The authors link time-to-event models with longitudinal models through shared latent variables when the time of the event of interest is known only to lie within an interval. The context of tree growth and mortality studies presents a natural application of shared parameter joint modelling where a latent feature of each tree impacts both mortality and growth. The authors' developments are motivated by such an application, with the additional caveat that event-times are not known exactly, since the trees are subject to intermittent observation, with the time between measurements extending into decades or longer. Such interval censoring is a common occurrence in similar long-term experiments in resource management, ecology and health research. The additional numerical complexity resulting from interval censored time-to-event data often makes inference for joint models prohibitive. The authors examine properties of three event-time imputation methods that enable application of now standard joint modelling techniques to interval censored time-to-event data. The imputation techniques include the midpoint method, a kernel smoothing method, and a backsolve method which incorporates information from the longitudinal trajectory. Joint analysis of a designed, long-term, forestry experiment is presented, accompanied by a simulation study investigating the properties of the three event-time imputation techniques. The Canadian Journal of Statistics 39: 438–457; 2011 © 2011 Statistical Society of Canada
Les auteurs relient les modeles de duree de vie avec les modeles longitudinaux a l'aide de variables latentes lorsque la duree de vie n'est pas observee exactement, mais que nous savons qu'elle appartient a un certain intervalle. Le contexte des etudes de croissance et de mortalite des arbres est un domaine naturel d'application de modelisation de parametres partages ou la caracteristique latente de chaque arbre a un impact sur la mortalite et la croissance. Les developpements sont motives par une telle application, mais il faut specifier que les durees de vie ne sont pas observees exactement puisque les arbres sont observes que de facon intermittente et qu'il peut se passer des dizaines d'annees (sinon plus) entre deux observations. Une telle censure par intervalle est frequente dans les experiences a long terme utilisees dans la recherche en gestion des ressources, en ecologie ou encore en sante. La complexite numerique additionnelle decoulant de la censure par intervalle des donnees de durees de vie rend prohibitive l'inference dans les modeles conjoints. Les auteurs etudient les proprietes de trois methodes d'imputation pour les durees de vie qui permettent l'application des techniques de modelisation conjointe devenues maintenant standards. Les techniques d'imputation incluent les methodes du point milieu, du lissage par noyau et une methode de projection en arriere qui incorpore l'information sur la trajectoire longitudinale. Une analyse conjointe d'une experience forestiere planifiee a long terme est presentee et elle est accompagnee d'une etude de simulation portant sur les proprietes des trois techniques d'imputation des durees de vie. La revue canadienne de statistique 39:438–457;2011 © 2011 Societe statistique du Canada
[1]
J. Ibrahim,et al.
A Bayesian semiparametric joint hierarchical model for longitudinal and survival data.
,
2003,
Biometrics.
[2]
S. Titus,et al.
A generalized logistic model of individual tree mortality for aspen, white spruce, and lodgepole pine in Alberta mixedwood forests
,
2001
.
[3]
T. Duchesne,et al.
Local likelihood density estimation for interval censored data
,
2005
.
[4]
James O. Ramsay,et al.
Nonparametric density estimation for censored survival data: Regression‐spline approach
,
1992
.
[5]
D. Zeng,et al.
Asymptotic results for maximum likelihood estimators in joint analysis of repeated measurements and survival time
,
2005,
math/0602240.
[6]
Geert Verbeke,et al.
Fully exponential Laplace approximations for the joint modelling of survival and longitudinal data
,
2009
.
[7]
Raymond J. Carroll,et al.
Estimation and comparison of changes in the presence of informative right censoring by modeling the censoring process
,
1988
.
[8]
S. Zeger,et al.
Joint analysis of longitudinal data comprising repeated measures and times to events
,
2001
.
[9]
M. Wulfsohn,et al.
A joint model for survival and longitudinal data measured with error.
,
1997,
Biometrics.
[10]
M. C. Jones,et al.
A reliable data-based bandwidth selection method for kernel density estimation
,
1991
.
[11]
Xihong Lin,et al.
Semiparametric Modeling of Longitudinal Measurements and Time‐to‐Event Data–A Two‐Stage Regression Calibration Approach
,
2008,
Biometrics.
[12]
Bradley P Carlin,et al.
Separate and Joint Modeling of Longitudinal and Event Time Data Using Standard Computer Packages
,
2004
.
[13]
D. Rubin,et al.
Maximum likelihood from incomplete data via the EM - algorithm plus discussions on the paper
,
1977
.
[14]
Probability Distributions as Models for Mortality
,
1985
.
[15]
R. Monserud,et al.
Modeling individual tree mortality for Austrian forest species
,
1999
.
[16]
Ning Li,et al.
A Joint Model for Longitudinal Measurements and Survival Data in the Presence of Multiple Failure Types
,
2008,
Biometrics.
[17]
Dimitris Rizopoulos,et al.
JM: An R package for the joint modelling of longitudinal and time-to-event data
,
2010
.
[18]
IMPUTATION BASED ON LOCAL LIKELIHOOD DENSITY ESTIMATION FOR INTERVAL CENSORED SURVIVAL DATA WITH APPLICATION TO TREE MORTALITY IN BRITISH COLUMBIA
,
2009
.
[19]
J. Ware,et al.
Random-effects models for longitudinal data.
,
1982,
Biometrics.
[20]
H. Temesgen,et al.
An Individual-Tree Mortality Model for Complex Stands of Southeastern British Columbia
,
2005
.
[21]
F. Hsieh,et al.
Joint Modeling of Survival and Longitudinal Data: Likelihood Approach Revisited
,
2006,
Biometrics.
[22]
Amanda G Chetwynd,et al.
Joint modelling of repeated measurements and time‐to‐event outcomes: The fourth Armitage lecture
,
2008,
Statistics in medicine.