pIndexControl function

Auxiliary function for pIndex fitting

Auxiliary function for pIndex fitting. Typically only used internally by 'pIndexFit', but may be used to construct a control argument to either function.

pIndexControl(method = c("Efron", "Elc", "Elw", "Pic"), 
                model = c("default", "local", "threshold"), 
                ci = c("Bootstrap", "Jackknife"), weights = NULL, 
                kernel = NULL, h = 0.1, w  = seq(0.05, 0.95, 0.05), 
                alpha = 0.05, B = 0, pct = 0.5, tau=NULL)

Arguments

• method: choose either Efron' for Efron method, Elc' for conditional empirical likelihood, Elw' for weighted empirical likelihood method, and Pic' for piecewise exponential distribution. The default value is `Efron'
• model: default' for default pIndex model, local' for kernel method, `threshold' for threshold method
• ci: Method to construct confidence interval, Bootstrap' for Bootstrap method and Jackknife' for Jackknife method
• weights: case weight
• kernel: kernel funtion types, including "gaussian", "epanechnikov", "rectangular", "triangular", "biweiht", "cosine", "optcosine". The default value is `gaussian'
• h: bandwidth, defaul is 0.1
• w: percentile of biomarker value for local fit
• B: number of Bootstrap sample
• alpha: significance level (e.g. alpha=0.05)
• pct: Percentile of threshold (i.e. the cut point), default is 0.5
• tau: maximum time tau to be used for pIndex

Details

Control is used in model fitting of `pIndex'.

Returns

This function checks the internal consisitency and returns a list of value as inputed to control model fit of pIndex.

Author(s)

Bingshu E. Chen

Note

Based on code from Bingshu E. Chen.

See Also

bhm, pIndex

Examples

## To calculate the probability index for a biomarker with conditional empirical likelihood method, 
## and the corresponding 90 percent CI using Bootstrap method with 10000 bootstrap sample

ctl = pIndexControl(method = 'Elc', ci = 'Bootstrap', B = 10000, alpha = 0.1)

##
## then fit the following model
##
#  fit = pIndex(y~x1 + x2, family = 'surv', control = ctl)
##