Computational Statistics

, Volume 34, Issue 1, pp 395–414 | Cite as

Fusion learning algorithm to combine partially heterogeneous Cox models

  • Lu TangEmail author
  • Ling Zhou
  • Peter X. K. Song
Original Paper


We propose a fusion learning procedure to perform regression coefficients clustering in the Cox proportional hazards model when parameters are partially heterogeneous across certain predefined subgroups, such as age groups. One major issue pertains to the fact that the same covariate may have different influence on the survival time across different subgroups. Learning differences in covariate effects is of critical importance to understand the model heterogeneity resulted from the between-group heterogeneity, especially when the number of subgroups is large. We establish a computationally efficient procedure to learn the heterogeneous patterns of regression coefficients across the subgroups in Cox proportional hazards model. Utilizing a fusion learning algorithm coupled with the estimated parameter ordering, the proposed method mitigates greatly computational burden with little loss of statistical power. Extensive simulation studies are conducted to evaluate the performance of our method. Finally with a comparison to some popular conventional methods, we illustrate the proposed method by a vehicle leasing contract renewal analysis.


Fused lasso Regression coefficient clustering Extended BIC Cox proportional hazards model 



We are grateful to three anonymous reviewers for their valuable comments that have led to an improvement of this paper. This research is partially supported by the National Science Foundation DMS 1513595 and the National Institutes of Health R01 ES024732.


  1. Aitkin M (1996) A general maximum likelihood analysis of overdispersion in generalized linear models. Stat Comput 6(3):251–262MathSciNetCrossRefGoogle Scholar
  2. Andersen PK, Borgan O, Gill RD, Keiding N (2012) Statistical models based on counting processes. Springer, BerlinzbMATHGoogle Scholar
  3. Azzimonti L, Ieva F, Paganoni AM (2013) Nonlinear nonparametric mixed-effects models for unsupervised classification. Comput Stat 28(4):1549–1570MathSciNetCrossRefzbMATHGoogle Scholar
  4. Bonetti M, Gelber RD et al (2000) A graphical method to assess treatment–covariate interactions using the Cox model on subsets of the data. Stat Med 19(19):2595–2609CrossRefGoogle Scholar
  5. Chen J, Chen Z (2008) Extended bayesian information criteria for model selection with large model spaces. Biometrika 95(3):759–771MathSciNetCrossRefzbMATHGoogle Scholar
  6. Cheng X, Lu W, Liu M (2015) Identification of homogeneous and heterogeneous variables in pooled cohort studies. Biometrics 71(2):397–403MathSciNetCrossRefzbMATHGoogle Scholar
  7. Cox DR (1972) Regression models and life tables (with discussion). J R Stat Soc Ser B (Stat Methodol) 34(2):187–220zbMATHGoogle Scholar
  8. Efron B (1993) Bayes and likelihood calculations from confidence intervals. Biometrika 80(1):3–26MathSciNetCrossRefzbMATHGoogle Scholar
  9. Egger M, Smith GD, Schneider M, Minder C (1997) Bias in meta-analysis detected by a simple, graphical test. BMJ Br Med J 315(7109):629–634CrossRefGoogle Scholar
  10. Fisher RA (1956) Statistical methods and scientific inference. Oliver and Boyd, EdinburghzbMATHGoogle Scholar
  11. Friedman J, Hastie T, Höfling H, Tibshirani R (2007) Pathwise coordinate optimization. Ann Appl Stat 1(2):302–332MathSciNetCrossRefzbMATHGoogle Scholar
  12. Friedman J, Hastie T, Tibshirani R (2010) Regularization paths for generalized linear models via coordinate descent. J Stat Softw 33(1):1CrossRefGoogle Scholar
  13. Gao X, Song PXK (2010) Composite likelihood bayesian information criteria for model selection in high-dimensional data. J Am Stat Assoc 105(492):1531–1540MathSciNetCrossRefzbMATHGoogle Scholar
  14. Gasperoni F, Ieva F, Paganoni A, Jackson C, Sharples L (2017) Nonparametric shared frailty model for classification of survival data. In: Proceedings of the conference statistics and data science: new challenges, new generations, pp 451–456Google Scholar
  15. Hakulinen T, Tenkanen L (1987) Regression analysis of relative survival rates. J R Stat Soc Ser C (Appl Stat) 36(3):309–317Google Scholar
  16. Higgins J, Thompson SG (2002) Quantifying heterogeneity in a meta-analysis. Stat Med 21(11):1539–1558CrossRefGoogle Scholar
  17. Klein JP (1992) Semiparametric estimation of random effects using the cox model based on the EM algorithm. Biometrics 48(3):795–806CrossRefGoogle Scholar
  18. Ma S, Huang J (2017) A concave pairwise fusion approach to subgroup analysis. J Am Stat Assoc 112(517):410–423MathSciNetCrossRefGoogle Scholar
  19. Nielsen GG, Gill RD, Andersen PK, Sørensen TI (1992) A counting process approach to maximum likelihood estimation in frailty models. Scand J Stat 19(1):25–43MathSciNetzbMATHGoogle Scholar
  20. Tang L, Song PXK (2016) Fused lasso approach in regression coefficients clustering–learning parameter heterogeneity in data integration. J Mach Learn Res 17(113):1–23MathSciNetzbMATHGoogle Scholar
  21. Tibshirani R, Saunders M, Rosset S, Zhu J, Knight K (2005) Sparsity and smoothness via the fused lasso. J R Stat Soc Ser B (Stat Methodol) 67(1):91–108MathSciNetCrossRefzbMATHGoogle Scholar
  22. Wang F, Wang L, Song PXK (2016) Fused lasso with the adaptation of parameter ordering in combining multiple studies with repeated measurements. Biometrics 72(4):1184–1193MathSciNetCrossRefzbMATHGoogle Scholar
  23. Zou H (2006) The adaptive lasso and its oracle properties. J Am Stat Assoc 101(476):1418–1429MathSciNetCrossRefzbMATHGoogle Scholar

Copyright information

© Springer-Verlag GmbH Germany, part of Springer Nature 2018

Authors and Affiliations

  1. 1.Department of BiostatisticsUniversity of MichiganAnn ArborUSA

Personalised recommendations