NEWS.md

In easysurv: Simplify Survival Data Analysis and Model Fitting

easysurv 2.0.1

• Removed wrapping of examples in if(FALSE){}. In write_to_xl.Rd, we now use \dontrun{} to prevent an Excel file from being created, saved and launched.
• Updated the Description field in DESCRIPTION, correcting the erroneous reference to 'flexsurvspline' as a package; it is a function. The same error was addressed in the README file and in fit_models() documentation.
• Removed dependency on the fs package. fs::path_package() was replaced with system.file().
• Wrapped external functions in square brackets in roxygen2 documentation to support auto-linking to external documentation.

easysurv 2.0.0

Major changes

• Transitioned to a tidymodels framework for survival analysis. Several updates were required to reflect this change including to function names, arguments, supporting documentation, and templates.
  • The tidymodels framework is a collection of R packages for modeling and machine learning using tidyverse principles. From the tidymodels framework, we take advantage of the parsnip and censored packages to specify models and predict survival outputs.
  • The parsnip package provides an interface to many different modeling packages, allowing for a consistent syntax for fitting models and making predictions.
  • The censored package is a parsnip extension that provides engines for various models to handle censored data in survival analysis.
• New functions!
  • inspect_surv_data() allows quick inspection of survival data.
  • get_km() replaces quick_KM() as the function to fit Kaplan-Meier curves.
  • test_ph() replaces quick_PH() as the function to test the proportional hazards assumption.
  • fit_models() allows for additional covariates to be specified in model fitting.
  • predict_and_plot() separates the generation of predictions and associated plots from the main model fitting function.

Other improvements and bug fixes

• No longer requires any additional font installation.
• Introduced print methods with the cli package for key functions to simplify and summarise outputs.
• Changed primary plotting package from ggsurvplot to ggsurvfit. ggsurvplot generates warning messages when median survival lines are added and generates misaligned risk tables as of ggplot2 version 3.5.0, while ggsurvfit is being actively maintained.
• To facilitate interpretation of Schoenfeld residual plots, scaled Schoenfeld residuals are now used, and facets are no longer produced per strata.
• plot() run on the output of fit_models() acts as a call to predict_and_plot(), generating predictions and plots for the specified model.

easysurv 1.1.0

• plot_fits() now uses flexsurv to generate survival predictions in plots, which matches the prediction method in predict_fits(). The original prediction method (via survHE) can be used instead by setting the new argument plot_predictions = "survHE" for plot_fits() and any functions that use plot_fits() (e.g., quick_fit_select(), quick_fit()).
• predict_fits() now outputs a list object that includes 95% confidence intervals for the predicted survival probabilities. CIs can be excluded by setting include_ci = FALSE. The include_ci argument is available for all functions that use predict_fits() or objects generated by predict_fits() (e.g., quick_fit_select(), quick_fit()).
• A dependency on the cli package has been added to support aesthetic and informative warning messages, such as in the quick_to_XL() function.

easysurv 1.0.0

• Initial GitHub release.

easysurv documentation built on June 24, 2024, 9:09 a.m.