pred.simml

simml.obj

a (n-by-p) matrix of new values for the covariates X at which predictions are to be made.

newX

a (n-by-L) matrix of new values for the treatment A at which predictions are to be made.

newA

a (n-by-q) matrix of new values for the covariates associated with the fitted main effect Xm at which predictions are to be made.

newXm

a length n vector specifying new values for the single-index at which predictions are to be made; the default is <code>NULL</code>.

single.index

the type of prediction required; the default "response" is on the scale of the response variable; the alternative "link" is on the scale of the linear predictors.

type

the default is <code>TRUE</code>, assuming a larger value of the outcome is better; if <code>FALSE</code>, a smaller value is assumed to be prefered.

maximize

This function makes predictions from an estimated SIMML, given a (new) set of pretreatment covariates.
The function returns a set of predicted outcomes for each treatment condition and a set of recommended treatment assignments (assuming a larger value of the outcome is better).

A major challenge in estimating treatment decision rules from a randomized clinical trial dataset with covariates measured at baseline lies in detecting relatively small treatment effect modification-related variability (i.e., the treatment-by-covariates interaction effects on treatment outcomes) against a relatively large non-treatment-related variability (i.e., the main effects of covariates on treatment outcomes). The class of Single-Index Models with Multiple-Links is a novel single-index model specifically designed to estimate a single-index (a linear combination) of the covariates associated with the treatment effect modification-related variability, while allowing a nonlinear association with the treatment outcomes via flexible link functions. The models provide a flexible regression approach to developing treatment decision rules based on patients' data measured at baseline. We refer to Park, Petkova, Tarpey, and Ogden (2020) <doi:10.1016/j.jspi.2019.05.008> and Park, Petkova, Tarpey, and Ogden (2020) <doi:10.1111/biom.13320> (that allows an unspecified X main effect) for detail of the method. The main function of this package is simml().

pred.simml: SIMML prediction function

Description

Usage

Arguments

Value

See Also