plot.curelps: Plot estimated survival functions and cure probability for the promotion time cure model.

Description

The routine takes as input an object of class curelps and plots the estimated baseline survival curve, the population survival curve for a specific covariate profile and a a smooth curve for the cure probability. Approximate pointwise credible intervals are available.

Usage

# S3 method for curelps
plot(x, cred.int = 0.95, curvetype = c("baseline", "population",
     "probacure"), covar.profile, fit.col = "black", shade.col = "gray75",
      plot.cred = FALSE, np = 50, show.legend = TRUE, ...)

Arguments

x: An object of class curelps.
cred.int: The level for an approximate pointwise credible interval to be computed for the smooth curves. Default is 0.95.
curvetype: The curve to be plotted; baseline (the default) will plot the estimated baseline survival, population the population survival function for a profile of covariates given in covar.profile, and probacure the probability to be cured (for a profile of covariates given in covar.profile) given that the subject has survived until time t.
covar.profile: A numeric vector of the same length as the number of covariates in the model. This corresponds to the profile of covariates for which to compute the population survival function and cure probability estimates. The order of the covariates in covar.profile is the same as the order specified in formula of the curelps routine. Each component of covar.profile should be in the range of the observed values for the corresponding covariate. If covar.profile is left unspecified by the user, the default will be to take the median covariate values.
fit.col: The color used for the estimated survival curve.
shade.col: The color used for the shading of the credible intervals.
plot.cred: Logical. Should the credible intervals be plotted? Default is FALSE.
np: The number of points used to plot the smooth curves. Default is 50 and allowed values are between 20 and 200.
show.legend: Show the legend? Default is TRUE.
...: Further arguments to be passed to plot.

Author

Oswaldo Gressani oswaldo_gressani@hotmail.fr.

Details

When plot.cred is FALSE, the routine omits to compute the approximate pointwise credible intervals for plotting and hence is less computationally intensive.

Examples

Run this code


# Example on phase III clinical trial e1684 on melanoma data

data(ecog1684)

# Kaplan-Meier curve
plot(survfit(Surv(time, status) ~ 1, data = ecog1684), mark.time = TRUE)
fit <- curelps(Surv(time, status) ~ lt(age + trt + sex) +
             st(age + trt + sex), data = ecog1684, K = 20, penorder = 2)
fit
profile1 <- c(0, 1, 1, 0, 1, 1) # Mean age, trt = IFN, sex = Female.
profile2 <- c(0, 0, 1, 0, 0, 1) # Mean age, trt = control, sex = Female.
opar <- par(no.readonly = TRUE)
par(mfrow = c(1, 2))
plot(fit, curvetype = "probacure", plot.cred = TRUE, ylim = c(0,1),
     covar.profile = profile1, cred.int = 0.90,
     main = "Mean age, trt = IFN, sex = Female", cex.main = 0.8,
     show.legend = FALSE)
plot(fit, curvetype = "probacure", plot.cred = TRUE, ylim = c(0,1),
     covar.profile = profile2, cred.int = 0.90,
     main = "Mean age, trt = control, sex = Female", cex.main = 0.8,
     show.legend = FALSE)
par(opar)