This code was developed using R version 3.4.1 and runs exclusively on the datasets from the SPRINT trial (NEJM 2015)
This data can be requested through BioLINCC: https://biolincc.nhlbi.nih.gov/studies/sprint_pop/
Needed libraries and their versions (more recent versions will probably work as well): ggplot2_2.2.1,sas7bdat_0.5,survival_2.41-3 impute_1.50.1,PRROC_1.3,rms_5.1-2
This code assumes all datasets and scripts are in the current working directory (i.e. under "./"). Namely the needed dataset files are:
baseline.csv
outcomes.csv
safety.csv
bp.csv
bloodpressure.sas7bdat
accord_key.sas7bdat
lipids.sas7bdat
otherlabs.sas7bdat
concomitantmeds.sas7bdat
cvdoutcomes.sas7bdat
f07_baselinehistoryphysicalexam.sas7bdat
microvascularoutcomes.sas7bdat
in order to reproduce the results correctly please execute the code as follows:
1) run the source() function for the following scripts: 
generate_sprint_raw_data.R
fix_framingham_score.R
generate_accord_raw_data.R
extract_dynamics_SPRINT.R
extract_dynamics_ACCORD.R
predictors.R
run_predictor.R
treatment_reccomendation.R
(or run the first lines of "analyze_sprint.R")
2) reproduce the results on the SPRINT database (including the treatment assignment simulation) by running "analyze_sprint.R" top to bottom
3) reproduce the results on the ACCORD database by running "analyze_accord.R" top to bottom (note that running "analyze_sprint.R" is obligatory before this step)