Study analysis · guided forms · client-neutral

Vaccine immunogenicity
& reactogenicity

Turn mapped assay results and symptom diaries into descriptive study outputs. Choose the rules in forms, review the assumptions, then run R and download the results.

Run analyses in your browser Explore the learning companion

A study setup you can review

Open the browser workspace, choose outputs and map your CSV columns. R runs inside the browser on your computer; no R installation, account or toolkit download is needed. Start blank or deliberately load the synthetic example.

Choose & map

Select immunogenicity, diary outputs, or both. Upload the requested files and map the participant IDs, assay fields, doses and diary observations.

Specify & preview

Set the assay, visits, populations, censoring rules, response definition and diary coverage. Review validation messages and the planned assumptions.

Run & inspect

Validate the setup, run the selected analyses and inspect the tables and figures. Download the report ZIP, individual results and reusable settings to keep them.

No code editing or command-line steps are required for the supported workflow. The browser downloads its R runtime automatically on first use. Inputs and results remain in this session; reset or reload clears them. The optional local toolkit remains available for larger studies and persistent private projects.

Synthetic worked example

Try the workflow before using study data

The example deliberately contains quantification limits and incomplete diaries. Its fictional assay, thresholds and schedule demonstrate the controls; they are not scientific defaults for a new study.

Open the browser app and choose Load synthetic example. This explicitly loads the four fictional CSVs and example study choices.
Review the Study, Inputs, Immunogenicity and Reactions tabs. The three study-confirmation boxes start unchecked.
Confirm the applicable definitions, choose Validate setup, then Run selected analyses.
Inspect all six outputs. Compare complete-window incidence with partial-diary positives and recorded maximum grades, then download the complete report ZIP.

You can also inspect the fictional CSVs directly: participants, assay results, administered doses and diary. The synthetic settings profile remains available for the optional local toolkit. Reset the browser workspace and supply your own prespecified settings for study work.

Input limits and larger studies

Each CSV supports 100,000 records and 16 MB. A revision supports 50 arm labels and 50 labels each for follow-up visits, doses and symptoms. Expected participant/visit and diary grids are limited to 1,000,000 rows. Preview checks capacity before execution. Select fewer visits, doses or symptoms per revision while retaining the prespecified population and window. Diary figures are divided by dose and into groups of at most four symptoms.

Six descriptive outputs

Output	What you get	Key choice
Immunogenicity `immunogenicity`	Geometric means by arm and visit, with log-t confidence intervals and included/excluded counts.	One assay/antigen/unit combination, population and confidence level; read the result as a GMT or GMC according to the assay.
Fold rise `fold_rise`	Paired geometric mean fold rise from the selected baseline.	Baseline and follow-up visit labels; participants must have both usable values.
Response `response`	Counts and percentages meeting a configured rule, with exact Clopper–Pearson intervals.	Threshold, fold-rise or baseline-conditional rule; no universal protection cutoff.
Titer / concentration plot `titer_plot`	Descriptive geometric-mean trends across the selected visits.	Visit order and assay definition; each visit may include a different evaluable set.
Reactogenicity `reactogenicity`	Participant incidence and maximum recorded grade by arm, dose and symptom, with coverage counts.	Inclusive day window, symptom list, grade scale and denominator policy.
Diary profile `diary_profile`	Daily symptom counts and observation completeness against the administered-dose roster.	The same expected days and symptoms; absent or blank entries remain missing.

Both export routes use R for the runnable analysis. All confidence intervals are pointwise; no multiplicity adjustment is performed. Native SAS reference exports are supplied for review and adaptation; they have not been executed or reconciled against R.

Bring analysis-ready CSVs

Map your source column names to these fields in the forms. Treatment labels, assay names and study rules come from your project. A selected output determines which files are needed.

Participants

USUBJID · ARM · IMMFL · SAFEFL

One row per participant. Supply explicit immunogenicity and safety population flags. Population membership is not inferred from having an assay result or diary entry.

Assay results

USUBJID · ASSAY · ANTIGEN · UNIT · VISIT · AVAL · QUALIFIER · LLOQ · ULOQ

One assigned result per participant, assay, antigen, unit and visit. Use EXACT for a positive quantified result, BLQ/ALQ for a limit-qualified result, or MISSING. AVAL is blank for BLQ, ALQ and MISSING. Supply the relevant LLOQ/ULOQ for censored records; map your own qualifier labels in the form.

