Insilico’s rentosertib: AI discovery and Phase 2a evidence
Insilico Medicine used AI in target discovery and molecule design for rentosertib. A June 2025 Nature Medicine paper reports a small Phase 2a trial. The findings support further study; they do not establish regulatory approval or confirmatory efficacy. This is third-party research, not a Paul Okhrem client result.
The published trial record
| Study feature | Reported detail | Interpretation limit |
|---|---|---|
| Design | 71 participants, 21 sites in China, 12 weeks | Small arms and limited follow-up. |
| Primary outcome | Safety and tolerability | Does not establish marketing approval. |
| Lung function | Mean FVC change at 60 mg daily: +98.4 mL; placebo: −20.3 mL | Exploratory efficacy signal, not confirmatory proof. |
Nature Medicine trial report, reviewed 29 October 2026.
Separate discovery from clinical validation
The research describes AI-assisted target selection and molecule design. Subsequent laboratory work, clinical testing and regulatory decisions remain distinct stages. The trial cannot isolate a causal speed advantage against a matched non-AI discovery programme.
Questions for an investment review
Paul Okhrem’s analysis: evaluate the evidence at each development stage. An interesting discovery method and a clinically validated product are different investment propositions.
- Which result is measured and which is projected?
- What evidence must the next study produce?
- Who owns clinical, regulatory and funding decisions?
See AI due diligence for the transaction review scope.
Limits of the evidence
The authors identify a small sample, geographic concentration and short follow-up. Larger and longer trials are needed to assess safety and efficacy. The page describes this dated publication, not the candidate’s current approval status.
AI drug-discovery evidence: common questions
Does a Phase 2a result establish an approved drug?
No. A development-stage result must be assessed against its endpoints, limitations and the evidence required for later clinical and regulatory decisions.
Does a discovery timeline prove an AI return on investment?
No. A return calculation also needs costs, a defensible comparison and the probability and value of later outcomes.
Change log
- : Reviewed service scope, source evidence, links and buyer answers for the October release.
- : Clarified the Phase 2a study population, duration and exploratory outcomes; separated trial evidence from approval claims.
Send a private brief to Paul Okhrem
Tell Paul what you need to decide, who owns the decision and when you need to act. Review his background and published terms.