Expert role · Biostatistician

The SAP and the sample size, drafted against the real cohort.

A statistical analysis plan written from the protocol's own endpoint structure, a sample-size estimate that knows how much your eligibility criteria shrink the pool, and a DSMB charter drafted alongside them. Reviewed and signed by your statistician.

HUMAN SIGN-OFF

Power against the effect size the design actually assumes.

The job · the trial statistician who owns the analysis plan

The plan is judgement. The document around it is not.

The statistical decisions in a trial — the estimand, the analysis populations, how missing data is handled, what the interim looks at — are judgement calls that belong to a statistician. What surrounds them is a long, structured document that has to be internally consistent with a protocol that is still moving, and a sample-size calculation that has to be redone every time the design or the population does.

This workspace drafts that document and that calculation from the protocol design and the cohort analyses, and recomputes the sample size when an amendment changes the assumptions. It does not make the judgement calls. It gets the draft to the point where the statistician is spending their time on the estimand instead of on the formatting.

What it covers

  • Draft the statistical analysis plan from the protocol design
  • Estimate and re-estimate sample size as the design and population move
  • Author data-monitoring-board charter materials
  • Recompute sample size when a protocol amendment changes the assumptions

In the role registry
SAP / DSMB / sample-size authoring

What it drafts

Deliverables — drafts, until signed.

Each artefact arrives with its citations attached and its status explicit. Nothing below is final until a named human signs it.

Statistical analysis plan (SAP)

Drafted from the protocol's design and endpoint structure so the plan and the protocol start out consistent with each other — the estimand, populations, and analysis structure laid out for a statistician to take over.

Signed off by
a qualified biostatistician

Sample-size estimate

An estimate built against the design's assumed effect size and the cohort analysis — so the number is answerable to the population that actually exists, not to a convenient one.

Signed off by
a qualified biostatistician

DSMB charter

Data-monitoring-board charter materials drafted alongside the SAP, so the monitoring plan and the analysis plan are written from the same design rather than reconciled afterwards.

Signed off by
a qualified biostatistician

Sample-size recomputation on amendment

When an amendment changes the design or the population, the estimate is recomputed against the new assumptions and the previously signed version is flagged for re-review.

Signed off by
the statistician who signed the original

What it reads

It drafts from the assessment, not from memory

The role has no private knowledge. Every claim it writes traces to an analysis in one of the six pillars, and through that to a public source record.

Design

Power, the design's distance from what the indication normally does, and the amendment risk that will force the calculation to be redone.

Statistical powerDesign vs. indication normProtocol-amendment riskWhy comparable trials failed

Cohort

A sample size is only as honest as the pool it assumes. Eligibility complexity is what turns a feasible N into an infeasible one.

Cohort sizingEligibility complexityReal-world cohort

Verdict

A directional read on how likely the trial is to read out, as context for the assumptions the plan has to make.

Probability of success
The six pillars in full

How it is governed

A role drafts. A human signs.

This is the part that matters, so it is stated without hedging: the role has no authority to approve anything, and its output has no standing until a qualified person puts their name on it.

  1. The role drafts.

    It produces a draft deliverable with its citations attached. A draft has no standing: it is not approved, not final, and not actionable. The role has no authority to change it into any of those things.

  2. A named human reviews.

    The draft lands as a task for a specific person on your team. They can edit it, reject it, or send it back for a re-draft against the current evidence. Nothing moves until someone with a name has read it.

  3. Sign-off is a Part 11 electronic signature.

    Approval is recorded as a 21 CFR Part 11 electronic signature — attributable to an authenticated identity, time-stamped, and bound to the exact version of the record that was signed.

  4. The whole chain is hash-chained.

    Every draft, edit, review and signature is appended to a tamper-evident, hash-chained audit trail. It is reproducible and exportable, so your QA team can reconstruct who approved what, on what evidence, and when.

  5. Stale approvals are flagged.

    If the underlying evidence later changes, the signed decision is automatically flagged for re-review. Nothing stays approved on data that has since moved.

This role does not replace a qualified biostatistician.

It is a drafting assistant working under human review. It is not an autonomous agent, it does not make regulatory decisions, and its drafts carry no regulatory standing of their own — the signature does.

See a verdict you can actually check.

Send us a protocol — or just a molecule and an indication. We'll return a fully cited feasibility assessment you can trace, line by line, back to public data — yours to defend in a bid, take to your board or investment committee, or hand to a regulator.