seizlong: data frame on seizure counts arising from a study of...

In yags: Yet Another GEE Solver

Description

data frame on seizure counts arising from a study of progabide

Usage

data(seizlong)

Format

A data frame with 236 observations on the following 5 variables.

resp

a numeric vector of 4 successive two-week seizure counts per patient (N=59)

lbas

a numeric vector of log of the baseline two-week count

lage

a numeric vector log patient age

tx

a numeric vector 1 if progabide 0 otherwise

id

a numeric vector patient identifier

Details

obtained from Table 2 of Thall and Vail 1990

Source

Thall PF, Vail SC. Some Covariance Models for Longitudinal Count Data with Overdispersion. Biometrics 1990;46:657-671.

Examples

data(seizlong)
v4 = rep(c(0,0,0,1), 59)
f1 = yags(resp~lbas*tx+lage+v4, id=id, data=seizlong, family=poisson(), corstr="exchangeable", alphainit=.2)
f1
with(seizlong, m2LG(f1, resp, cbind(1,lbas,tx,lage,v4,lbas*tx), id=id, tim=id, invlink=function(x)exp(x), hetfac=function(m) m))
f1@del1
f2 = yags(resp~lbas*tx+lage+v4, id=id, data=seizlong, family=quasi(link="log", var="mu^2"), corstr="exchangeable", alphainit=.2)
f2
with(seizlong, m2LG(f2, resp, cbind(1,lbas,tx,lage,v4,lbas*tx), id=id, tim=id, invlink=function(x)exp(x), hetfac=function(m) m^2))
f2@del1
f3 = yags(resp~lbas*tx+lage+v4, id=id, data=seizlong, family=poisson(), corstr="ar1", cor.met=rep(1:4,59), alphainit=.2)
with(seizlong, m2LG(f3, resp, cbind(1,lbas,tx,lage,v4,lbas*tx), id=id, tim=rep(1:4,59), invlink=function(x)exp(x), hetfac=function(m) m))
f3@del1
f4 = yags(resp~lbas*tx+lage+v4, id=id, data=seizlong, family=quasi(link="log", var="mu^2"), corstr="ar1", cor.met=rep(1:4,59), alphainit=.2)
with(seizlong, m2LG(f4, resp, cbind(1,lbas,tx,lage,v4,lbas*tx), id=id, 
       tim=rep(1:4,59), invlink=function(x)exp(x), hetfac=function(m) m^2))
f4@del1

yags documentation built on May 2, 2019, 5:46 p.m.