no records to leak, because none were ever collected: the populations live in the arithmetic and nowhere else.
Nothing to clear
No ethics board submission, no HIPAA exposure, no animal committee. None of it applies to data that was computed rather than collected.