sim_data: Simulating time-to-event data with time-dependent covariates

In Wenyi-Xie/DSP: Support Vector Machine for Dynamic Survival Prediction with Time-Dependent Covariates

Simulating time-to-event data with time-dependent covariates

Description

sim_data is used to simulate longitudinal time-to-event data. Users are allowed to specify the sample size, event rate and number of time-independent noise variables. Currently, only one time-dependent covariate can be incorporated. The inclusion of multiple time-dependent covariates will be permitted in the future versions. Users can also generate long dataset at user-specified time-points. The event time can be generated from two different distributions, either from a proportional hazard model or an accelerated failure time model.

Usage

sim_data(n, er, dist = "PH", noise = 0, time_points = NULL)

Arguments

n	the sample size
er	the event rate (range 0-1)
dist	a character variable that specify the distribution of event time. "PH" for proportional hazard model (Default), "AFT" for accelerated failure time model.
noise	the number of time-independent noise variables.
time_points	an optional numeric vector containing the user-specified time points.

Value

a list of three objects relating simulated time-to-event data with time-dependent covariates.

• T_failure is a subject-level dataframe that contains the underlying non-censored event time and observation status (1: non-censored; 0: censored).

• Z is the longitudinal dataset of time-dependent covariates at unique event time or user-specified time points (when the argument time_point is supplied). Data points are preserved even for subjects who are no longer at risk. The last four columns, in order, are time, event status (1:event; 0:non-event), unique identifier for subject and at-risk status (1:at risk; 0: not at risk). The rest leading columns are time-dependent covariates.

• T_90 is the 90th percentile of event time.

Examples

set.seed(1236)

dat <- sim_data(n=100, er=0.6)

dat <- sim_data(n=200, er=0.3, dist="AFT", noise=10)

dat <- sim_data(n=100, er=0.6, time_points = seq(from=0.1, to=1, by=0.1))

Wenyi-Xie/DSP documentation built on Sept. 14, 2022, 10:03 p.m.