NEWS.md

In psfmi: Prediction Model Pooling, Selection and Performance Evaluation Across Multiply Imputed Datasets

Package 1.4.0

• Added a new function nri_cox.R for reclassification analyses of Cox models.
• Added the possibility to pool stratified Cox models in the function psfmi_coxr.

Package 1.1.0

• Added df_com to the D1 and D2 method for small sample correction.
• Update strata option in functions cv_MI and cv_MI_RR.

Package 1.0.0

• Added the functions psfmi_lm, psfmi_lm_fw, psfmi_lr_bw for pooling and backward and forward selection of linear regression models.
• Added the functions glm_bw, glm_fw, coxph_bw and coxph_fw for backward and forward selection of linear, logistic and Cox models in a single dataset based on the likelihood ratio statistic.
• Function psfmi_perform is now deprecated, use psfmi_validate instead.
• Function bw_single is now deprecated, use glm_bw instead.
• Added the function hoslem_test and implemented this in the function pool_performance.
• added pooled concordance and R-squared measures for Cox regression to function pool_performance.
• Added the function pool_D2, to pool chi-square statistics.
• Added the function pool_D4, to pool likelihood ratio tests.
• Added the internal function pool_performance_internal, used internally by psfmi_perform.
• Option plot.indiv in function pool_performance and mivalext_lr is deprecated, use plot.method instead.
• created a new vignette for the psfmi_lm function and updated the other ones.
• corrected other bug fixes.
• Updated package website with pkgdown (included citation).

Package 0.7.1

• Added the functions pool_compare_models and pool_reclassification.
• applied some bug fixes in the function bw_single where selection with anova was replaced with Anova.
• added internal function RR_diff_prop to pool difference in proportions and related SE with RR.
• Updated vignettes.
• Updated package website with pkgdown.
• Added datasets to be used in tutorials.

Package 0.5.0

• Added forward selection to the psfmi_lr and psfmi_coxr functions.
• Added the functions psfmi_lr_fw, psfmi_lr_bw, psfmi_coxr_fw and psfmi_coxr_bw that are called by psfmi_lr or psfmi_coxr for forward and backward selection after MI with logistic and Cox regression models.
• Extended the function psfmi_perform by including the possibility of using cross-validation in combination with multiple imputation.
• Added the functions cv_MI, cv_MI_RR, MI_cv_naive, boot_MI and MI_boot, that are called by psfmi_perform to combine cross-validation or bootstrapping with MI.
• Added the function bw_single for backward selection in a single dataset.
• Added the functions pool_auc, rsq_nagel and scaled_brier.
• Added the internal functions mean_auc_log, clean_P, miceImp.
• Updated vignettes.
• Created a package website with pkgdown.

Package 0.2.0

• Added functions pool_intadj.R, pool_performance.R, psfmi_mm.R, psfmi_mm_multiparm, psfmi_perform, psfmi_stab.
• Added the class statement smodsmi to the functions psfmi_lr, psfmi_coxr and psfmi_mm.
• Added the following output object information to psfmi_lr, psfmi_coxr and psfmi_mm: model_type, predictors_in, predictors_out, fit.formula, predictors_final and predictors_initial.
• Excluded function miperform_lr.
• Changed name object multiparm_p into multiparm.
• Updated vignettes.

psfmi documentation built on July 9, 2023, 7:02 p.m.