Research topics
Virtual hospital
Running a clinical trial in computation first, using medical data and predictive models — Virtual Hospital.
What this research is trying to do
A virtual hospital raises the question of the virtual cell to the level of the patient: if a cell's response can be estimated by a model, can a patient population's response be estimated too, by integrating omics with medical data? Building virtual patient cohorts and running a trial in computation beforehand — the in-silico trial — is the central problem of this research.
What would change
Clinical trials are the most expensive stage of drug development and the latest one to fail. An in-silico trial cannot stand in for a real one, but it can stand in front of it: sweep first, in computation, for which patient subgroups the effect diverges in, where the assumptions behind a dose design break down, which biomarker should narrow the population — and the real trial goes in on a tighter, sturdier hypothesis.
Two things have to hold for this to work. A predictive model that runs from the molecular level through to phenotype, and a data foundation that integrates heterogeneous medical data so it can be handled in the same frame as omics. Neither the model nor the data alone makes a virtual patient.
Where it stands now
This work stands on the two foundations before it. The model that runs from molecule to phenotype is being built up by the virtual cell work, and the data foundation starts from omics processing (AURORA) and is widening toward medical data. The virtual patient cohort is deliberately left as the problem that comes after those two interlock.
Last updated ·