glmCensRd: Maximum likelihood estimator (MLE) for censored predictor in...

In sarahlotspeich/glmCensRd: A generalized algorithm to fit GLMs with censored covariates in R

Maximum likelihood estimator (MLE) for censored predictor in generalized linear models (GLM)

Description

Maximum likelihood estimator (MLE) for censored predictor in generalized linear models (GLM)

Usage

glmCensRd(
  Y,
  W,
  D,
  Z = NULL,
  data,
  distY = "normal",
  distX = "normal",
  rightCens = TRUE,
  robcov = TRUE,
  subdivisions = 100,
  steptol = 1e-06,
  iterlim = 100,
  verbose = FALSE
)

Arguments

Y	name of outcome variable.
W	name of observed (censored) version of X.
D	name of event indicator, defined to be = 1 if X was uncensored and 0 otherwise.
Z	(optional) name(s) of additional fully observed covariates. If none, Z = NULL (the default).
data	a dataframe containing at least columns Y, X, C, Z.
distY	distribution assumed for Y given X and Z. Default is "normal", but other options are "bernoulli", "lognormal", "gamma", "weibull", "exponential", or "poisson".
distX	distribution assumed for X given Z. Default is "normal", but other options are "lognormal", "gamma", "weibull", "exponential", or "poisson".
rightCens	logical. If TRUE (the default), right censoring is assumed. If FALSE, left censoring is assumed.
robcov	logical. If TRUE (the default), the robust sandwich covariance estimator of parameter standard errors is included in the output.
subdivisions	(fed to integrate) the maximum number of subintervals used to integrate over unobserved X for censored subjects. Default is 100.
steptol	(fed to nlm) a positive scalar providing the minimum allowable relative step length. Default is 1E-6.
iterlim	(fed to nlm) a positive integer specifying the maximum number of iterations to be performed before the program is terminated. Default is 100.
verbose	logical. If TRUE, text is printed to show progress. Default is FALSE.

Value

A list with the following elements:

outcome_model	a list containing details of the fitted model for the outcome.
predictor_model	a list containing details of the fitted model for the predictor.
code	an integer indicating why the optimization process terminated. See ?nlm for details on values.
vcov	variance-covariance matrix of the model parameters.
rvcov	robust sandwich variance-covariance matrix of the model parameters.

sarahlotspeich/glmCensRd documentation built on Aug. 19, 2024, 4:11 p.m.