For consultants and in-house teams

Choose a browser app or a private local workspace.

Open browser tools directly, including the vaccine analysis forms. The optional downloaded toolkit adds a local control panel for private projects, numerical examples, QC checks and generated study packages. Study definitions and names are yours; no client is preselected.

Start here: Run vaccine analyses in your browser with no installation or account · Browse all 23 modules · Download the optional local toolkit ZIP. Each module identifies its browser or local workflow.

Vaccine analyses: no installation or toolkit download

Open the vaccine application to run all six outputs using the same R numerical core as the desktop module. R runs inside the browser; no account, local R installation or command-line steps are required.

Choose your study setup. Start with a blank workspace or deliberately load the synthetic example. Upload the requested CSVs and review source-column mappings.
Confirm the definitions. Specify populations, assay and visit labels, quantification-limit rules, response thresholds and applicable diary definitions. Manually confirm the study choices, validate, then run.
Review and download. Explore complete tables and figures, missingness counts and assumptions. Download the complete ZIP, extract it, and open analysis_report.html; individual files, the setup profile and configured SAS reference are also available.

Study files stay in the browser session. Changes to inputs or settings invalidate earlier results; reset, closing or reloading clears the session. Download work you want to retain. Browser capacities and the separate desktop limits are listed in the vaccine workflow guide. Native SAS exports remain unexecuted reference code, and completed computation does not grant scientific approval.

Optional local toolkit: first use

Download and extract the ZIP. Keep the complete extracted folder together. Do not start it from inside the ZIP.
Install Python 3.10 or later if it is missing. Use the official Python installer or your organization's software portal. The launcher's missing-runtime message provides the next step. On Windows, enable the installer's PATH option if offered.
Double-click the Start Toolkit file for your system. Use Start Toolkit.command on macOS or Start Toolkit.cmd on Windows. A small local server opens the control panel in your browser. On Linux, use the supplied Start Toolkit.sh launcher through your desktop's run/executable-file option.
Create a private project from examples. Give it a neutral name such as Study workspace. The launcher makes a separate, complete copy of the toolkit and its fictional example data. Keep one project for each study or engagement.
Choose a card and click its check or start action. Missing R packages can be installed with that card's Install button. Browser tools and guides need no R. The next table explains the separate runtime installers.

If your operating system blocks a downloaded launcher, review the file and use its normal trusted-app opening workflow or your organization's software process. Keep the launcher running while local apps or jobs are active; use the control panel's Stop action for a running app or job.

Local workflows: install only the runtime your task needs

R apps and numerical examples

Install R through its normal installer or your approved software portal. The five apps require R 4.1 or later; the R template library documents R 4.4 or later. The worked study records the environment in its lockfile. Use Check, then the relevant Install action, and read the resulting log. Packages are installed in user storage. No terminal commands are required for these launcher actions.

Worked-study reports

Install Quarto using its installer. The study card checks for it and restores the study's R packages before reproduction. Restoring the lockfile may take time and may require approved system build tools on your machine. A failed check or install remains visible in the job log.

R analyses with SAS reference exports

Use guided study forms to select a design and outputs, upload CSVs, map columns and set analysis assumptions. Generate a revision, check its R environment, install missing packages through the button and run. The report and results ZIP include complete outputs and execution QC. Both TLF setup routes use R; native SAS reference exports remain available and unexecuted.

Local language-model projects

The LOCALMIND and interpretation cards prepare settings and open their source projects. Their guides describe the separately installed model runtime and model files. The toolkit does not download weights, accept model licenses or promise a working production model. The interpretation card's runnable action checks the deterministic number validator without a model.

