Canada Workshop 2017
New features
-
Added new functions
dynSurv()
anddynLong()
, which generates survival probabilities and xpected longitudinal predictions, respectively, for a new subject conditional on their last measurement time and longitudial history. Prediction can be implemented using either a first order approximation or a Monte Carlo simulation approach. -
Added an associated
print()
method fordynSurv
anddynLong
objects. -
Added an associated
plot()
method fordynSurv
anddynLong
objects. -
Added a function
baseHaz()
for extracting the centered and uncetered estimates of the baseline hazard function. -
print()
andsummary()
now report the total computation time in addition to just the EM algorithm time. This was deemed useful after some examples showed that the time to get initial values was more expensive than the time for the MCEM algorithm to converge.
Bug patches
-
Fixed a bug that prevented models being fitted with no covariates in the survival sub-model, i.e.
Surv() ~ 1
. -
Correction to the vignette description of
mjoint()
arguments. -
Removed
enumintem
package for Sweave vignette to satisfy CRAN checks on macOS (release).
Housekeeping
-
Updated
Makevars
andMarkvars.win
to allow for OpenMP, which can be used by RcppArmadillo. -
Minor tidy-up of in-code comments.
-
Minor updates and corrections to documentation.
-
Added unit tests for new features.