sim_data {BioPETsurv}

Simulating Biomarker and Survival Observations

Description

This function simulates biomarkers and generates survival observations depending on biomarker values. The simulated data can be used to explore prognostic enrichment using surv_enrichment.

Usage

sim_data(n = 500, biomarker = "normal", effect.size = 1.25,
         baseline.hazard = "constant", end.time = 10,
         end.survival = 0.5, shape = NULL, seed = 2333)

Arguments

n

The number of observations to simulate.

biomarker

Character specifying the shape of the biomarker distribution. Choices are normal for a symmetric distribution and lognormal for a right-skewed distribution.

effect.size

The hazard ratio corresponding to one standard deviation increment in the biomarker.

baseline.hazard

Character ("constant"/"increasing"/"decreasing") specifying whether the overall hazard in the population is constant, increasing or decreasing over time.

end.time

The length of observation in the simulated dataset. In the data simulation, any events after this time will be censored at this time.

end.survival

The survival rate in the population at the end of observation.

shape

(Optional) the Weibull shape parameter for the baseline hazard. Values smaller and larger than 1 correspond to decreasing and increasing respectively.

seed

(Optional) specify the random seed used for simulation.

Details

The biomarker will be simulated from a standardized normal or lognormal distribution. It is important that effect.size should correspond to a 1 SD increment in the biomarker. Conditioning on the biomarker values and assuming proportional hazards, survival times are simulated from a Weibull distribution with user-specified shape parameter, and the scale parameter is determined by the specified event rate and effect size.

Value

Returns a list of the following items:

data

A data frame with 4 columns: the value of biomarker, observed event time, event indicator and the true event time.

km.plot

The Kaplan-Meier survival curves of the simulated dataset at enrichment levels 0, 25%, 50% and 75%.

Examples

  ## Simulate a dataset with 500 observations,
  ## where the biomarker is Normally distributed (with SD=1).
  ## The hazard ratio corresponding to every one unit of increament in the biomarker is 1.25.
  ## The observation period is 10 months,
  ## and the survival probability of the population at the end of observation is 0.5.
  ## Hazards are constant over time.
  sim_obj <- sim_data(n = 500, biomarker = "normal", effect.size = 1.25,
                     baseline.hazard = "constant", end.time = 10, end.survival = 0.5)
  dat <- sim_obj$data