Type: Package
Title: Joint Longitudinal and Survival Model for Big Data
Version: 0.1.3
Maintainer: Atanu Bhattacharjee <atanustat@gmail.com>
Description: Provides analysis tools for big data where the sample size is very large. It offers a suite of functions for fitting and predicting joint models, which allow for the simultaneous analysis of longitudinal and time-to-event data. This statistical methodology is particularly useful in medical research where there is often interest in understanding the relationship between a longitudinal biomarker and a clinical outcome, such as survival or disease progression. This can be particularly useful in a clinical setting where it is important to be able to predict how a patient's health status may change over time. Overall, this package provides a comprehensive set of tools for joint modeling of BIG data obtained as survival and longitudinal outcomes with both Bayesian and non-Bayesian approaches. Its versatility and flexibility make it a valuable resource for researchers in many different fields, particularly in the medical and health sciences.
Imports: JMbayes2,joineRML,rstanarm,FastJM,dplyr,nlme,survival,ggplot2
License: GPL-3
Encoding: UTF-8
LazyData: true
Depends: R (≥ 2.10)
RoxygenNote: 7.3.1
NeedsCompilation: no
Repository: CRAN
Author: Atanu Bhattacharjee [aut, cre, ctb], Bhrigu Kumar Rajbongshi [aut, ctb], Gajendra K Vishwakarma [aut, ctb]
Packaged: 2025-01-19 20:41:40 UTC; atanubhattacharjee
Date/Publication: 2025-01-19 21:00:02 UTC

Bootstrapped CI using FastJM

Description

Bootstrapped CI for predicted survival probability

Usage

cisurvfitJMCS(object)

Arguments

object

a survfitJMCS object

Value

Bootstrap CI for the survival probability and other relevant information for predicted survival plot

Examples

  
##
library(survival)
library(dplyr)
jmcs1<-jmcsBig(dtlong=data.frame(long2),
dtsurv = data.frame(surv2),
longm=y~ x7+visit,
survm=Surv(time,status)~x1+visit,
rd= ~ visit|id,
samplesize=200,id='id')
mod2<-jmcs1
P2<-survfitJMCS(model<-mod2,ids<-c(5),u<-seq(surv2[surv2$id==5,]$time,
surv2[surv2$id==5,]$time+10,0.2),obs.time='time')
bootci<-cisurvfitJMCS(P2)
print(bootci)
##

Joint model for BIG data using JMbayes2

Description

function for joint model in BIG DATA using JMbayes2

Usage

jmbayesBig(
  dtlong,
  dtsurv,
  longm,
  survm,
  samplesize = 50,
  rd,
  timeVar,
  nchain = 1,
  id,
  niter = 2000,
  nburnin = 1000
)

Arguments

dtlong

longitudinal dataset, which contains id,visit time,longitudinal measurements along with various covariates

dtsurv

survival dataset corresponding to the longitudinal dataset, with survival status and survival time

longm

fixed effect model for longitudinal response

survm

survival model

samplesize

sample size to divide the Big data

rd

random effect model part

timeVar

time variable in longitudinal model, included in the longitudinal data

nchain

number of chain for MCMC

id

name of id column in longitudinal dataset

niter

number of iteration for MCMC chain

nburnin

number of burnin sample for MCMC chain

Value

returns a list containing various output which are useful for prediction.

Author(s)

Atanu Bhattacharjee, Bhrigu Kumar Rajbongshi and Gajendra Kumar Vishwakarma

References

Rizopoulos, D., G. Papageorgiou, and P. Miranda Afonso. "JMbayes2: extended joint models for longitudinal and time-to-event data." R package version 0.2-4 (2022).

See Also

jmcsBig,jmstanBig,joinRMLBig

Examples


 
##
library(survival)
library(nlme)
library(dplyr)
fit5<-jmbayesBig(dtlong=long2,dtsurv = surv2,longm=y~ x7+visit,survm=Surv(time,status)~x1+visit,
rd= ~ visit|id,timeVar='visit',nchain=1,samplesize=200,id='id')
ydt<-long2%>%filter(id%in%c(900))
cdt<-surv2[,'id']%>%filter(id%in%c(900))
newdata<-full_join(ydt,cdt,by='id')
P2<-predJMbayes(model<-fit5,ids<-c(900),newdata=newdata,process = 'event')
plot(P2$p1[[1]])
##

Joint model for BIG data using FastJM

Description

function for joint model in BIG DATA using FastJM

Usage

jmcsBig(dtlong, dtsurv, longm, survm, samplesize = 50, rd, id)

Arguments

dtlong

longitudinal dataset, which contains id,visit time,longitudinal measurements along with various covariates

dtsurv

survival dataset corresponding to the longitudinal dataset, with survival status and survival time

longm

model for longitudinal response

survm

survival model

samplesize

sample size to divide the Big data

rd

random effect part

id

name of id column in longitudinal dataset

Value

returns a list containing various output which are useful for prediction.

