- Fixed noLD checks issue as reported by CRAN (Issue 156).
- Fixed issue 143 and return the data invisibly with
plot.singleEventCB()whentype = "hr". - Removed
family = "gbm"as it wasn't properly tested. - Added
confint.singleEventCBto compute confidence bands for the risk (or survival) function. - Updated
ERSPCdata so that the exposure variable is categorical. This may break previous code explicitly making this conversion, or somehow relying on the numerical coding.
- Fixed issue with
plot.singleEventCB()whenvisregpackage is not loaded. - Improved error message when using
family = "glmnet"with a single covariate. - Introduced
summarymethod for objects of classsingleEventCB, and improved the output ofprintby displaying the appropriate function call.
This is a Major new release
- The output of
absoluteRisk()now always contains the time variable in the first column, regardless of the length oftime. This will break earlier code that depended on the previous behaviour. - Population time plots now use
ggplot2::geom_ribbon()instead ofggplot2::geom_segment(). - Population time functions now allow for more flexible plots with user defined arguments including sequentially adding base, case, and competing event series. These are now passed as a list to the
*.paramsarguments. Several arguments are now deprecated. - Removed
popTimeExposureclass and the correspondingplotmethod.popTime()now returns anexposureattribute which contains the name of the exposure variable in the dataset. The plot method for objects of classpopTimewill use this exposure attribute to create exposure stratified population time plots.
- Major refactoring of
absoluteRisk(). Trapezoidal rule to perform numerical integration for absolute risk estimation, providing significant speed up. - Users now have further control on the output of
absoluteRisk()using the argumentstypeandaddZero. - New plotting method for time-dependent hazard functions and hazard ratios. These include confidence intervals. See
plot.singleEventCB(). The hazard function plot requires thevisregpackage to be installed. - New plotting method for cumulative incidence and survival curves. See
plot.absRiskCB(). - When
timeis unspecified,absoluteRisk()now computes the cumulative incidence atntimesequidistant points between 0 and the max failure time. absoluteRisk()can now compute the cumulative incidence for a"typical"covariate profile withnewdata = "typical". "Typical" corresponds to the median for continuous variables and the mode for factors (each variable is summarised independently).- Added
eprchd,brcancer,supportandsimdatdatasets to the package. - Implemented
riskRegression::predictRisk()method forsingleEventCBobjects.
- No longer importing the entire namespace of
data.tableandggplot2. - Moved from make the docs to pkgdown for package website.
- A warning is given when
family="gbm"and nonlinear functions of time or interactions are specified. - Add
singleEventCBclass to object returned byfitSmoothHazard() - Add
absRiskCBclass to object returned byabsoluteRisk() - Use
glmnet::prepareXto convert factors into indicator variables
- Added a
NEWS.mdfile to track changes to the package. - First release of the casebase package