Predictions for a Cox Model Fit via MPL

Compute predicted instantaneous risk or survival probabilities for a fitted coxph_mpl model.

Usage

# S3 method for class 'coxph_mpl'
predict(
  object,
  se = "M2QM2",
  type = "risk",
  i = NULL,
  time = NULL,
  upper.quantile = 0.95,
  ...
)

# S3 method for class 'predict.coxph_mpl'
plot(x, ...)

Arguments

object

A fitted model of class "coxph_mpl".

se

Inference method for confidence intervals. One of "H", "M2QM2", or "M2HM2". Default "M2QM2".

type

Prediction type: "risk" for instantaneous risk or "survival" for survival probability. Default "risk".

i

Optional integer index of the observation whose covariates are used. Defaults to mean covariates.

time

Optional numeric vector of times at which to predict. Defaults to 1000 equally spaced times over the outcome range.

upper.quantile

Quantile of the response used to bound the x-axis when plotting. Default 0.95.

...

Additional plotting parameters.

x

An object of class "predict.coxph_mpl".

Value

A data frame of class "predict.coxph_mpl" with columns time, risk or survival, se, low, and high.

Details

Predictions incorporate the baseline hazard or cumulative baseline hazard, giving absolute (not relative) risk and survival estimates. Standard errors and confidence intervals are computed via the delta method and truncated to the parameter range.

See also

[coxph_mpl()], [coxph_mpl.control()], [residuals.coxph_mpl()], [summary.coxph_mpl()]

Examples

if (FALSE) { # \dontrun{
data(lung, package = "survival")
fit_mpl <- coxph_mpl(Surv(time, status == 2) ~ age + sex + ph.karno + wt.loss,
                     data = lung)
plot(predict(fit_mpl))
} # }