Use your own inputs and keep your work

Start from a known example. Run it once and read its output and log before adapting it.
Use the card's input files. Download or open a listed CSV, preserve its headers and documented types/units, then upload the replacement into the selected private project. Upload validation checks the file structure; the app's checks and your review determine whether the input makes sense. UAT local input files are the parent uat_irt/templates/ files.
Set up template studies with guided forms. The SAS/R libraries, monitoring, LOCALMIND and TLF interpretation cards offer study choices, CSV upload and column mapping, a preview, and a Generate action. Save a profile to reuse the choices; download the generated ZIP and read its setup report. No study-path or variable-name code edits are needed for supported choices. Generated TLF, monitoring and aggregate interpretation packages have Check, Install and Run actions. Each run retains separate results and diagnostics. LOCALMIND requires a separately configured model service. Use Open folder or Open in editor to inspect supplied programs. The sandbox, carryover, C-QTc and worked-study run actions rebuild their documented synthetic examples. They are not general analysis forms for uploaded participant data. In particular, the worked-study SAP settings JSON is generated by its R source and overwritten on reproduction.
Save app exports before closing. Hours offers its ledger and change-order text; TRIALMON offers its worklist. UAT exports test results only, so preserve defect and traceability changes separately. Lifecycle and risk sliders are temporary what-ifs. Browser-only continuity and tracker workspaces need their own JSON/CSV exports.
Export the private project. Retain the input files, edited source, outputs and run logs together. A project ZIP preserves files; it does not capture unsaved Shiny sessions or browser localStorage. The shared tracker's database and settings live outside the workspace ZIP. Use the tracker's separate backup controls to preserve those records and settings.

Optional local study packages with guided forms

The browser vaccine app runs directly without these installation or generation steps. For local generated revisions, choose Set up R study outputs, Set up SAS study outputs, Set up monitoring checks, Set up language review , Set up vaccine analyses or Set up TLF draft report on the relevant module card.

Choose the study and scope. Select a design, outputs or checks. No client is preselected.
Map CSV inputs. Upload the requested files and map source headers. Confirm population flags, units, parameters, comparison labels and analysis windows against the study plan. Dose models require actual numeric doses.
Preview. Resolve missing columns and invalid values. Read model assumptions and exclusions. Interpretation shows an editable factual paragraph; LOCALMIND findings start unapproved.
Generate a revision. Its package retains configured inputs and programs, provenance, a setup report and a reusable profile. Generation itself does not run an analysis.
Check and run. TLF, monitoring, vaccine and interpretation revisions offer Check, Install packages and Run analysis. R is the runnable default, including the SAS TLF route. An installed R runtime is required; the toolkit links to its graphical installer. No command-line steps are required.
Review and download. Each attempt has a separate results ZIP, HTML report, QC record and software versions. Tables/listings include all rows in CSV/RDS and paginated HTML; figures include PDF/PNG. Failed outputs make the run fail and retain diagnostics. No run approves itself.
Reuse settings. Import a profile in another private project, add that project's data and preview again. Profiles transfer choices, not study data.

Native SAS exports remain unexecuted reference code. LOCALMIND needs a separately installed model service; no model, email service, scheduler or live feed is contacted by the deterministic analysis runner. Synthetic QC establishes the tested software behavior, not suitability or approval for a particular study.

Every module has a setup and use path

All local modules are included in a private project. The eCRF apps and their evaluation are explicitly separate, external projects. The cards distinguish runnable examples, browser tools, templates and references.

