Statistical Learning Methods for Optimizing Dynamic Treatment Regimes R package DTRlearn2 version 2.1 CRAN Task View: CausalInference (https://CRAN.R-project.org/view=CausalInference) We provide a comprehensive software to estimate general K-stage DTRs from SMARTs with Q-learning and a variety of outcome-weighted learning methods. Penalizations are allowed for variable selection and model regularization. With the outcome-weighted learning scheme, different loss functions - SVM hinge loss, SVM ramp loss, binomial deviance loss, and L2 loss - are adopted to solve the weighted classification problem at each stage; augmentation in the outcomes is allowed to improve efficiency. The estimated DTR can be easily applied to a new sample for individualized treatment recommendations or DTR evaluation. Software Depends: R(>= 2.10) Imports: kernlab, MASS, Matrix, foreach, glmnet, WeightSVM Yuan Chen <irene.yuan.chen@gmail.com> Comprehensive R Archive Network (CRAN) Yuan Chen [aut, cre], Ying Liu [aut], Tianchen Xu [ctb] (ORCID: <https://orcid.org/0000-0002-0102-7630>), Donglin Zeng [ctb], Yuanjia Wang [ctb] GPL-2 2026-09-10 application/tgz https://CRAN.R-project.org/package=DTRlearn2 doi:10.32614/CRAN.package.DTRlearn2