BLUP: Calculate best linear unbiased predictors (BLUPs) for a mixed...

In mdbrown/partlyconditional: Partly Conditional Logistic and Cox models

Description

Smooth predictors/markers through time using mixed effect models and estimate BLUPs.

Usage

BLUP(marker, measurement.time, fixed, random = NULL, id, data)

Arguments

marker	name of continuous marker to be modeled using mixed effects model. This will be used as the outcome for the mixed effects model.
measurement.time	name of time of measurement from baseline.
fixed	name of variables to be used as fixed effects.
random	name of variables to be used as random effects.
id	character name of subject id in data
data	data.frame consisting of id, marker, measurement.time, fixed and random effects used for modeling. Observations with missing data in any of these variables will be removed.

Value

An object of class "PC_BLUP" which is a list containing:

model	A 'lme' fit object from the nlme package.
fit	A data.frame consisting of id, measurement time, marker, 'fitted.blup' giving the fitted blup estimates for the marker at each measurement time for the complete data. There are also columns giving the fixed and random components of the blup for each variable in the mixed effect model. These could be used to calculate, for example, an blup estimate for the rate of change in marker values over time.
id, marker, measurement.time, fixed, random	Function inputs.

mdbrown/partlyconditional documentation built on May 22, 2019, 12:38 p.m.