JointMM-package: A joint modeling tool for analyzing zero-inflated...

In JiyuanHu/JointMM: A joint modeling tool for analyzing zero-inflated longitudinal proportions and time-to-event data

Description

This package is developed to investigate the association between zero-inflated longitudinal proportions and time to an event,e.g., disease onset. It is in particular applicable to microbiome proportion data from prospective studies, specifically designed to handle the zero-inflated and highly skewed longitudinal microbial proportion data to examine whether the temporal pattern of microbial presence and/or the non-zero microbial proportions are associated with differences in the time to an event.

Details

A joint modeling tool for analyzing zero-inflated longitudinal proportions and time-to-event data This package was not yet installed at build time.
Index: This package was not yet installed at build time.
This package is developed to investigate the association between zero-inflated longitudinal proportions and time to an event,e.g., disease onset. It is in particular applicable to microbiome proportion data from prospective studies, specifically designed to handle the zero-inflated and highly skewed longitudinal microbial proportion data to examine whether the temporal pattern of microbial presence and/or the non-zero microbial proportions are associated with differences in the time to an event. The demo dataset 'dat' illustrates how to organize the longitudinal and survival data for the joint modeling analysis. The main function of this package 'JointMM.func()'. In the example we illustrated how to fit JointMM to the demo data to perform joint modeling analysis.

Author(s)

Jiyuan Hu Jiyuan Hu Maintainer: Jiyuan HuJiyuan Hu <[e-mail]>

References

Hu J, Wang C, Blaser M, Li H (2021). Joint modeling of zero-inflated longitudinal proportions and time-to-event data with application to a gut microbiome study. Biometrics(Accepted).

Examples

require(JointMM)
data(dat) #load in the long format of longitudinal and survival data

# the proportion of zero RAs for the demo dataset is 52%. We recommend that a reduced model of JointMM is fitted when the zero proportion is too low (for example, <10%) and this threshold depends on the sample size and the number of observations for the data
zero.prop = mean(dat$Y==0)  

#1. full model; joint modeling
res1 = JointMM.func(data=dat,cov.name.long=c('obstime','treatment'),cov.name.surv='treatment',
                  threshold.zero.prop = 0.1,is.longi.model.only = FALSE, quad.n=10) 
#2. full model; modeling of the longitudinal part only - this is applicable when only longitudinal data is available
res2 = JointMM.func(data=dat,cov.name.long=c('obstime','treatment'),cov.name.surv='treatment',
                  threshold.zero.prop = 0.1,is.longi.model.only = TRUE, quad.n=10) 
#3. reduced model-- joint modeling analysis when the proportion of zero RA is lower than the threshold.zero.prop; Here the threshold of 0.6 is just for illustration. For sample size such like in our demo dataset, a threshold of 10% is recommended.
res3 = JointMM.func(data=dat,cov.name.long=c('obstime','treatment'),cov.name.surv='treatment',
                  threshold.zero.prop = 0.6,is.longi.model.only = FALSE, quad.n=10) 

#4. reduced model-- used when the proportion of zero RA is lower than the threshold.zero.prop; modeling of the longitudinal part only - this is applicable when only longitudinal data is available
res4 = JointMM.func(data=dat,cov.name.long=c('obstime','treatment'),cov.name.surv='treatment',
                  threshold.zero.prop = 0.6,is.longi.model.only = TRUE, quad.n=10) 
#names(res1) and names(res3) 
#c("par.est.Wald","SEs","est.hessian","Wald.Ts","pvals.Wald","status.Wald") 

#names(res2) and names(res4) 
#c("par.est.Longonly","SEs","est.hessian","Wald.Ts.Longonly","pvals.Wald.Longonly","status.Wald.Longonly")

#5. Evaluate the cross-part correlation in the longitudinal sub-model
res5 = JointMM.func(data=dat,cov.name.long=c('obstime','treatment'),cov.name.surv='treatment',
                    threshold.zero.prop = 0.1,is.longi.model.only = TRUE, cross.part.corr.eval = TRUE,quad.n=10) 

JiyuanHu/JointMM documentation built on Dec. 18, 2021, 1:37 a.m.