Your AI second opinion on running a trial

A pre-mortem for your glioblastoma trial.

Match your idea against 31 historical brain cancer trials and see which past failures it most resembles.

Why this cancer

Glioblastoma is the deadliest brain cancer, and treatment hasn't fundamentally changed in 20 years.

~15 mo Median survival from diagnosis
<5% Patients alive at 5 years
2005 Last fundamental treatment update

Every GBM tumor is different. When a trial gives the same drug to everyone, the few patients who would actually respond get drowned out by the patients who never had a chance. That's the failure pattern this tool is built to catch.

Read more on the methodology page →

What this does

Describe a glioblastoma (brain cancer) trial idea: what the treatment is, who would get it, and what you'd compare it against. The tool flags which past failures your design most resembles, and why.

Example input → output An anti-angiogenic drug, given alongside standard chemo and radiation, in 600 newly-diagnosed patients (no specific subgroup), measuring both survival and tumor progression. → "HIGH risk. 3 of 3 trials with this exact profile (AVAglio, RTOG 0825, EORTC 26101) failed."

First time? Try one of these

Your trial idea

8 questions

The treatment

The patients

The trial design

A few seconds

Cross-referencing 31 trialsa few seconds

Risk verdict

Clinical detail

Failure patterns flagged

Most similar past trials

Recommendations

    What this is and isn't A research aid surfacing patterns from 31 historical glioblastoma trials. Not a regulatory or statistical predictor. A structured second opinion, not a verdict.

    Retrospective validation

    What if we had this tool in 2013?

    2,026 patients were enrolled across these three Phase 3 trials over twelve cumulative years. All three were flagged HIGH risk by the same pattern matcher, for the exact failure mode that ultimately killed each trial.

    2,026Patients enrolled
    ~$400MEstimated combined R&D spend
    3 / 3Flagged HIGH in advance
    CENTRIC · 2013
    Match

    Cilengitide added to standard care, n=545, MGMT-methylated newly-diagnosed GBM, OS endpoint.

    Tool verdict

    HIGH

    Phase 2 → Phase 3 effect-size collapse

    Actual outcome

    FAILED

    mOS 26.3 vs 26.3 mo
    HR 1.02 · p=0.86

    AVAglio · 2013
    Match

    Bevacizumab + standard chemo and radiation, n=921, all-comers newly-diagnosed GBM, co-primary OS and PFS.

    Tool verdict

    HIGH

    Pseudoresponse (anti-angiogenic class)

    Actual outcome

    FAILED

    PFS met, OS missed
    HR 0.88 · p=0.10

    CheckMate-498 · 2019
    Match

    Nivolumab vs temozolomide, n=560, MGMT-unmethylated newly-diagnosed GBM, OS endpoint.

    Tool verdict

    HIGH

    Cold-tumor immune therapy failure

    Actual outcome

    FAILED

    mOS 13.4 vs 14.9 mo
    HR 1.31 · p=0.0037

    The database

    Browse all 31 historical trials

    Each trial includes a plain-English deep-dive of why it succeeded or failed. Filter by treatment class, click any trial for the breakdown.

    Open the database→