BioinvestGPT ApS · Research use only
BeatSoC Virtual Clinical Trial (BVCT)
BeatSoC Virtual Clinical Trial (BVCT), developed by BioinvestGPT ApS, is a patient-data-free clinical AI infrastructure that simulates the mechanistic interaction between a drug and a disease-matched virtual human body, then outputs quantified clinical effect size against named comparators — including comparators never tested head-to-head in humans — while also providing the underlying mechanistic rationale. Research use only.
Last updated · Maintained by BioinvestGPT ApS
What is BVCT?
BVCT is developed by BioinvestGPT ApS for drug research and development. It models a drug's clinical effect size relative to a named comparator, usually the standard of care (SoC), to support decisions about targets, drug candidates, clinical programs, acquisitions and indication expansion.
BVCT outputs are modeled estimates for research use. They do not establish clinical efficacy and do not replace clinical trials in human participants.
What goes into a virtual trial?
BVCT is patient-data-free: it does not require patient records. A virtual trial is specified by the drug (its design and mechanism of action), the disease and population, the comparator, the endpoint and the trial's eligibility criteria. BVCT draws on public scientific knowledge, such as biomedical knowledge graphs, published literature and trial registries, which is distinct from private patient data.
Define the intervention, population, comparator, endpoint and time horizon before interpreting a result. A result for one combination should not be assumed to apply to another.
What does BVCT output?
A quantified clinical effect size against the named comparator: a hazard ratio (HR) for time-to-event endpoints, or an odds ratio (OR) or Relative Ratio (RR) for binary endpoints. Effect sizes are reported at 0.1 precision, meaning to one decimal place (for example, an illustrative HR 0.6), together with the underlying mechanistic rationale and a GO/NO-GO interpretation. Read each ratio with its endpoint, comparator and direction of benefit; the glossary defines each measure. A modeled value is not an observed trial result.
Where does it fit in drug development?
| Decision | Question to investigate |
|---|---|
| Target prioritization | What clinical effect could the proposed mechanism support? |
| Candidate selection | How could design differences change the effect size against the comparator? |
| Clinical development | What outcome is predicted for the specified protocol and endpoints? |
| Asset assessment | Which claims are supported by prospective evidence? |
| Indication expansion | Do the rationale and evidence support investigating a different population? |
What is BVCT's track record?
As of 24 September 2026, BVCT's company-reported prospective track record is 95.6% accuracy (674/705 scored readouts), 89.7% positive predictive value (234/261) and 99.1% negative predictive value (440/444). A further 5 readouts could not be adjudicated. Every adjudicated prediction is listed on the public prediction dashboard; the evidence page explains the cohort and how to check a prediction.