Available with limitations

From clinical data to reproducible statistical results.

A statistician should be able to name the snapshot, the analysis plan and the software versions that produced a number. ValidTrials can store that chain for the methods it runs. Confirmatory analyses still need a prespecified plan and a qualified review. A convenient screen is not a statistical analysis plan.

What this capability does

Descriptive summaries

Counts and summaries can be run on a frozen snapshot.

Diagnostic accuracy

Sensitivity, specificity, Wilson and Clopper–Pearson intervals, and ROC with DeLong, where those methods are implemented.

Method comparison

Bland–Altman, Deming and Passing–Bablok are available for the studies that use them.

Plans

The analysis plan is versioned. Running a job does not approve the plan.

Outputs

A result stores the method and the software version. It is not a journal-ready table until a person checks it.

Limits

Not every model in a textbook is implemented. If a method is absent, the page and the product say so.

How it works

Freeze

The dataset checksum is recorded.

Specify

The plan names the population and the endpoint.

Run

The job uses that snapshot.

Review

A statistician accepts or rejects the output.

Connected modules and services

Questions

Does the software choose the analysis?

No. A person declares it.

Is an R session included for every method?

Only the methods that are actually implemented. There is no general claim of a full R runtime for arbitrary scripts.

Available with limitations means the described software behavior exists. A professional service is work a person does under a separate scope. Nothing here is a certification.

See the workflow on your protocol.

The conversation is about your study design and what the software can and cannot do today.