BioinvestGPT ApS · Research use only
Patient-Data-Free · TRL-9 · Deployed ·
The Genome Is Enough — A TRL-9 Existence Proof
On , BioinvestGPT set out BVCT as an existence proof: a patient-data-free system that predicts a candidate drug’s clinical-benefit hazard ratio versus standard of care from the genome. On the frozen cohort to 24 Sep 2026, it scored 95.6% (674 of 705).
An existence proof for patient-data-free clinical superintelligence
Clinical biology · 95.6% across 705 scored predictions · timestamp-audited
BioinvestGPT has built, and is operating today, a domain-specific instance of experience-derived intelligence.
As far as we can tell, it is the world’s first TRL-9, deployed, patient-data-free clinical superintelligence.
Zero human-trial data ingested. The substrate is the genome — and the laws governing how a body unfolds from it.
What BVCT is
The system is BVCT (BeatSoC Virtual Clinical Trials). It predicts the clinical-benefit Hazard Ratio of any candidate drug versus the standard of care — at preclinical, Phase-1 or Phase-2 stage, before a single human is dosed.
It does so without ingesting patient data. None. Ever. Instead, it mathematically de-compresses causal disease and drug biology directly from the human genome, instantiating two million human-equivalent virtual patients on which any molecule, any modality, any indication can be simulated head-to-head against any control.
Inputs are public trial-design parameters only; outputs are HR / OR / RR vs. SoC at 0.1 precision plus a mechanistic causal rationale per prediction.
The proof
Ex-ante, prospective, cryptographically timestamped before readout, timestamp-audited.
- 705 predictions. Sent to pharmaceutical clients across sponsors, drug assets, therapeutic areas and modalities; frozen cohort as of 24 Sep 2026.
- 95.6% accuracy. 674 of 705 correct; two-sided p < 0.0001.
- 99.1% NPV · 89.7% PPV. F1 = 93.8%, DOR 953.
- 82.1% first-in-class. Share of the ledger — i.e. the system operates exactly where statistical AI is structurally blind, because no historical data exists.
- EU top-20 pharma. Auditing live predictions in an exclusive data room since April 2024; independent due diligence in April 2026 returned 9 / 10 predictions correct on the partner’s own clinical trials, median lead time ~9 months before readout.
- 2−181 null excluded. The probability that the track record is a lucky guess is below 2−181 — more than 9.0 × 1015 (253) times stronger than the 128-bit cryptographic standard (2−128) that protects the world’s financial secrets. We have, in a precise sense, mathematically excluded the null.
- Transformative. BVCT collapses a large head-to-head Phase-III trial (which can exceed €100M and 3 years) to < 0.1% of the cost in ~1 week.
- Beneficial. Every NO-GO call avoids dosing real patients in trials that the system has already predicted will fail.
- Timely. TRL 9 today. Live alpha at bvct.bioinvestgpt.com since April 2026.
- Experiential. Zero human-trial data ingested; the substrate is the genome and the laws governing how a body unfolds from it.
- General. One architecture, no retraining, working across small molecules, ADCs, in vivo CAR-T, gene therapy, mRNA, peptides, multispecific antibodies, degraders — all 14 modalities (our classification).
- Continual. The public ex-ante ledger at data.bioinvestgpt.com is extended weekly and re-validated against every new trial readout in the world.
BVCT is, on its surface, a clinical-decision platform. Underneath, it is something more interesting — an empirical existence proof that domain-general, patient-data-free, experience-derived intelligence is achievable today, in a domain where statistical AI has achieved only modest accuracy for two decades.
On its surface, a clinical-decision platform. Underneath, an existence proof that domain-general, patient-data-free, experience-derived intelligence is achievable today.
Live evidence: data.bioinvestgpt.com
Public alpha: bvct.bioinvestgpt.com
Methodology document, full evidence dossier available on request under NDA.
We are based in Copenhagen.