Fitted survival curves from a cML-NMR fit

Ported from multinma::plot.surv_nma_summary() (Phillippo et al. 2020). Draws the model-implied survival function for each study arm, averaged over that arm's own covariate distribution (the IPD covariates for an IPD arm, the integration points for an aggregate arm), with posterior credible bands. Add geom_km() to overlay the observed Kaplan-Meier curves; systematic departures indicate that the baseline hazard is too rigid.

Usage

plot_survival(object, ..., times = NULL, ndraws = 200L, level = 0.95)

Arguments

object

A cmlnmr() fit with family = "survival".

...

Unused.

times

Time grid. Defaults to 100 points spanning the observed times.

ndraws

Number of posterior draws to summarize over. Default 200.

level

Credible level for the band. Default 0.95.

Value

A ggplot object.

Details

The baseline hazard is whatever the model was fitted with: a piecewise exponential step function, or a continuous cubic M-spline. Its posterior enters the curve through the integrated basis, so the bands include baseline uncertainty.

See also

geom_km(), cmlnmr()

Examples

if (FALSE) {
plot_survival(fit) + geom_km(fit)
}