glimem {dfmeta}

The logistic regression mixed effect model.

Description

This function corresponds to the generalized logistic regression mixed effect model. This model is a model-based method for combining toxicities across the trials and cycles. We model:

log(\frac{R_{ij}(x)}{1 - R_{ij}(x)}) = \mu_{ij}(x) + Z_i,

where Z_{i}'s are assumed to be i.i.d \mathcal{N}(0, \sigma^{2}(x)). Notice that \mu_{i1}(x), \mu_{i2}(x),..., \mu_{iP}(x) = \bar{\mu}_i represents the mean toxicities on the logit scale across the cycles.

Usage

glimem(simData, sim0, sim1, family = binomial, link = "logit", nAGQ, 
  control = glmerControl(optimizer = "bobyqa"))

Arguments

simData

a data frame containing the variables named in the formula (i.e. a toxicity data for each patient at each simulation and trial.

sim0

the simulation starting point; defaults to the minimum number of the simulation in the given data frame.

sim1

the simulation ending point; defaults to the maximum number of the simulation in the given data frame.

family

a distribution family for the response variable; defaults to binomial distribution. See glm and glmer for details.

link

a specification for the model link function. This can be a name/expression, a literal character string or a length-one character vector; defaults to logit link function. See glmer for details.

nAGQ

an integer scalar - the number of points per axis for evaluating the adaptive Gauss-Hermite approximation to the log-likelihood; defaults to 1, corresponding to the Laplace approximation. Values greater than 1 produce greater accuracy in the evaluation of the log-likelihood at the expense of speed.

control

a list (of correct class, resulting from lmerControl() or glmerControl() respectively) containing control parameters, including the nonlinear optimizer to be used and parameters to be passed through to the nonlinear optimizer, see the *lmerControl* documentation for details.

Value

A list is returned, consisting of the generalized logistic regression mixed effect model's results. The output generated by this function contains the following components:

m

a summary of the generalized logistic regression model.

coeff

the random and the fixed effects coefficients for each explanatory variable for each level of each grouping factor.

simData

a data frame including the predicted values and the residuals for the selected simulation.

Author(s)

Artemis Toumazi <[e-mail]>, Sarah Zohar <[e-mail]>, Anand N. Vidyashankar <[e-mail]>, Jie Xu <[e-mail]> and Moreno Ursino <[e-mail]>

See Also

MA_estimates

Examples

  ## Not run: 
    ################################################################
    ### Give a toxicity data for each simulation and each trial. ###
    ################################################################
    
    ### Using a toxicity data for each simulation and each trial including in the dfmeta package.
    data("Toxdata")
    dim(Toxdata)
    colnames(Toxdata)
    
    ## NOTE: The toxicity dataset must have the same structure as below. ##
    ## Check it before running the VarWT function! ##
    str(Toxdata)
    
    analyf0 <- Toxdata[order(Toxdata$simulation), ]
    
    ## Using the random effect analysis's function including in the dfmeta package ##
    simData <- doseRecords(analyf0)
    NewData <- simData$doseRecords
    
    ########################################################################################
    ## In the following example the function glimem is using to find the predicted random ##  
    ###################### effect center mu only for the simulation 10 #####################
    ########################################################################################
    
    mod <- glimem(NewData, 10, 10, family = binomial, link = "logit", nAGQ = 1, 
                  control = glmerControl(optimizer = "bobyqa"))
    mod
  
## End(Not run)