Module	Point-and-click actions	Runtime	What to keep in mind
Open Study Tracker	Open offline tracker; Start shared-tracker setup	Browser; Python 3.10+ for shared setup	Operational records only. Offline imports use the last supplied values; shared editing checks row versions. SMTP delivery and shared hosting need your own settings.
Continuity and handoff workspace	Open continuity workspace	Browser only	The page makes no network requests. Browser storage is tied to the browser and origin; use exports to keep projects separate.
TRIALMON safety review	Check R and packages; Install app packages; Start local app	R 4.1+ and app packages	A review aid with synthetic thresholds. It does not diagnose liver injury, adjudicate DLTs or decide escalation. Planted teaching cases are separate from the dosed participants.
Study Lifecycle Monitor	Check R and packages; Install app packages; Start local app	R 4.1+ and app packages	Operational signals only. Source-system connectors are not included. Read-only CSV exports must follow the documented schema.
IRT / RTSM UAT tracker	Check R and packages; Install app packages; Start local app	R 4.1+ and app packages	The gate packages evidence for a named human reviewer; it does not approve an IRT system. The workbook and Shiny checks are separate.
Operational risk signals	Check R and packages; Install app packages; Start local app	R 4.1+ and app packages	This is a synthetic operational illustration. Signal values, weights and thresholds need review for the intended study. No Shinylive build is included.
Find Hours reconciliation	Check R and packages; Install app packages; Start local app	R 4.1+ and app packages	Hours and effort only: no prices, rates or headcount. No connection to budgets or timesheets; no Shinylive build is included.
Worked bioequivalence study	Check study environment; Restore study packages; Run synthetic study	R, study package lockfile and Quarto	The study is synthetic and always regenerates its example data. QC is same-author re-derivation. No conformant SDTM, define.xml, ADRG or submission package is produced.
PK, NCA and bioequivalence sandbox	Run synthetic sandbox	Base R	Synthetic ADaM-shaped data with home-grown checks. NCA conventions and the simple model have stated limits; this is not CDISC conformance or a validated analysis.
Residual carryover study	Check carryover known answers; Run carryover study	Base R; optional PKNCA cross-check	Exploratory synthetic method study. Residual subtraction is a sensitivity analysis, not a regulatory recommendation.
C-QTc hysteresis study	Check C-QTc known answers; Run C-QTc study	Base R; nlme for a test cross-check	An exploratory synthetic method study. It does not replace a prespecified clinical C-QTc analysis or regulatory review.
Paired SAS and R TLF libraries	Browse SAS templates; Browse R templates; Parse R templates; Set up SAS study outputs; Set up R study outputs	R 4.4+; packages installed through the local toolkit	Software QC uses synthetic fixtures and known answers. Review methods, exclusions, units and model assumptions for each study. Native SAS reference code has not been executed; execution success is separate from study approval.
Vaccine immunogenicity and reactogenicity	Open browser app; upload and map CSVs or deliberately load the example; confirm, validate, run and download. Optional local setup remains available.	Browser only; no account, installation or toolkit download. Optional desktop route requires R.	All six outputs use the shared R numerical core. Confirm assay limits, populations, visits and missingness rules. Download before resetting or reloading. SAS references are unexecuted; no protection or efficacy conclusion is inferred.
Validation harness	Run QC harness	R and Python	A passing example does not validate a new study. Manual QC and AI-use sign-off remain separate; SAS/R parity is structural, not independently reconciled execution.
Seeded-defect check	Check defect-test environment; Install defect-test packages; Run seeded-defect check	R and Python; test packages for seeded defects	Counts describe a hand-picked list, not an estimated defect-detection rate. A change with no effect on shipped data may be classified equivalent.
TLF text number validator	Run number-validator self-test; Set up TLF draft report	Browser and local toolkit for guided previews; R/Quarto for optional report regeneration	Supports the documented AE-overview and PK-summary contracts. Number checks do not establish that a number supports the right clinical claim. Generated drafts remain unapproved and require human review.
SAS/R monitoring automation	Set up monitoring checks; Open monitoring guide	R and jsonlite for local checks; scheduler and SMTP configured separately	Eight mapped-CSV checks run locally through the generated revision. Native SAS exports remain unexecuted; scheduling, live feed connections and monitoring emails remain separate.
SAS/R monitoring coverage wiki	Open coverage wiki	Browser or PDF reader	Reference material and examples do not configure feeds or authorize clinical decisions.
Coverage playbook and runbook	Open coverage playbook	Browser or PDF reader	The playbook is a reference. It does not automatically provide cover, schedule jobs or send messages.
LOCALMIND SAS/R helpers	Set up language review; Open helper guide	Licensed SAS or R with httr2/jsonlite; separately installed local model service	All generated findings start unapproved. Installing a model service and evaluating its outputs remain separate. A configured helper is not evidence of model quality.
Biostatistics deliverables register	Open deliverables register; Open pipeline map	Browser, PDF reader or spreadsheet application	A reference map, not a record of completed study work or a generated submission package.
eCRF assist apps	Open eCRF demos	Browser; separately hosted Python runtime downloads on first use	Uploads are disabled in the public demos. Placeholder coding is not MedDRA. The host/CDN sees your IP address; this is not a validated or Part 11 system.
Within-sibling coding evaluation	Open evaluation	Browser	Synthetic evaluation results do not establish performance on a new study or coding dictionary.

Shared tracking and email

The tracker card opens its existing guided setup for accounts, permissions and optional SMTP. Preview messages before connecting your mail service. This local toolkit control panel does not publish a server, create a client deployment or send monitoring-template emails. A shared deployment uses your chosen host and its own settings.

When a check needs attention

Open the job log first: it records the attempted action, runtime findings and error. Install the named runtime through its graphical installer or your software portal, reopen the launcher if needed, and rerun the check. If a package requires a compiler or system library unavailable on your managed machine, use your approved R environment. Keep the example and modified project separate so a data or method change can be compared with the baseline.