Understanding Your Results

The GO / NO-GO Decision

Every BVCT result resolves to one clear, board-ready recommendation. The badge sits at the top of the result summary and encodes whether the predicted effect size clears the target threshold you set when you configured the run.

The badge is a decision-support summary, not investment advice. A CONDITIONAL GO in particular is an instruction to read the stratified breakdown — do not treat it as a soft GO. The prospective track record behind these calls is published at data.bioinvestgpt.com.

1

Read the result summary

Open a completed BVCT from the sidebar to land on the result summary. The badge is the first thing you see, followed by the per-stratum protocol cards. Read the badge first, then drop into the effect-size numbers that justify it.

Tip: A NO-GO is just as valuable as a GO — the whole point of an in-silico trial is to fail cheaply before a real one fails expensively.

Effect-Size Metrics (HR, OR, CI)

The badge is the headline; the effect-size metrics are the evidence. Each predicted effect size is reported in standard clinical-trial statistics so it slots straight into your own analyses and decks.

Reading a CI quickly: for both HR and OR, the value 1.0 means "no difference from control." An HR confidence interval that stays entirely below 1.0 (e.g. 0.48–0.88) is a significant benefit; one that straddles 1.0 (e.g. 0.78–1.21) is not yet conclusive.

Where the numbers come from

Effect sizes are model-based predictions, not observed outcomes from a physical trial. They are produced per patient stratum, which is why a single BVCT can return one HR for a biomarker-positive subgroup and a different HR for the unselected population. The Result Reliability Grid tells you how many independent data sources fed each prediction.

Always read effect size with its CI. A point estimate of HR = 0.55 looks strong, but if the interval is 0.30–1.02 it crosses 1.0 and the call should be treated as CONDITIONAL GO until the subgroups are reviewed.

Protocol Cards

Below the badge, the result lays out one protocol card per patient stratum. Each card is a self-contained mini-trial design for that subgroup, with its own predicted effect size and powering.

1

Inspect a protocol card

Each protocol card shows:

• Treatment Name — editable. Click to rename it to match your internal program code.
• Predicted Effect Size — the HR or OR for that specific patient stratum.
• Sample Size — the number of patients needed to reach 80% statistical power.
• Duration — the expected trial follow-up period.
• Download — the full protocol PDF for that stratum.
• Verify — mark the card as reviewed, which adds a green checkmark.

Tip: Renaming a treatment is cosmetic and local to your record — it does not re-run the prediction.

Verify All

Above the row of protocol cards, click Verify All to mark every protocol as reviewed in one action instead of verifying each card individually. Use it once you have read through the stratified results and are ready to sign off on the whole set.

Sample size and duration are powered for the predicted effect size. If you renegotiate the target threshold, re-run the BVCT rather than hand-editing these figures — the powering math depends on the predicted effect.

Result Reliability Grid

After a result is approved, the view renders an iOS-style grid of evidence icons. The grid is the answer to "how much should I trust this prediction?" — each icon is one independent data source feeding the model.

• Hover any icon to see its details: sensitivity, specificity, and the underlying source.
• Click the label below an icon to edit its nickname so the grid reads in your own terms.
• More icons means a more reliable prediction — a dense grid is backed by more converging evidence than a sparse one.

Read the grid alongside the CI. A tight confidence interval backed by many reliability icons is the strongest possible signal; a tight interval backed by a single source warrants more scrutiny. The grid is the qualitative complement to the quantitative effect size.

The reliability grid is also why two BVCTs with the same headline HR are not equally trustworthy: the one with more independent sources in its grid is the more defensible call.

Audio Narration & Story Player

Every completed result can narrate itself. Click the Play button on the result summary to hear a spoken walkthrough of the BVCT in clinical language — the predicted effect sizes, the GO/NO-GO rationale, and the key risks.

• Play — starts the spoken summary of the result.
• Speed selector — switch playback between 1×, 1.5×, and 2× to skim or to listen carefully.

The narration is a fast way to brief a colleague who is driving or away from the screen — it covers the same effect sizes and rationale you would read off the protocol cards, in a single spoken pass.

The narration is generated from the same model output that produced the badge and effect sizes — it summarizes the result, it does not add new claims beyond what the BVCT computed.

Visualize Rationale (Back to the Graph)

A BVCT answers "does this work and how much?" The Visualize Rationale link answers "why?" — it closes the loop back to the Hypothesis Generator's 3D knowledge graph.

Click Visualize Rationale on any completed result to navigate to the Generator with the result's evidence chain pre-loaded. The graph centers the biology that drove the prediction:

• The disease at the center of the rationale.
• The drug node being evaluated.
• The pathway genes connecting the drug to the disease.
• Publication nodes carrying the supporting — and contradicting — evidence.

This is the platform's core loop: validate a hypothesis in the Validator, then visualize why the model called it that way in the Generator. You can open the Generator directly at bvct.bioinvestgpt.com/#/generator, but launching from the Visualize Rationale link is what passes the specific result's evidence chain across.

1

Resume any record from the sidebar

The left sidebar tracks every BVCT record across its pipeline panels. Use the status filters to find what you need:

• Pending — still in progress.
• Validated — complete, with a GO/NO-GO call attached.
• Failed — an error occurred during generation.

Click any record to resume it or jump to its results, where the Visualize Rationale link lives.

Tip: Visualize Rationale is only meaningful for Validated records — Pending or Failed runs have no completed evidence chain to plot.

Research Deep-Dive, Infographic & Exports

Once you trust a result, the platform produces several layers of output — from a transmitted protocol package down to a publication-quality infographic.

Research Protocol Deep-Dive

Click Research Protocol for the most detailed view the platform produces. It assembles the full evidentiary context behind the prediction:

• The drug profile.
• The disease information page.
• High-impact publications pulled from PubMed.

Infographic Report

Click Generate Report to produce a publication-quality infographic PDF, ready for board presentations. It renders:

• Forest plots of the predicted effect sizes.
• Patient stratification breakdowns.
• Market positioning.

Per-protocol PDF download

Each protocol card carries its own Download action for that stratum's full protocol PDF. Investigator fields from your Profile auto-populate these documents, so download them after your Profile is set.

1

Confirm a transmitted result

The transmitted view confirms a completed BVCT and its outputs in one place — the GO/NO-GO call, the protocol set, and the available exports. From here you can hand the result off, generate the infographic, or open the research deep-dive.

Tip: Generate the infographic report after you have read the per-stratum cards — the forest plots and stratification panels are most useful once you know which subgroup drove the call.

Outputs are decision-support analyses. Every export — protocol PDF, deep-dive, or infographic — is a model-based prediction to inform a go/no-go decision, not a guarantee of trial outcomes. The prospective track record is published at data.bioinvestgpt.com.