Doubly robust survival trees

Jon Arni Steingrimsson; Liqun Diao; Annette M Molinaro; Robert L Strawderman

doi:10.1002/sim.6949

Doubly robust survival trees

Stat Med. 2016 Sep 10;35(20):3595-612. doi: 10.1002/sim.6949. Epub 2016 Mar 31.

Authors

Jon Arni Steingrimsson¹, Liqun Diao², Annette M Molinaro³, Robert L Strawderman⁴

Affiliations

¹ Department of Biostatistics, Johns Hopkins University, Baltimore, 14853, MD, 21205 U.S.A.
² Department of Statistics and Actuarial Science, University of Waterloo, Waterloo, N2L 3G1, ON, CANADA.
³ Department of Neurological Surgery, University of California, San Francisco, 94143-0372, CA, U.S.A.
⁴ Department of Biostatistics and Computational Biology, University of Rochester, Rochester, 14642, NY, U.S.A.

Abstract

Estimating a patient's mortality risk is important in making treatment decisions. Survival trees are a useful tool and employ recursive partitioning to separate patients into different risk groups. Existing 'loss based' recursive partitioning procedures that would be used in the absence of censoring have previously been extended to the setting of right censored outcomes using inverse probability censoring weighted estimators of loss functions. In this paper, we propose new 'doubly robust' extensions of these loss estimators motivated by semiparametric efficiency theory for missing data that better utilize available data. Simulations and a data analysis demonstrate strong performance of the doubly robust survival trees compared with previously used methods. Copyright © 2016 John Wiley & Sons, Ltd.

Keywords: CART; censored data; inverse probability of censoring weighted estimation; loss estimation; regression trees; semiparametric estimation.

MeSH terms

Data Accuracy*
Humans
Models, Statistical*
Mortality
Risk Factors
Survival Analysis*

Grants and funding

R01 CA163687/CA/NCI NIH HHS/United States