Author(s)

Atanu Bhattacharjee, Bhrigu Kumar Rajbongshi and Gajendra Kumar Vishwakarma

References

Li, Shanpeng, et al. "Efficient Algorithms and Implementation of a Semiparametric Joint Model for Longitudinal and Competing Risk Data: With Applications to Massive Biobank Data." Computational and Mathematical Methods in Medicine 2022 (2022).

See Also

jmbayesBig,jmstanBig,joinRMLBig

Examples

  
##
library(survival)
library(dplyr)
fit2<-jmcsBig(dtlong=data.frame(long2),dtsurv = data.frame(surv2),
longm=y~ x7+visit,survm=Surv(time,status)~x1+visit,rd= ~ visit|id,samplesize=200,id='id')
print(fit2)
##

Joint model for BIG data using rstanarm

Description

function for joint model in BIG DATA using rstanarm package

Usage

jmstanBig(
  dtlong,
  dtsurv,
  longm,
  survm,
  samplesize = 50,
  time_var,
  id,
  nchain = 1,
  refresh = 2000
)

Arguments

dtlong

longitudinal dataset, which contains id,visit time,longitudinal measurements along with various covariates

dtsurv

survival dataset corresponding to the longitudinal dataset, with survival status and survival time

longm

model for longitudinal response

survm

survival model

samplesize

sample size to divide the Big data

time_var

time variable in longitudinal model, included in the longitudinal data

id

name of id column in longitudinal dataset

nchain

number of chain for MCMC

refresh

refresh rate for MCMC chain

Value

returns a list containing various output which are useful for prediction.

Author(s)

Atanu Bhattacharjee, Bhrigu Kumar Rajbongshi and Gajendra Kumar Vishwakarma

References

Goodrich, B., et al. "rstanarm: Bayesian applied regression modeling via Stan. R package version 2.17. 4." Online< http://mc-stan. org (2018).

See Also

jmbayesBig,jmcsBig,joinRMLBig

Examples


 
##
library(survival)
library(dplyr)
fit3<-jmstanBig(dtlong=long2,dtsurv = surv2,longm=y~ x7+visit+(1|id),
survm=Surv(time,status)~x1+visit,samplesize=200,time_var='visit',id='id')
P2<-postTraj(model<-fit3,m<-1,ids<-c(1,2,100))
pp1<-plot(P2$p1[[1]],plot_observed = TRUE)
pp2<-plot(P2$p1[[2]],plot_observed = TRUE)
pp3<-plot(P2$p1[[3]],plot_observed = TRUE)
##

Joint model for BIG data using joineRML

Description

function for joint model in BIG DATA using joineRML

Usage

joinRMLBig(dtlong, dtsurv, longm, survm, samplesize = 50, rd, timeVar, id)

Arguments

dtlong

longitudinal dataset, which contains id,visit time,longitudinal measurements along with various covariates

dtsurv

survival dataset corresponding to the longitudinal dataset, with survival status and survival time

longm

model for longitudinal response

survm

survival model

samplesize

random effect part

rd

random effect part

timeVar

time variable in longitudinal model, included in the longitudinal data

id

name of id column in longitudinal dataset

Value

returns a list containing various output which are useful for prediction.

Author(s)

Atanu Bhattacharjee, Bhrigu Kumar Rajbongshi and Gajendra Kumar Vishwakarma

References

Hickey, Graeme L., et al. "joineRML: a joint model and software package for time-to-event and multivariate longitudinal outcomes." BMC medical research methodology 18 (2018): 1-14.

