joineRML
is an extension of the joineR package for fitting joint
models of time-to-event data and multivariate longitudinal data. The
model fitted in joineRML is an extension of the Wulfsohn and Tsiatis
(1997) and Henderson et al. (2000) models, which is comprised of
(K+1)-sub-models: a Cox proportional hazards regression model (Cox,
1972) and a K-variate linear mixed-effects model - a direct extension
of the Laird and Ware (1982) regression model. The model is fitted using
a Monte Carlo Expectation-Maximization (MCEM) algorithm, which closely
follows the methodology presented by Lin et al. (2002).
As noted in Hickey et al. (2016), there is a lack of statistical
software available for fitting joint models to multivariate longitudinal
data. This is contrary to a growing methodology in the statistical
literature. joineRML
is intended to fill this void.
The main workhorse function is mjoint
. As a simple example, we use the
heart.valve
dataset from the package and fit a bivariate joint model.
library(joineRML)
data(heart.valve)
hvd <- heart.valve[!is.na(heart.valve$log.grad) & !is.na(heart.valve$log.lvmi), ]
set.seed(12345)
fit <- mjoint(
formLongFixed = list("grad" = log.grad ~ time + sex + hs,
"lvmi" = log.lvmi ~ time + sex),
formLongRandom = list("grad" = ~ 1 | num,
"lvmi" = ~ time | num),
formSurv = Surv(fuyrs, status) ~ age,
data = list(hvd, hvd),
timeVar = "time")
The fitted model is assigned to fit
. We can apply a number of
functions to this object, e.g. coef
, logLik
, plot
, print
,
ranef
, fixef
, summary
, AIC
, getVarCov
, vcov
, confint
,
sigma
, update
, formula
, resid
, and fitted
. In addition,
several special functions have been added, including dynSurv
,
dynLong
, and baseHaz
, as well as plotting functions for objects
inheriting from the dynSurv
, dynLong
, ranef
, and mjoint
functions. For example,
summary(fit)
plot(fit, param = 'gamma')
mjoint
automatically estimates approximate standard errors using the
empirical information matrix (Lin et al., 2002), but the bootSE
function can be used as an alternative.
If you spot any errors or wish to see a new feature added, please file an issue at https://github.com/graemeleehickey/joineRML/issues or email Graeme Hickey.
For an overview of the model estimation being performed, please see the technical vignette, which can be accessed by
vignette('technical', package = 'joineRML')
For a demonstration of the package, please see the introductory vignette, which can be accessed by
vignette('joineRML', package = 'joineRML')
This project is funded by the Medical Research Council (Grant number MR/M013227/1).
To install the latest developmental version, you will need R version (version 3.3.0 or higher) and some additional software depending on what platform you are using.
If not already installed, you will need to install Rtools. Choose the version that corresponds to the version of R that you are using.
If not already installed, you will need to install Xcode Command Line Tools. To do this, open a new terminal and run
$ xcode-select --install
The latest developmental version will not yet be available on CRAN.
Therefore, to install it, you will need devtools
. You can check you
are using the correct version by running
Once the prerequisite software is installed, you can install joineRML
by running the following command in an R console
library('devtools')
install_github('graemeleehickey/joineRML')
Compatibility with broom
Tidiers methods for objects of class mjoint
(i.e. models fit with
joineRML
) are included in the
broom
package; this provides
methods that allow extracting model estimates, predictions, and
comparing models in a straightforward way.
See vignette(topic = "joineRML-broom", package = "joineRML")
for
further details and examples.
-
Cox DR. Regression models and life-tables. J R Stat Soc Ser B Stat Methodol. 1972; 34(2): 187-220.
-
Henderson R, Diggle PJ, Dobson A. Joint modelling of longitudinal measurements and event time data. Biostatistics. 2000; 1(4): 465-480.
-
Hickey GL, Philipson P, Jorgensen A, Kolamunnage-Dona R. Joint modelling of time-to-event and multivariate longitudinal outcomes: recent developments and issues. BMC Med Res Methodol. 2016; 16(1): 117.
-
Laird NM, Ware JH. Random-effects models for longitudinal data. Biometrics. 1982; 38(4): 963-974.
-
Lin H, McCulloch CE, Mayne ST. Maximum likelihood estimation in the joint analysis of time-to-event and multiple longitudinal variables. Stat Med. 2002; 21: 2369-2382.
-
Wulfsohn MS, Tsiatis AA. A joint model for survival and longitudinal data measured with error. Biometrics. 1997; 53(1): 330-339.