Changelog • survivalmodels

Add fully parametric survival models
Fix broken CI

Slight speed up in Akritas model by adding parameters to control number of unique time-points

Add cpp method for asserting survival matrices

Upgraded DNNSurv for keras 2.11.0

Added examples to readme
Fixes for distr6 1.6.8

Exported surv_to_risk for reducing survival matrix distributions to relative risks / rankings

Better handling of improper distributions
Fixes bug in Python conversion

Removed deprecated cutoff parameter from simsurvdata
Update risk calculations to follow https://arxiv.org/abs/2112.04828
Minor internal changes

Added tutorial with survivalmodels and mlr3proba
Bugfix for distr6 1.6.0

Add example of custom model to DNNSurv
Change default of pip to TRUE in install_ functions

Fix export of set_seed and cindex

Add set_seed for easier setting of seeds within R and Python environments

Fixed bug in risk return type when distr6 = FALSE
Added cindex for evaluating concordance of risk

Improved risk return type for discrimination

Previous risk predictions returned values where lower ranks implied higher risk, now higher rank implies higher risk. Calculated as the negative mean survival time.

Minor patch for CRAN

Patch for clang-ASAN, gcc-ASAN, valgrind

Initial release