See Also

jmbayesBig,jmstanBig,jmcsBig

Examples

  
##
library(survival)
library(dplyr)
fit4<-joinRMLBig(dtlong=long2,dtsurv = surv2,longm=y~ x7+visit,survm=Surv(time,status)~x1+visit,
rd=~ visit|id,timeVar='visit',samplesize=200,id='id')
P2<-predJRML(model<-fit4,ids<-c(10),dtlong=long2,dtsurv=surv2)
pp1<-plot(P2$plong[[1]])
pp1<-plot(P2$psurv[[1]])
##

longitudinal data

Description

A longitudinal dataset with single marker , with different numeric and categorical covariate

Usage

data(long2)

Format

a tibble of 13 columns and 5639 observations,

id

id value for subjects

status

survival status

time

survival time

y

longitudinal marker

visit

visit time of longitudinal measurements

x1,x2,...,x7

different numeric and categorical variable

longitudinal- survival dataset

Description

A longitudinal dataset with single marker , with different numeric and categorical covariate

Usage

data(longsurv)

Format

a tibble of 13 columns and 5639 observations,

id

id value for subjects

status

survival status

time

survival time

y

longitudinal marker

visit

visit time of longitudinal measurements

x1,x2,...,x7

different numeric and categorical variable

Plot for cisurvfitJMCS object

Description

prediction of survival probability and longitudinal marker using FastJM for BIG data

Usage

plot_cisurvfitJMCS(object)

Arguments

object

fitted survfitJMCS object

Value

Plot for predicted survival probability

Examples

  
##
library(survival)
library(dplyr)
jmcs1<-jmcsBig(dtlong=data.frame(long2),
dtsurv = data.frame(surv2),
longm=y~ x7+visit,
survm=Surv(time,status)~x1+visit,
rd= ~ visit|id,
samplesize=200,id='id')
mod2<-jmcs1
P2<-survfitJMCS(model<-mod2,ids<-c(5),u<-seq(surv2[surv2$id==5,]$time,
surv2[surv2$id==5,]$time+10,0.2),obs.time='time')
P3<-cisurvfitJMCS(P2)
plot_cisurvfitJMCS(P3)
##

Prediction using rstanarm

Description

posterior survival probability estimates from rstanarm for BIG data

Usage

postSurvfit(model, ids, ...)

Arguments

model

fitted model

ids

value of id

...

other parameter option, see posterior_survfit

Value

list of predicted value for the given id

Examples

 
##
library(survival)
library(dplyr)
jmstan<-jmstanBig(dtlong=long2,
         dtsurv = surv2,
         longm=y~ x7+visit+(1|id),
         survm=Surv(time,status)~x1+visit,
         samplesize=200,
         time_var='visit',id='id')
mod1<-jmstan
P2<-postSurvfit(model<-mod1,ids<-c(1,2,210))
pp1<-plot(P2$p1[[1]])
pp1
pp2<-plot(P2$p1[[2]])
pp2
pp3<-plot(P2$p1[[3]])
pp3
 ##

Prediction using rstanarm

Description

prediction of the posterior trajectory for longitudinal marker while using rstanarm for Big data

Usage

postTraj(model, m, ids, ...)

Arguments

model

fitted model object

m

m for posterior_traj function

ids

value of id

...

other parameter option, see posterior_traj

Value

list of predicted values for the given id

Examples

 
##
library(survival)
library(dplyr)
fit6<-jmstanBig(dtlong=long2,dtsurv = surv2,longm=y~ x7+visit+(1|id),
         survm=Surv(time,status)~x1+visit,samplesize=200,time_var='visit',id='id')
P2<-postTraj(model<-fit6,m<-1,ids<-c(1,2,100))
pp1<-plot(P2$p1[[1]],plot_observed = TRUE)
pp2<-plot(P2$p1[[2]],plot_observed = TRUE)
pp3<-plot(P2$p1[[3]],plot_observed = TRUE)
##

Prediction using JMbayes2

Description

prediction of survival probability and longitudinal marker using jmBayes2 for BIG data

Usage

predJMbayes(model, ids, process = "longitudinal", newdata, ...)

Arguments

model

fitted model object

ids

value of id

process

see jm

newdata

dataset having covariate information for the ids mentioned above.

...

other parameter options, see predict.jm

Value

list of predicted value for the given id

Examples


 
##
library(survival)
library(nlme)
library(dplyr)
jmcs1<-jmbayesBig(dtlong=long2,
dtsurv = surv2 ,
longm=y~ x7+visit,
survm=Surv(time,status)~x1+visit,
rd= ~ visit|id,
timeVar='visit',
nchain=1,
samplesize=200,
id='id')
mod3<-jmcs1
ydt<-long2%>%filter(id%in%c(900))
names(ydt)
cdt<-surv2[,'id']%>%filter(id%in%c(900))
names(cdt)
newdata<-full_join(ydt,cdt,by='id')
P2<-predJMbayes(model<-mod3,ids<-c(900),newdata=newdata,process = 'event')
plot(P2$p1[[1]])

##

Prediction using joineRML

Description

prediction of survival probability and longitudinal marker using joineRML for BIG data

Usage

predJRML(model, ids, dtlong, dtsurv, ...)

