Changes in version 1.1.5 (2026-04-03)                  

Bug fixes

  - Reduced total check time from >10 min to ~3 min by sharing
    pre-computed MCMC fixtures across test files and reducing vignette
    iterations.

  - Added inst/WORDLIST for domain-specific terms (HAI,
    hemagglutination, titer) flagged by CRAN spell checker.

  - Fixed GPL license badge URL in README that caused a connection
    timeout on CRAN's Windows pre-test server.

                        Changes in version 1.1.3                        

Bug fixes

  - Platform-conditional LAPACK linkage via
    configure/Makevars.in/cleanup scripts, fixing macOS Abort trap from
    double-linking while retaining Linux dpotrf_ symbol resolution.

  - .onLoad() now respects pre-existing RCPP_PARALLEL_NUM_THREADS before
    capping to 2.

  - Plot test examples wrapped in pdf(NULL)/dev.off() to prevent
    Rplots.pdf artifacts.

  - plot_diagnostics() example limited to 2 parameters to avoid graphics
    device overflow in pkgdown.

                        Changes in version 1.1.2                        

Bug fixes

  - Added LAPACK/BLAS linkage to src/Makevars for Linux/Debian, fixing
    undefined symbol: dpotrf_ installation failure on CRAN's Debian
    (clang-21) pre-test environment.

                        Changes in version 1.1.1                        

Bug fixes

  - Fixed variable length array (VLA) warnings in C++ code flagged by
    GCC 14 on Windows (-Wvla). Replaced with std::vector<double>.

  - Removed obsolete CXX_STD = CXX11 from Makevars files and
    SystemRequirements — the default C++ standard is sufficient.

                        Changes in version 1.1.0                        

New features

  - sero_reconstruct() gains a group_by argument for subgroup analysis —
    fit independent MCMCs for age groups, vaccination status, or other
    strata defined by a one-sided formula (e.g., ~age_group).

  - sero_reconstruct() gains a shared argument for joint models — run a
    single MCMC that shares measurement error ("error") and/or antibody
    boosting/waning ("boosting_waning") across groups while estimating
    group-specific infection risk. Measurement error is always shared
    when comparing groups of the same virus subtype.

  - sero_reconstruct() gains a subject_ids argument for ID-based
    individual lookup in plot_trajectory().

  - Multi-season support — add a 0-indexed integer season column to
    input data; the model estimates season-specific infection risk and
    HAI protection parameters.

  - S3 classes seroreconstruct_fit, seroreconstruct_joint, and
    seroreconstruct_multi with print() and summary() methods.

  - New plot functions: plot_trajectory(), plot_boosting(),
    plot_waning(), plot_infection_prob(), plot_diagnostics().

  - New table functions: table_parameters(), table_infections().

  - simulate_data() — generate synthetic HAI titer datasets for
    validation and power analysis.

Deprecated

  - output_model_estimate() is deprecated. Use table_parameters() and
    table_infections() instead.

                        Changes in version 1.0.0                        

  - Initial release. Core Bayesian MCMC framework for inferring
    influenza infection status, antibody dynamics, and individual
    infection risks from longitudinal HAI titer data.

  - Based on Tsang TK et al. (2022) Nat Commun 13:1557.
    https://doi.org/10.1038/s41467-022-29310-8