Regression trees for interval-censored failure time data based on censoring unbiased transformations and pseudo-observations

被引:0
作者
Yang, Ce [1 ]
Li, Xianwei [2 ]
Diao, Liqun [2 ]
Cook, Richard J. [2 ]
机构
[1] Vertex Pharmaceut, Boston, MA USA
[2] Univ Waterloo, Dept Stat & Actuarial Sci, Waterloo, ON, Canada
来源
CANADIAN JOURNAL OF STATISTICS-REVUE CANADIENNE DE STATISTIQUE | 2024年 / 52卷 / 04期
基金
加拿大健康研究院;
关键词
Censoring unbiased transformations; interval censoring; prediction; pseudo-observations; regression trees; variable selection; SURVIVAL TREES; RISK-FACTORS; ARTHRITIS; ALGORITHM; CANCER;
D O I
10.1002/cjs.11807
中图分类号
O21 [概率论与数理统计]; C8 [统计学];
学科分类号
020208 ; 070103 ; 0714 ;
摘要
Interval-censored data arise when a failure process is under intermittent observation and failure status is only known at assessment times. We consider the development of predictive algorithms when training samples involve interval censoring. Using censoring unbiased transformations and pseudo-observations, we define observed data loss functions, which are unbiased estimates of the corresponding complete data loss functions. We show that regression trees based on these loss functions can recover the tree structure and yield good predictive accuracy. An application is given to a study involving individuals with psoriatic arthritis where the aim is to identify genetic markers useful for the prediction of axial disease within 10 years of a baseline assessment. Les donn & eacute;es censur & eacute;es par intervalle surviennent lorsqu'un processus de d & eacute;faillance est observ & eacute; de mani & egrave;re intermittente et que l'& eacute;tat de d & eacute;faillance n'est connu qu'aux temps d'& eacute;valuation. Les auteurs de ce travail & eacute;tudient le d & eacute;veloppement d'algorithmes pr & eacute;dictifs lorsque les & eacute;chantillons d'entra & icirc;nement impliquent une censure par intervalle. En utilisant des transformations sans biais de censure et des pseudo-observations, ils d & eacute;finissent des fonctions de perte de donn & eacute;es observ & eacute;es, qui sont des estimations sans biais des fonctions de perte des donn & eacute;es compl & egrave;tes correspondantes. Ils d & eacute;montrent que les arbres de r & eacute;gression construits & agrave; partir de ces fonctions de substitution permettent de retrouver la structure r & eacute;elle sous-jacente tout en assurant de bonnes performances pr & eacute;dictives. Cette m & eacute;thodologie est appliqu & eacute;e & agrave; une & eacute;tude clinique sur l'arthrite psoriasique, visant & agrave; identifier des marqueurs g & eacute;n & eacute;tiques pr & eacute;dictifs de l'apparition d'une forme axiale de la pathologie dans les 10 ans suivant une & eacute;valuation initiale.
引用
收藏
页数:21
相关论文
共 42 条
[1]   ANALYSIS OF MULTITYPE RECURRENT EVENTS IN LONGITUDINAL-STUDIES - APPLICATION TO A SKIN-CANCER PREVENTION TRIAL [J].
ABULIBDEH, H ;
TURNBULL, BW ;
CLARK, LC .
BIOMETRICS, 1990, 46 (04) :1017-1034
[2]   Pseudo-observations in survival analysis [J].
Andersen, Per Kragh ;
Perme, Maja Pohar .
STATISTICAL METHODS IN MEDICAL RESEARCH, 2010, 19 (01) :71-99
[3]   Regression analysis of restricted mean survival time based on pseudo-observations [J].
Andersen, PK ;
Hansen, MG ;
Klein, JP .
LIFETIME DATA ANALYSIS, 2004, 10 (04) :335-350
[4]   Generalised linear models for correlated pseudo-observations, with applications to multi-state models [J].
Andersen, PK ;
Klein, JP ;
Rosthoj, S .
BIOMETRIKA, 2003, 90 (01) :15-27
[5]   An Efficient Implementation of the EMICM Algorithm for the Interval Censored NPMLE [J].
Anderson-Bergman, Clifford .
JOURNAL OF COMPUTATIONAL AND GRAPHICAL STATISTICS, 2017, 26 (02) :463-467
[6]  
Andriyashin A., 2005, FINANCIAL APPL CLASS
[7]   Presenting the Regression Tree Method and its application in a large-scale educational dataset [J].
Assis Gomes, Cristiano Mauro ;
Jelihovschi, Enio .
INTERNATIONAL JOURNAL OF RESEARCH & METHOD IN EDUCATION, 2020, 43 (02) :201-221
[8]  
Banerjee A, 2005, J MACH LEARN RES, V6, P1705
[9]   A General and Adaptive Robust Loss Function [J].
Barron, Jonathan T. .
2019 IEEE/CVF CONFERENCE ON COMPUTER VISION AND PATTERN RECOGNITION (CVPR 2019), 2019, :4326-4334
[10]  
Bogaerts K., 2017, Survival analysis with interval-censored data: A practical approach with examples in R, SAS, and BUGS