			Home About us Contact

Survival Trees (survival + tree)

Distribution by Scientific Domains

Mathematics and Statistics	50%

Selected Abstracts

Multivariate Survival Trees: A Maximum Likelihood Approach Based on Frailty Models

BIOMETRICS, Issue 1 2004
Xiaogang Su
Summary. A method of constructing trees for correlated failure times is put forward. It adopts the backfitting idea of classification and regression trees (CART) (Breiman et al., 1984, in Classification and Regression Trees). The tree method is developed based on the maximized likelihoods associated with the gamma frailty model and standard likelihood-related techniques are incorporated. The proposed method is assessed through simulations conducted under a variety of model configurations and illustrated using the chronic granulomatous disease (CGD) study data. [source]

Identification of MRI and 1H MRSI parameters that may predict survival for patients with malignant gliomas

NMR IN BIOMEDICINE, Issue 1 2004
Xiaojuan Li
Abstract Although MR imaging (MRI) and MR spectroscopic imaging (MRSI) have been applied in the diagnosis and treatment planning for brain tumors, their prognostic significance has not yet been determined. The goal of this study was to identify pre-treatment MRI and MRSI parameters for patients with malignant glioma that may be useful in predicting survival. Two populations of patients with newly-diagnosed malignant glioma were examined with MRI and three-dimensional proton (1H) MRSI. Thirty-nine patients (22 grade 3 and 17 glioblastoma multiforme, GBM) were studied prior to surgery, and 33 GBM patients were studied after surgery but prior to treatment with radiation and chemotherapy. Signal intensities of choline (Cho), creatine (Cr), N -acetyl aspartate (NAA), and lactate/lipid (LL) were estimated from the spectra. Recursive partitioning methods were applied to parameters that included age, histological grade, MRI and MRSI variables to generate survival trees. Patients were grouped into high and low risk categories and the corresponding Kaplan,Meier curves were plotted for comparison between groups. The parameters that were selected by recursive partitioning as being predictive of poor outcome were older age, larger contrast enhancement, higher Cho-to-Cr, higher Cho-to-NAA, higher LL and lower Cr-to-NAA abnormalities. The survival functions were significantly different between the sub-groups of patients obtained from the survival tree for both pre-surgery and post-surgery data. The results of this study suggest that pre-treatment MRI and three-dimensional 1H-MRSI provide information that predicts outcome for patients with malignant gliomas and have drawn attention to variables that should be examined prospectively in future studies using these techniques. Copyright © 2004 John Wiley & Sons, Ltd. [source]

Discrete-time survival trees

THE CANADIAN JOURNAL OF STATISTICS, Issue 1 2009
Imad Bou-hamad
MSC 2000: Primary 62N99; secondary 62G08 Abstract Tree-based methods are frequently used in studies with censored survival time. Their structure and ease of interpretability make them useful to identify prognostic factors and to predict conditional survival probabilities given an individual's covariates. The existing methods are tailor-made to deal with a survival time variable that is measured continuously. However, survival variables measured on a discrete scale are often encountered in practice. The authors propose a new tree construction method specifically adapted to such discrete-time survival variables. The splitting procedure can be seen as an extension, to the case of right-censored data, of the entropy criterion for a categorical outcome. The selection of the final tree is made through a pruning algorithm combined with a bootstrap correction. The authors also present a simple way of potentially improving the predictive performance of a single tree through bagging. A simulation study shows that single trees and bagged-trees perform well compared to a parametric model. A real data example investigating the usefulness of personality dimensions in predicting early onset of cigarette smoking is presented. The Canadian Journal of Statistics 37: 17-32; 2009 © 2009 Statistical Society of Canada Arbres de survie à temps discret Les méthodes d'arbres sont fréquemment utilisées lors d'études impliquant des données censurées. La structure d'un arbre ainsi que la facilité avec laquelle il peut être interprété font de lui un outil utile afin d'identifier des facteurs de pronostique et de prédire les probabilités de survie conditionnelles d'un individu étant donné ses covariables. Les méthodes existantes ont été développées pour traiter une variable temporelle continue. En pratique, il arrive fréquemment que la variable mesurant le temps de survie soit mesurée selon une échelle discrète. Les auteurs proposent une nouvelle méthode pour construire un arbre qui est spécialement adaptée aux variables de survie à temps discret. Le critère de division peut être vu comme étant une extension, au cas de censure à droite, du critère d'entropie pour une variable catégorielle. La sélection de l'arbre final est basée sur une méthode d'élagage combinée avec une correction bootstrap. Les auteurs présentent également une méthode simple pour améliorer, potentiellement, la performance d'un seul arbre avec le bagging. Une étude par simulation montre que des arbres seuls et des arbres "baggés" performent bien comparativement à un modèle paramétrique. Les auteurs présentent aussi une illustration de la nouvelle méthode avec des vraies données qui investiguent l'utilité d'utiliser des dimensions de la personnalité afin de prévoir le début de l'utilisation de la cigarette. La revue canadienne de statistique 37: 17-32; 2009 © 2009 Société statistique du Canada [source]