Description Usage Arguments Details Value Examples
Fits the linear ERR model on matched casecontrol data and performs first and second order jackknife correction
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 
data 
data frame containing matched casecontrol data, with a number of columns for doses to different locations, a column containing matched set numbers, a column containing the case's tumor location (value between 1 and the number of locations, with location x corresponding to the xth column index in 
doses 
vector containing the indices of columns containing dose information. 
set 
column index containing matched set numbers. 
status 
column index containing case status. 
loc 
column index containing the location of the matched set's case's second tumor. 
corrvars 
vector containing the indices of columns containing variables to be corrected for. 
ccmethod 
choice of method of analysis: one of meandose, CCML, CCAL or CL. Defaults to CCAL 
repar 
reparametrize to β=exp(ξ)? Defaults to 
initpars 
initial values for parameters, default is 0 for all parameters. If supplying a different vector, use a vector with an initial value for β or ξ, one for all of the other location effects and one for each other covariate (in that order). Note that if 
fitopt 
list with options to pass to 
uplimBeta 
upper limit for β=exp(ξ), default value 5. This is used for constraining the MLE estimation in some settings and for the jackknife inclusion criteria, and can be infinite except when Brent optimization is used (see details) 
profCI 
boolean: compute 95% profile likelihood confidence interval for β/ξ? Default value TRUE. 
doJK1 
perform first order jackknife correction? Automatically set to 
doJK2 
perform second order jackknife correction? Caution: this can take a very long time to run. Default value FALSE 
jkscorethresh 
square L2 norm threshold for leaveoneout and leavetwoout estimates to be included in the computation of the first and second order jackknife corrected estimate, respectively 
jkvalrange 
range of leaveoneout and leavetwoout beta/xi estimates to be allowed in the computation of the first and second order jackknife corrected estimate, respectively 
This is the main function of the package, used for fitting the linear ERR model in matched casecontrol data. Use this function to estimate the MLE (including a profile likelihood confidence interval for the dose effect) and to perform first and second order jackknife corrections.
The model being fit is HR=∑(1+β d_l)exp(α_l+X^Tγ), where the sum is over organ locations. Here β is the dose effect, α are the location effects and γ are other covariate effects. The model can be reparametrized to HR=∑(1+exp(ξ) d_l)exp(α_l+X^Tγ) using repar=TRUE
. In the original parametrization, β is constrained such that HR cannot be negative. There are different choices for the design used to estimate the parameters: mean organ dose, CCML, CL, and CCAL. Mean organ dose (ccmethod='meandose'
) uses the mean of the supplied location doses and compares that mean dose between case and matched controls. The other choices (CCML, CL and CCAL) use the tumor location for the case and compare either only between patients (CCML), only within patients (CL) or both between and within patients (CCAL). CCML only compares the same location between patients, and hence cannot be used to estimate location effects. Similarly, CL compares within patients and cannot be used to estimate covariate effects other than dose, meaning corrvars
should not be supplied for CL.
For onedimensional models (i.e., mean dose or CCML without additional covariates), the Brent algorithm is used with a search interval (10,log(uplimBeta
)) when repar=TRUE
and (L,uplimBeta
) otherwise, where L is determined by the positivity constraint for HR. For other optimizations, the LBFGSB algorithm (with constraint uplimBeta
) is used when repar=FALSE
, and the unconstrained NelderMead is used when repar=TRUE
. For details refer to the function optim
, also for fitopt
settings. Note that when supplying ndeps
to fitopt
, a value needs to be specified for every free parameter in the model. For more flexibility in optimizion, use linERRloglik
and optimize directly.
The jackknife procedure allows for filtering of the leaveoneout and leavetwoout estimates, which is important as the model can be unstable and produce extreme estimates. All estimates reaching the maximum number of iterations are excluded, as well as estimates larger than uplimBeta (if applicable). Further, the user can set a threshold for the square L2 norm of the score for an estimate (default .01), as well as an allowed value range for the β/ξ estimate itself. When the jackknife is run, the output object contains an element details
, allowing the user to inspect the produced leaveoneout and leavetwoout estimates.
Object with components MLE
and jackknife
. MLE
has components:
coef 
estimated model coefficients 
sd 
estimated standard deviation for all coefficient estimates 
vcov 
variancecovariance matrix for all estimates 
score 
score in the MLE 
convergence 
convergence code produced by the optimizer (for details refer to 
message 
convergence message produced by the optimizer 
dosepval 
pvalue for the LRT comparing the produced model with a model without dose effect. Note that the null model this is based on uses the same optimization algorithm used for the MLE, meaning onedimensional NelderMead is used when 
profCI 
the 95% profile likelihood confidence interval. In some cases one or both of the bounds of the CI cannot be obtained automatically. In that case, it is possible to use the 
fitobj 
Fit object produced by linearERRfit 
jackknife
has components firstorder
and secondorder
. Both of these have components:
coef 
the jackknifecorrected coefficient estimates 
details 
data frame with information on leaveoneout or leavetwoout estimates, with columns:

Note that the details
for the second order jackknife only include leavetwoout estimates. To access leaveoneout estimates, use details
for the first order jackknife.
1 2 3 4 5 6 7  data(linearERRdata1)
fitCCML < linearERR(data=linearERRdata1, set=1, doses=2:6, status=8,
loc=7, corrvars=9, repar=FALSE, ccmethod="CCML", doJK1=TRUE)
fitCCML$MLE$coef
fitCCML$jackknife$firstorder$coef

Add the following code to your website.
For more information on customizing the embed code, read Embedding Snippets.