Introduction to MIIPW

Introduction

This is a vignette for our package ‘MIIPW’. It contains function for fitting GEE model for repeated measurement data. We have included mainly five function

We have included a repeated measured gene expression data in our package. Models are fitted to the dataset available in our package as below.

library(MIIPW)
data("srdata1")
head(srdata1)
#>    ID Visit   C6kine ActivinRIB ActivinRIIA ActivinRIIAB Adiponectin     AgRP
#> 5   2     1 1543.212   589.8966    277.3228     20377.32   10834.171 389.7555
#> 6   2     2 2130.282   820.4673    390.4100     38432.00   18354.472 560.1543
#> 7   2     4 1444.763   833.0092    459.7626     16478.16    8141.389 373.6979
#> 8   2     5 2661.616   860.1184    355.3753     26138.09   21286.228 506.2943
#> 9   3     1 2064.330   532.0388    435.7554     25176.20    7440.891 605.6970
#> 10  3     2 2341.346   579.0088    375.4292     24950.24   10418.850 608.5809
#>       ALCAM
#> 5  77.70429
#> 6  63.06272
#> 7        NA
#> 8  54.20246
#> 9  66.67085
#> 10 71.75843
apply(srdata1,2,anyNA)
#>           ID        Visit       C6kine   ActivinRIB  ActivinRIIA ActivinRIIAB 
#>        FALSE        FALSE         TRUE         TRUE         TRUE        FALSE 
#>  Adiponectin         AgRP        ALCAM 
#>         TRUE         TRUE         TRUE
mice::md.pattern(srdata1[,-c(1,2)],plot = TRUE)

#>     ActivinRIIAB ActivinRIB ActivinRIIA Adiponectin AgRP C6kine ALCAM   
#> 124            1          1           1           1    1      1     1  0
#> 8              1          1           1           1    1      1     0  1
#> 9              1          1           1           1    1      0     1  1
#> 1              1          1           1           1    1      0     0  2
#> 5              1          1           1           1    0      1     1  1
#> 6              1          1           1           0    1      1     1  1
#> 1              1          1           1           0    1      1     0  2
#> 1              1          1           1           0    0      1     1  2
#> 4              1          1           0           1    1      1     1  1
#> 1              1          1           0           1    0      1     1  2
#> 1              1          1           0           0    0      1     1  3
#> 2              1          0           1           1    1      1     1  1
#> 1              1          0           0           1    0      1     1  3
#>                0          3           7           9    9     10    10 48

Meanscore method

Here we have considered the response model for C6kine, depending on the other covariates in dataset srdata1. Formula object below defines the model struture. Imputation model for the methods described in can be specified through the predictor matrix function available in mice package.

formula<-C6kine~ActivinRIB+ActivinRIIA+ActivinRIIAB+Adiponectin+AgRP+ALCAM
pMat<-mice::make.predictorMatrix(srdata1[names(srdata1)%in%all.vars(formula)])
m1<-MeanScore(data=srdata1,
formula<-formula,id='ID',
visit='Visit',family='gaussian',init.beta = NULL,
init.alpha=NULL,init.phi=1,tol=.00001,weights = NULL,
corstr = 'exchangeable',maxit=50,m=2,pMat=pMat)
summary_meanscore(m1)
#> $call
#> MeanScore(data = srdata1, formula = formula <- formula, id = "ID", 
#>     visit = "Visit", family = "gaussian", init.beta = NULL, init.alpha = NULL, 
#>     init.phi = 1, tol = 1e-05, weights = NULL, corstr = "exchangeable", 
#>     maxit = 50, m = 2, pMat = pMat)
#> 
#> $inference
#>      Estimates           SE    z value   Pr(>|z|)
#> 1 672.80209372 2.333119e+02  2.8837023 0.00393030
#> 2   0.42917075 4.349044e-01  0.9868163 0.32373272
#> 3   0.35191319 6.701037e-01  0.5251623 0.59947035
#> 4   0.05025117 9.886384e-03  5.0828660 0.00000037
#> 5  -0.02466605 1.271584e-02 -1.9397898 0.05240524
#> 6   0.04980042 2.804215e-01  0.1775913 0.85904394
#> 7  -1.28003491 5.009943e-01 -2.5549889 0.01061912
#> 
#> $phi
#> [1] 319700
#> 
#> $Ralpha
#>            [,1]       [,2]       [,3]       [,4]
#> [1,]  1.0000000 -0.1103776 -0.1103776 -0.1103776
#> [2,] -0.1103776  1.0000000 -0.1103776 -0.1103776
#> [3,] -0.1103776 -0.1103776  1.0000000 -0.1103776
#> [4,] -0.1103776 -0.1103776 -0.1103776  1.0000000
#> 
#> attr(,"class")
#> [1] "summary_meanscore"

The summary_meanscore() and summary_ipw() summarises the results from fitted object obtained from meanscore and ipw functions. It provides the list of parameter estimates, wald statistics, p-value, phi value.

SIPW, AIPW, miSIPW, miAIPW

The inverse probability weighted method can be used through the function . Similarly we need to specify a predictor matrix for the imputation of score fucntion missing due to incomplete data. The argument takes the predictor matrix to be used in inside the function. The demo code for this model as follows :

m2<-SIPW(data=srdata1,formula<-formula,id='ID',
visit='Visit',family='gaussian',corstr = 'exchangeable',maxit=5)

m3<-AIPW(data=srdata1,
formula<-formula,id='ID',
visit='Visit',family='gaussian',init.beta = NULL,
init.alpha=NULL,init.phi=1,tol=.00001,weights = NULL,
corstr = 'exchangeable',maxit=50,m=3,pMat=pMat)

m4<-miSIPW(data=srdata1,
formula<-formula,id='ID',
visit='Visit',family='gaussian',init.beta = NULL,
init.alpha=NULL,init.phi=1,tol=0.001,weights = NULL,
corstr = 'exchangeable',maxit=50,m=2,pMat=pMat)

m1<-miAIPW(data=srdata1,
formula<-formula,id='ID',
 visit='Visit',family='gaussian',init.beta = NULL,
init.alpha=NULL,init.phi=1,tol=.00001,weights = NULL,
corstr = 'exchangeable',maxit=4,m=2,pMat=pMat)

Model Selection Crietrion QIC

The function provides the list of various model selection criterion based on quasi liklihood. The demo code is as follows

m1<-MeanScore(data=srdata1,
             formula<-formula,id='ID',
             visit='Visit',family='gaussian',init.beta = NULL,
             init.alpha=NULL,init.phi=1,tol=.00001,weights = NULL,
             corstr = 'exchangeable',maxit=50,m=2,pMat=pMat)
 m11<-MeanScore(data=srdata1,
             formula<-formula,id='ID',
             visit='Visit',family='gaussian',init.beta = NULL,
             init.alpha=NULL,init.phi=1,tol=.00001,weights = NULL,
            corstr = 'independent',maxit=50,m=2,pMat=pMat)
QICmiipw(model.R=m1,model.indep=m11,family="gaussian")
#>       QIC      QICu Quasi Lik 
#>  171.7651  485.0000  -78.5000
##