                 Changes in version 1.2.0 (2026-09-17)                  

  - Fix bug on power_best_binomial()
  - Improve performance of sim_power_best_binomial()
  - Add test to confirm power_best_binomial() produces a valid result
    using simulations.
  - constrain parameter in wcs_power_best_binomial() to avoid search
    probabilities lower than the difference.
  - Modify sim_power_nbinom() to explicitly indicate the direction of
    the alternative hypothesis as it could be wrongly inferred from the
    rr alone.
  - Modify multz() and multq() to follow R convention in distributions.
    A new parameter lower.tail is added
  - Fix a bug an ss_best_normal() returns now an integer value
  - ss_ni_ve() now computes the total number of events using Freedman's
    (1982) log-rank sample size formula instead of
    gsDesign::nBinomial1Sample(), which better reproduces Table 1 of
    Fleming et al. (2021) and removes the gsDesign package dependency
  - Added ve_exp/ve_ac parameters to ss_ni_ve(), allowing the
    experimental vaccine's assumed efficacy to differ from the active
    comparator's, which reproduces Table 2 of Fleming et al. (2021)
  - Added a true_diff parameter to sim_power_ni_normal() so simulations
    are no longer restricted to a true difference of 0 between groups,
    allowing sensitivity evaluation when the experimental group is truly
    slightly worse or better than control
  - Fixed prophr() to accept vector inputs for p0/hr without an
    "condition has length > 1" error
  - Corrected documentation: sim_power_best_bin_rank() selects the group
    with the highest total rank as best (roxygen said "lowest");
    power_best_normal() returns a probability, not an integer sample
    size; sim_power_ni_normal()'s and sim_power_equivalence_normal()'s
    t_level examples/docs clarified
  - Removed dead/superseded code and a missing-comma typo in
    sim_power_nbinom()'s example

                 Changes in version 1.1.0 (2026-09-06)                  

  - Added the power_events_rate() function to compute the exact binomial
    probability of observing at least a given number of events, across
    combinations of sample sizes and risks.
  - Updated the "Power for rare events" vignette (formerly "Detectable
    Event Rate for Safety Signal Detection") to cover
    power_events_rate() alongside power_single_rate().
  - Corrected several grammar, spelling, and terminology errors across
    the package vignettes.

                 Changes in version 1.0.0 (2025-06-12)                  

  - CRAN submission

                       Changes in version 0.15.0                        

  - Remove lf_config functions as they are not useful. Replace by the
    worst case scenario function.
  - Modify sim_best_binomial() function to select the best using ranks
    rather than the custom approach of which == max().
  - Make uniform the parameters names
  - Provide standardized output for the simulations

                       Changes in version 0.14.0                        

  - Re-factoring of the entire package.
  - Functions that calculate power based on simulation now start with
    the prefix sim_.
  - ni_fleming has been renamed to ni_ve.
  - The multz function was modified to correctly handle the case when
    there is only one group.
  - Improve documentation

                       Changes in version 0.13.0                        

  - Added the probhr function to estimate the probability of an event in
    the experimental group based on the control group probability and
    the hazard ratio, assuming proportional hazards.

                       Changes in version 0.12.0                        

  - Added the option to preserve the effect.
  - Modified the formula used in the Fleming method.

                       Changes in version 0.11.0                        

  - Added a 30% criterion to the ni_fleming function.

                       Changes in version 0.10.1                        

  - Updated the messages displayed by ni_fleming.

                       Changes in version 0.10.0                        

  - power_best_binomial and power_best_normal were updated to match
    their rank-based equivalents.
  - Now, dif represents the difference between the most promising group
    and the rest.
  - For binomial tests, prob is now the probability in the most
    promising group (previously it referred to the other groups), and
    dif defines how much lower the other groups are.

                        Changes in version 0.9.0                        

  - Added support for weights in power_best_norm_ranks and
    power_best_bin_rank.
  - Note: the API is not backward compatible for these two functions.

                        Changes in version 0.8.0                        

  - Added power_best_norm_rank to empirically estimate power for normal
    distributions based on ranks.

                        Changes in version 0.7.0                        

  - Added a function to calculate non-inferiority in vaccine trials
    following Fleming et al.

                        Changes in version 0.6.0                        

  - Fixed an error in power_test_binomial where, in the absence of ties,
    the sample function randomly selected a number instead of sampling
    from a range.
  - Fixed the same issue in power_best_normal.
  - power_best_binomial now returns a data frame with power and 95%
    confidence interval.
  - power_best_normal now returns a data frame with power and 95%
    confidence interval.
  - power_ni_normal now includes the number of simulations in the output
    data frame.
  - Added power_best_bin_rank, which selects the best group based on
    ranks, instead of assuming the highest value as in
    power_best_binomial.

                        Changes in version 0.5.0                        

  - Added the power_ni_normal function.

                        Changes in version 0.4.1                        

  - Fixed an error in the ggplot_prob_lowest_power graph.
  - Fixed some global variables and cleaned up imports.

                        Changes in version 0.4.0                        

  - Added the power_equivalence_normal function.

                        Changes in version 0.3.0                        

  - Added the power_best_binomial function and related helpers.

                        Changes in version 0.2.0                        

  - Added the power_best_normal function and related helpers.

                        Changes in version 0.1.1                        

  - Added a NEWS.md file to track package changes.
  - Added the power_single_rate function.
