Changes in version 0.1.0 (2026-06-24)                  

New Features

  - Added low_pct and high_pct arguments to derive_cutoffs_percentile()
    for flexible percentile specification (instead of
    hardcoded 80th/20th).
  - Added loess_span argument to derive_cutoffs_sensspec() to allow user
    control of smoothing in LOESS regression.
  - Added risk_col argument to dca_analysis() to support custom risk
    score column names.
  - Dynamic lower bound for decision curve plots (ylim_lower) to avoid
    clipping when net benefit is negative.
  - Biomarker name is now stored in result objects (cutoff.percentile,
    cutoff.sensspec) and referenced directly in print methods.

Improvements

  - evaluate_performance() documentation updated to clarify that
    indeterminate subjects are excluded from sensitivity/specificity
    calculations.
  - Print methods (print.cutoff.percentile, print.cutoff.sensspec) now
    display biomarker name explicitly instead of inferring from dataset.
  - Cleaner output from adjust_score() when validate = FALSE (no
    test_auc included).
  - Expanded test coverage in tests/testthat/:
      - Validation and error handling in adjust_score.
      - Edge cases and fallback cutoffs in derive_cutoffs_*.
      - Performance metrics validation for both cutoff methods.
      - New test files for compare_performance(), dca_analysis(), and
        plot_two_cutoff().

Bug Fixes

  - Fixed fragile biomarker inference in print methods.
  - Improved error messages for missing biomarker or non-binary outcome
    in adjust_score().

This NEWS file summarizes the major changes, new arguments,
documentation updates, and test coverage improvements for this release.