Changes in version 0.3.1                        

Documentation

  - A vignette carries one task end to end on synthetic or bundled data,
    offline, and is built and run by R CMD check.
  - Runnable @examples added or corrected across 3 help pages. Each runs
    offline in seconds, writes nothing outside tempdir(), and is
    executed by R CMD check; anything needing a device, a download or an
    optional backend is fenced with the reason stated.
  - The README's quick start runs as written: it attaches the package,
    builds its own inputs, and uses only hard dependencies.

                        Changes in version 0.3.0                        

  - doseResponse(): therapy dose-response model relating a
    continuous/ordinal dose (hours/repetitions/sessions) to an outcome
    or change-from-baseline, with confounder adjustment, optional
    subject random intercepts (via fitMixedModel()), and linear / log /
    natural-spline / Emax dose forms. Returns an AnalysisResult with the
    dose effect, a predicted dose-response curve and the coefficient
    table. Honest scope: every coefficient is estimated from the data
    (no hard-coded slope), associational unless dose was randomised.

                        Changes in version 0.2.1                        

  - Relaxed the two LAPACK-derived medsens sensitivity-curve comparisons
    (acme_control/d0, ade_control/z0) in the pinned mediation fixture
    test from 1e-12 to 1e-8. These curves are RNG-invariant analytic
    functions computed through matrix/QR operations and drift by up to
    ~2.1e-10 relative across BLAS/LAPACK implementations (measured
    OpenBLAS vs reference LAPACK); the fixture was pinned under
    reference LAPACK. RNG-reproduced quantities keep 1e-12. Fixes the
    r-universe binary-check failure.

                        Changes in version 0.2.0                        

Cross-tool validation

  - Added strict offline parity fixtures for uncertainty, SCED,
    mixed/MMRM, survival, recovery, imputation, and causal inference.
    The gate rejects missing or mutated fixtures, stale surface
    inventory, direction drift, unexpected warnings, and unexecuted
    required rows.
  - Added deterministic publication CSV, JSON, and Markdown tables plus
    an independent 100-row and 100-mutation audit.

Causal-inference workflows

  - Added causalMediation(), a guarded mediation::mediate() workflow
    that verifies retained model rows, preserves the caller's RNG state,
    reports control, treated, and average ACME/ADE estimates, and
    records supported medsens() diagnostics.
  - Added targetTrialProtocol() for explicit seven-component
    target-trial declarations with a separate time-zero rule and
    provenance representations for static and dynamic strategies.
  - Added targetTrialEmulate() for intention-to-treat and per-protocol
    clone-censor-weight analyses of long person-period data. Raw and
    truncated weights, censor reasons, positivity and balance
    diagnostics, and clone-clustered robust uncertainty are retained in
    the result.
  - Added guarded WeightIt and ipw parity paths for binary baseline
    IPTW.
  - Added pinned offline backend fixtures and a 100-seed independent
    numerical validation with deliberate future-information,
    cumulative-product-grouping, and treatment-coding mutations.

These workflows estimate contrasts only under their stated consistency,
exchangeability, positivity, time-zero, and model assumptions. They do
not turn observational data into a randomized trial or prove a
biological mechanism.

                        Changes in version 0.1.0                        

Initial release of PhysioClinStats as a standalone package in the
x-biosignal ecosystem. This package provides the clinical inference
layer for rehabilitation and physiological signal studies, with heavy
modelling backends kept as optional, guarded dependencies.

New Features

  - Optional-backend detection and guarding for the clinical inference
    engine. Modelling packages are loaded only when present, so the
    package installs and loads without pulling in heavy dependencies.
      - clinStatsBackends() reports availability of the optional
        modelling backends (mmrm, emmeans, lme4, and SingleCaseES) as a
        named logical vector, letting callers check support before
        running an analysis.
      - requireBackend() asserts that a named optional backend is
        installed and otherwise stops early with an actionable install
        message.

Documentation

  - Package-level documentation describing the intended scope:
    mixed-effects / MMRM longitudinal models, single-case (N-of-1)
    designs, estimands with multiple imputation, and per-estimate
    uncertainty.