fast_pseudo_calculation.R: Function for calculating pseudo-observations, which is faster than Rpackage pseudo.PSW_pseudo.R: Function for IPW-PO and OW-PO.cox_model.R: Function for estimator Cox and Cox-IPW.pseudo_G.R: Function for PO-UNADJ and PO-G.Mao_Method_func.R: Function for estimators in Mao's paper.AIPW_pseudo.R: Function for AIPW and AOW.
simu_main.R: Main script for running simulations.simu_utils.R: Utility function for simulations.simu_data_gen.R: Utility function for generating simulated data.simu_exe.sh: Bash script to run simulations in all settings.
data_preprocessing.R: Cleaning for data application.data_application.R: Analyze function for data application.
To run the simulation in the paper, you can run following the command and set the simulation as the working directory.
git clone https://github.com/zengshx777/OW\Survival_CodeBase
R CMD BATCH --vanilla '--args dependent.censoring=F multi.arm=T prop.hazard=F good_overlap=1 sample_size=150' simu_main.R R1.out
where dependent.censoring controls whether the censoring is independent of the covariates; multi.arm controls the number of arms in the data (T for J=3, F for J=2); prop.hazard controls whether the proportional hazard assumption is correct; good_overlap control the overlap conditions (1 for RCT, 2 for good overlap, 3 for poor overlap); sample_size control the sample size. One simple way to run many simulations in different settings in parallel is to run the simu_exe.sh directly (you can customize the scenario in this file). The current simu_main.R will run all estimators mentioned in the paper by default, which might be time-consuming. You can comment out certain estimators to speed up.
The results will be saved in the folder simulation_results To output the similar Figures and Tables in the paper, please refer to the scripts in folder output_utils
The NCDB data used in the case study is publicly available upon approval of the NCDB Participant User File application.