Administered doses

USUBJID · DOSE

One row per participant and actual administered dose. This roster defines who is expected to provide a diary for that dose.

Daily symptom diary

USUBJID · DOSE · DAY · SYMPTOM · GRADE

One record per participant, dose, relative day and symptom. Zero explicitly means observed absence. A blank grade or absent record means missing assessment.

Visit and grade assignment happens upstream.

Supply visit labels assigned according to the protocol and analysis plan, and confirm that assignment in setup. This version does not derive assay windows from raw collection timestamps, turn raw temperatures into grades, or adjudicate adverse events.

Make the analysis choices explicit

Geometric means and paired fold rise

The geometric mean is exp(mean(log x)) for the usable positive values. Its confidence limits exponentiate the mean log value ± a Student-t quantile times the standard error, using n − 1 degrees of freedom. Fold rise uses log(post) − log(baseline) within a participant before summarising.

These intervals assume independent participants and an appropriate distribution of log values or log ratios. Repeated visits are displayed descriptively; no longitudinal model or treatment-comparison hypothesis test is fitted. With one usable value, only the point estimate is available. With no usable values, an estimate is not available.

Quantile definition: R Student-t documentation.

Assay limits, missing values and response rules

For below-limit results, choose exclusion, half the lower limit of quantification (LLOQ), or the LLOQ. For above-limit results, choose exclusion or the upper limit of quantification (ULOQ). Substitution is an analysis assumption, not a recovered measurement; it can affect geometric means, fold rises and threshold classifications. The selected handling and counts are reported.

Choose a threshold response (post ≥ response threshold), a paired fold-rise response (post ≥ baseline × fold multiplier), or a baseline-conditional rule. For the conditional rule, baseline below the split uses the post-value threshold; baseline at or above the split uses the paired fold-rise rule. Review these inclusive response boundaries in the preview. A baseline-dependent rule needs a usable baseline as well as follow-up. A missing or excluded result is not automatically a nonresponse.

Response confidence intervals use the exact Clopper–Pearson method on the evaluable participant denominator. These can be conservative and do not account for informative missingness or validate the selected clinical threshold. R exact-binomial documentation.

Diary coverage, incidence and maximum grade

The default complete-window policy requires an assessed grade on every configured day for that participant, dose and symptom. The optional observed policy includes participants with at least one assessed day and labels the result observed-reaction incidence. Incomplete and missing coverage remains visible.

Only a recorded grade of zero establishes no symptom on that day. Maximum grade means the maximum of the available recorded grades in the configured window. Incomplete coverage can hide a higher unobserved grade. No missing assessment is filled with zero.

Grades are protocol-assigned integers from 0 through your configured maximum, which can be 1–5. Different symptoms can require different clinical grading criteria before upload. A maximum-grade table is not an SAE adjudication. FDA grading guidance.

Read the report alongside its denominator

Confirm the analysis definition. Check populations, assay/antigen/unit, visit order, censoring treatment, response rule and diary window.
Check coverage and exclusions. Compare the planned population with available observations, complete pairs, substituted results and diary completeness.
Review every selected output. Inspect the tables and plots, missing estimates, warnings and per-run QC status. Run completion does not approve a scientific conclusion.
Retain the revision and results. Download the report bundle with its data outputs, configuration and run record for review.

For context on what these measurements mean, open the illustrated learning companion. Its examples are synthetic and its optional active-inference discussion is separate from the analysis methods.

Built for study-specific descriptive work

This first version covers established descriptive immunogenicity and solicited reactogenicity. It does not fit assay dilution curves, estimate vaccine efficacy or individual protection, establish a correlate of protection, test immunobridging noninferiority, adjust for batch effects, or model longitudinal waning. Those analyses need their own prespecified methods and validation.

No client, sponsor, pathogen, vaccine platform, protection threshold or clinical decision is selected for you. Example data are synthetic. The supported forms let you configure the scope above; they do not replace the study’s protocol or statistical analysis plan.

Set up the toolkit Learn the science Technical guide