NEWS.md
In gmvjoint: Joint Models of Survival and Multivariate Longitudinal Data

gmvjoint version 0.4.0

Fairly minor updates
surv.formula now properly accepts factor variables. This shortcoming was brought to my attention by J. Piekos (Oxford); so thanks!
residuals.joint functions have been amended to actually work for Cox-Snell (there was an issue with e.g. spline time specifications).
plot.residuals.joint now produces a series of plots with meaningful legends.
ROC and dynPreds have been updated to correctly work with splines, and have been altered slightly.
joint now accepts the control argument inits which allows the user to overwrite initial conditions. The function boot.joint has been amended in line with this along with necessary documentation.
Added CITATION file with link to CSDA paper.

Fairly extensive re-writes to workhorse functions joint (and underlying EMUpdate); some computational efficiency has been gained here 'within' the EM algorithm.
Removed CreateDataMatrices function, in favour of simply taking ones from fits used in initial conditions (via glmmTMB); this greatly reduces computation times, particularly for larger models.
Added a bootstrapping function boot.joint: The approximate SEs created by the model are known to be underestimated, and so this bootstrapping function seeks to provide an alternative route; or means of comparison.
Spruced-up plotting of conditional distributions of the random effects.
Updates to S3 methods.