Abstract
We consider joint selection of fixed and random effects in general mixed-effects models. The interpretation of estimated mixed-effects models is challenging since changing the structure of one set of effects can lead to different choices of important covariates in the model. We propose a stepwise selection algorithm to perform simultaneous selection of the fixed and random effects. It is based on Bayesian Information criteria whose penalties are adapted to mixed-effects models. The proposed procedure performs model selection in both linear and nonlinear models. It should be used in the low-dimension setting where the number of ovariates and the number of random effects are moderate with respect to the total number of observations. The performance of the algorithm is assessed via a simulation study, which includes also a comparative study with alternatives when available in the literature. The use of the method is illustrated in the clinical study of an antibiotic agent kinetics.
Acknowledgement
The application was based on the data from the Impact trial (PI O. Pajot, Sponsor APHP, Paris, France, NCT00950222). We thank C. Burdet, O. Pajot, C. Couffignal, L. Armand-Lefèvre, A. Foucrier, C. Laouénan, M. Wolff, L. Massias and F. Mentré for providing us the data. We are also grateful to the INRAE MIGALE bioinformatics platform (https://migale.inrae.fr) for providing computational resources. We thank the Editors and the two reviewers for their useful comments.
-
Author contribution: All the authors have accepted responsibility for the entire content of this submitted manuscript and approved submission.
-
Research funding: None declared.
-
Employment or leadership: None declared.
-
Honorarium: None declared.
-
Conflict of interest statement: The authors declare no conflicts of interest regarding this article.
References
1. Pinheiro J, Bates D. Mixed-effects models in S and S-PLUS. 1st ed. 2nd printing Springer-Verlag New York Inc; 2009.Search in Google Scholar
2. Lavielle M. Mixed effects models for the population approach. Models, tasks, methods and tools. Chapman and Hall/CRC; 2014.10.1201/b17203Search in Google Scholar
3. Burdet C, Pajot O, Couffignal C, Armand-Lefevre L., Foucrier A, Laouénan C, et al. Population pharmacokinetics of single-dose amikacin in critically ill patients with suspected ventilator-associated pneumonia. Eur J Clin Pharmacol 2015;71:75–83. https://doi.org/10.1007/s00228-014-1766-y.Search in Google Scholar PubMed
4. Bondell H, Krishna A, Ghosh S. Joint variable selection for fixed and random effects in linear mixed-effects models. Biometrics 2010;66:1069–1077. https://doi.org/10.1111/j.1541-0420.2010.01391.x.Search in Google Scholar PubMed PubMed Central
5. Fan J, Li R. Variable selection in linear mixed effects models. The Ann Stat 2012;40:2043–2068. https://doi.org/10.1214/12-aos1028.Search in Google Scholar PubMed PubMed Central
6. Tibshirani R. Regression shrinkage and selection via the lasso. J.Roy Stat Soc B 1996;58:267–288. https://doi.org/10.1111/j.2517-6161.1996.tb02080.x.Search in Google Scholar
7. Pan J, Huang C. Random effects selection in generalized linear mixed models via shrinkage penalty function. Stat Comput 2014;24:725–738. https://doi.org/10.1007/s11222-013-9398-0.Search in Google Scholar
8. Chen Y, Fei Y, Pan J. Statistical inference in generalized linear mixed models by joint modelling mean and covariance of non-normal random effects. Open J Stat 2015;5:568–584. https://doi.org/10.4236/ojs.2015.56059.Search in Google Scholar
9. Schelldorfer J, Meier L, Bühlman P. Glmmlasso: An algorithm for high-dimensional generalized linear mixed models using l1-penalization. J Comput Graph Stat 2014;23:460–477. https://doi.org/10.1080/10618600.2013.773239.Search in Google Scholar
10. Hui F, Muller S, Welsh A. Joint selecytion in mixed models using regularized pql. J Am Stat Assoc 2017;112:1323–1333. https://doi.org/10.1080/01621459.2016.1215989.Search in Google Scholar
11. Kuznetsova A, Brockhoff PB, Christensen RHB. lmertest package : Tests in linear mixed effects models. J Stat Software 2017;82. https://doi.org/10.18637/jss.v082.i13.Search in Google Scholar
12. Baey C, Cournede PH, Kuhn E. Asymptotic distribution of likelihood ratio test statistics for variance components in nonlinear mixed effects models. Comput Stat Data Anal 2019;135:107–122. https://doi.org/10.1016/j.csda.2019.01.014.Search in Google Scholar
13. Bertrand J, Balding DJ. Multiple single nucleotide polymorphism analysis using penalized regression in nonlinear mixed-effect pharmacokinetik models. Pharmacogenetics Genom 2013;23:167–174. https://doi.org/10.1097/fpc.0b013e32835dd22c.Search in Google Scholar PubMed
14. Delattre M, Lavielle M, Poursat M-A. A note on bic in mixed effects models. EJS 2014;8:456–475. https://doi.org/10.1214/14-ejs890.Search in Google Scholar
15. Nie L. Convergence rate of the MLE in generalized linear and nonlinear mixed-effects models: Theory and applications. J Stat Plan Inference 2007;137:1787–1804. https://doi.org/10.1016/j.jspi.2005.06.010.Search in Google Scholar
16. Lebarbier E, Mary-Huard T. Une introduction au critère bic: fondements théoriques et interprétation. J de la SFdS 2006;147:39–57. http://www.numdam.org/item?id=JSFS_2006__147_1_39_0.Search in Google Scholar
17. Lavielle M, Mentre F. Estimation of population pharmacokinetic parameters of saquinavir in hiv patients with the monolix software. J Pharmacokinet Pharmacodyn 2007;34:229–249. https://doi.org/10.1007/s10928-006-9043-z.Search in Google Scholar PubMed PubMed Central
18. Delattre M, Lavielle M. Maximum likelihood estimation in discrete mixed hidden Markov models using the saem algorithm. Comput Stat Data Anal 2012;56:2073–2085. https://doi.org/10.1016/j.csda.2011.12.017.Search in Google Scholar
© 2020 Walter de Gruyter GmbH, Berlin/Boston