Arguments

model

fitted model object

ids

value of id

dtlong

longitudinal data

dtsurv

survival data

...

other parameter options, see dynSurv

Value

list of predicted values for the given id

Examples

   
##
library(survival)
library(dplyr)
jmcs1<-joinRMLBig(dtlong=long2,
dtsurv = surv2,
longm=y~ x7+visit,
survm=Surv(time,status)~x1+visit,
rd=~ visit|id,
timeVar='visit',
samplesize=200,
id='id')
mod4<-jmcs1
P2<-predJRML(model<-mod4,ids<-c(10),dtlong=long2,dtsurv=surv2)
plot(P2$plong[[1]])
plot(P2$psurv[[1]])
##

print.jmbayesBig

Description

print method for class 'jmbayesBig'

Usage

## S3 method for class 'jmbayesBig'
print(x, ...)

Arguments

x

fitted object

...

others

Value

prints table containing various parameter estimates, SE, P- value for both survival and longitudinal submodel, if the model is bayesian it includes their credible interval too.

Examples

 
##
library(survival)
library(dplyr)

#################################
mod3<-jmbayesBig(dtlong=long2,
dtsurv = surv2 ,
longm=y~ x7+visit,
survm=Surv(time,status)~x1+visit,
rd= ~ visit|id,
timeVar='visit',
nchain=1,
samplesize=200,
id='id')
print(mod3)

print.jmcsBig

Description

print method for class 'jmcsBig'

Usage

## S3 method for class 'jmcsBig'
print(x, ...)

Arguments

x

fitted object

...

others

Value

prints table containing various parameter estimates, SE, P- value for both survival and longitudinal submodel, if the model is bayesian it includes their credible interval too.

Examples

 
##
library(survival)
library(dplyr)
################################
mod2<-jmcsBig(dtlong=data.frame(long2),
dtsurv = data.frame(surv2),
longm=y~ x7+visit,
survm=Surv(time,status)~x1+visit,
rd= ~ visit|id,
samplesize=200,id='id')
print(mod2)

print.jmstanBig

Description

print method for class 'jmstanBig'

Usage

## S3 method for class 'jmstanBig'
print(x, ...)

Arguments

x

fitted object

...

others

Value

prints table containing various parameter estimates, SE, P- value for both survival and longitudinal submodel, if the model is bayesian it includes their credible interval too.

Examples


 
##
library(survival)
library(dplyr)
mod1<-jmstanBig(dtlong=long2,
         dtsurv = surv2,
         longm=y~ x7+visit+(1|id),
         survm=Surv(time,status)~x1+visit,
         samplesize=200,
         time_var='visit',id='id')
print(mod1)

print.joinRMLBig

Description

print method for class 'joinRMLBig'

Usage

## S3 method for class 'joinRMLBig'
print(x, ...)

Arguments

x

fitted object

...

others

Value

prints table containing various parameter estimates, SE, P- value for both survival and longitudinal submodel, if the model is bayesian it includes their credible interval too.

Examples

 
##
library(survival)
library(dplyr)
mod4<-joinRMLBig(dtlong=long2,
dtsurv = surv2,
longm=y~ x7+visit,
survm=Surv(time,status)~x1+visit,
rd=~ visit|id,
timeVar='visit',
samplesize=200,
id='id')
print(mod4)

survival data

Description

A survival dataset related the long2 dataset, with different numeric and categorical covariate

Usage

data(surv2)

Format

a tibble of 13 columns and 1000 observations,

id

id value for subjects

status

survival status

time

survival time

visit

visit time of longitudinal measurements

x1,x2,...,x7

different numeric and categorical variable

Prediction using FastJM

Description

prediction of survival probability using FastJM for BIG data

Usage

survfitJMCS(model, ids, u, method = "GH", obs.time)

Arguments

model

fitted model object

ids

value of id

u

see survfitjmcs

method

options are 'Laplace','GH'

obs.time

vector which represents time variable in the longitudinal data

Value

list of predicted value for the given id along with other information relevant for survival probability confidence plot

Examples

  
##
library(survival)
library(dplyr)
jmcs1<-jmcsBig(dtlong=data.frame(long2),
dtsurv = data.frame(surv2),
longm=y~ x7+visit,
survm=Surv(time,status)~x1+visit,
rd= ~ visit|id,
samplesize=200,id='id')
mod2<-jmcs1
P2<-survfitJMCS(model<-mod2,ids<-c(5),u<-seq(surv2[surv2$id==5,]$time,
surv2[surv2$id==5,]$time+10,0.2),obs.time='time')
print(P2)
##