A Study on Penalized Variable Selection in Accelerated Failure Time Models with Random Effects
- Alternative Title
- 변량효과를 갖는 가속화 실패 시간모형에서 벌점화 변수선택 방법에 대한 연구
- Abstract
- Accelerated failure time (AFT) model is a linear model under the log-transformation of survival time and has been introduced as a useful alternative to proportional hazards (PH) model. In this thesis we propose variable-selection procedures of fixed effects in the AFT model using penalized likelihood approaches. We use two popular penalty functions, LASSO (least absolute shirinkage and selection operator; Tibshirani, 1996) and SCAD (smoothly clipped absolute deviation; Fan and Li, 2001). With these procedures we can select important variables and estimate the fixed effects at once. The performance of the proposed method is evaluated using simulation studies under univariate survival data. The proposed method is also illustrated with a primary biliary cirrhosis (PBC) data set which is well known in the literature.
Furthermore, we extend the proposed method to AFT models allowing for random effects for analyzing multivariate survival data. For this purpose, we use penalized h-likelihood approach (Ha et al., 2014, 2017). We demonstrate the usefulness of the extended method through simulation studies and practical data such as kidney infection data and bladder-cancer recurrence data from multi-center clinical trial.
가속화 실패 시간(Accelerated failure time; AFT) 모형은 로그생존시간에 대한 선형모형이며 비례위험 모형의 유용한 대안이다. 본 연구에서는 벌점화 가능도 접근법을 사용하여 AFT 모형의 고정효과에 대한 변수선택 절차를 개발한다. 여기서 벌점화 함수에 대해 LASSO(Least absolute shirinkage and selection operator; Tibshirani, 1996)와 SCAD (Smoothly clipped absolute deviation; Fan and Li, 2001)를 사용하였다. 특히 이러한 절차는 중요한 변수선택과 고정효과의 추정을 동시에 하는 장점이 있다. 먼저 일변량 생존자료하에서 모의실험을 통해 제안된 방법의 타당성을 평가하고, 그 예증을 위해 문헌에서 잘 알려진 담즙성 간경화증 (Primary biliary cirrhosis; PBC) 자료를 사용한다.
나아가 다변량 생존자료를 분석하기 위해, 제안된 방법을 변량효과를 허용하는 AFT 모형으로 확장한다. 그 추론을 위해, 벌점화 다단계 가능도 접근법을 사용한다 (Ha et al., 2014, 2017). 모의실험 및 다기관 임상시험 자료와 신장감염 자료를 이용하여 확장된 방법의 타당성을 입증한다.
- Author(s)
- 박은영
- Issued Date
- 2018
- Awarded Date
- 2018.2
- Type
- Dissertation
- Keyword
- AFT model H-likelihood Penalized likelihood Random effects Variable selection
- Publisher
- 부경대학교
- URI
- https://repository.pknu.ac.kr:8443/handle/2021.oak/13874
http://pknu.dcollection.net/common/orgView/200000010714
- Affiliation
- 부경대학교 대학원
- Department
- 대학원 통계학과
- Advisor
- 하일도
- Table Of Contents
- Abstract v
Contents i
1. INTRODUCTION 1
2. PENALIZED VARIABLE SELECTION FOR AFT MODELS 4
2.1 AFT Model 4
2.2 Variable Selection Procedure 6
3. PENALIZEDVARIABLESELECTIONFORAFTRANDOMEFFECT MODELS 11
3.1 AFT Random-Effect Models 11
3.2 Extension of Variable Selection 13
4. SIMULATIONS 17
4.1 AFT Model without Random Effects 17
4.2 AFT Model with Random Effects 19
5. PRACTICAL EXAMPLES 21
5.1 Primary Biliary Cirrhosis Data 21
5.2 Kidney Infection Data 25
5.3 Multi-center Bladder Cancer Data 27
6. CONCLUSIONS 30
References 31
Appendix A Derivations 34
Appendix B R Codes 41
- Degree
- Master
-
Appears in Collections:
- 대학원 > 통계학과
- Authorize & License
-
- Files in This Item:
-
Items in Repository are protected by copyright, with all rights reserved, unless otherwise indicated.