The aim of this document is to keep track of the changes made to thedifferent versions of theR packagepencal.
The numbering of package versions follows the convention a.b.c, wherea and b are non-negative integers, and c is a positive integer. Whenminor changes are made to the package, a and b are kept fixed and only cis increased. Major changes to the package, instead, are made apparentby changing a or b.
Each section of this document corresponds to a major change in thepackage - in other words, within a section you will find all thosepackage versions a.b.x where a and b are fixed whereas x = 1, 2, 3, …Each subsection corresponds to a specific package version.
prepare_longdata is not adataframelcmm that broke the examples have been fixedrun = FALSE) to examples in?fit_mlpmms,?summarize_mlpmms and?fit_prcmlpmm: some changes have been introduced inlcmm version 2.2.0 which make the example withfit_mlpmms break. It’s unclear why this is happening, andit may take some time until the problem is solved. Until the source ofthe problem is found, the examples for the PRC MLPMM approach may failto work. The PRC LMM approach is still completely functional.getlmm andgetmlpmm functions have beenreplaced by twoS3 classes withsummarymethodssummary methods added for the output ofsteps 2summary methods for step3fitted_prclmm andfitted_prcmlpmm objectshave been refittedsurvplot_prc functionlandmark argument tosimulate_prclmm_data andsimulate_prcmlpmm_data. Examples updated accordingly andrefittedfit_prclmm andfit_prcmlpmmsurvpred_prclmm,survpred_prcmlpmm,performance_pencox andperformance_pencoxperformance_prc andperformance_pencox_baselinemetric argument toperformance_prcandperformance_pencox_baselinefit_prclmm andfit_prcmlpmmobjects so they are up to date with classes and methodspencox_baseline topencox andperformance_pencox_baseline toperformance_pencoxpbc2data and corresponding documentationCITATION file usingbibentry( ) toaddress CRAN noteDESCRIPTION file (addedbiocViews:to fixsurvcomp installation problems)LICENSE filesummarize_lmms andsummarize_mlpmms (this should yield computing time gainswith thousands of longitudinal predictors)prclmm andprcmlpmm) andcorresponding methods (print andsummary) tothe packagegetlmm andgetmlpmmcontrol argument tofit_lmms. Thisargument is used to pass control parameters tonlme::lme(see?nlme::lmeControl). See?fit_lmms for thedefaultssimulate_prclmm_data now outputs an extra element(theta.true) containing the true parameters used togenerate the dataeval( ) when creatingbaseline.covswithinsurvpred_prclmm andsurvpred_prcmlpmmseed argument tofit_lmms andfit_mlpmmssummarize_lmms in case estimation ofa LMM fails for a bootstrap replicatepfac.base.covs infit_prclmmsurvpred_prclmm whennew.longdata is provided. From this version, when allobservations of a longitudinal predictor for a given subject aremissing, a warning is produced and the corresponding random effects areset equal to 0 (population average). Previously, the function returnedan error due to theNAsstandardize argument indocumentation ofpencox_baselineperformance_prc andperformance_pencox_baseline extended to computations ofnaive tdAUC performancemax.ymissing argument tofit_lmms:with this change, it is now possible to estimate the LMMs within thePRC-LMM model even if there are subjects with missing measurements forsome (but not all) of the longitudinal outcomes. Withinsummarize_lmms, the predicted random effects when alongitudinal outcome is missing for a given subject are set = 0(marginal / population average). Settingmax.ymissing = 0disables such additional featuresummarize_lmms on subjects withoutany longitudinal information available (i.e., 100% missing on alllongitudinal variables used in step 1)purrr (now required bysummarize_lmms)CRAN dependency issue with examples insimulate_prclmm_data andsimulate_prcmlpmm_datatau.age argument tosimulate_prclmm_data andsimulate_prclmm_datafit_lmms (row181)survpred_prclmmsurvpred_prclmm fail when new datafor just 1 subject were supplied (added missingdrop = FALSE)survpred_prcmlpmmsurvpred_prc replaced by two distinctfunctions:survpred_prclmm for the PRC-LMM model, andsurvpred_prcmlpmm for the PRC-MLPMM modelfit_lmms is now more memory efficient(keep.data = F when calling lme)fit_mlpmms is now faster (parallelization implementedalso before the CBOCP is started)pencox_baseline andperformance_pencox_baselineT withTRUE)simulate_prcmlpmm_data,fit_mlpmms,summarize_mlpmms andfit_prcmlpmmperformance_prclmm toperformance_prc, andsurvpred_prclmm tosurvpred_prc (the functions work both for the PRC-LMM, andthe PRC-MLPMM)survpred_prclmm, which computespredicted survival probabilities from the fitted PRC-LMM modelfitted_prclmm data object and relateddocumentation (it is used in the examples ofperformance_prclmm)pencal package.It comprises the skeleton around which the rest of the R package will bebuiltsimulate_t_weibull andsimulate_prclmm_data);fit_lmms,summarize_lmms andfit_prclmm);performance_prclmm)R package that is user-friendly,comprehensive and well-documented is an effort that takes months,sometimes even years. This package iscurrently under activedevelopment, and many additional features and functionalities(including vignettes!) will be added incrementally with the nextreleases. If you notice a bug or something unclear in the documentation,feel free to get in touch with the maintainer of the package!