Reproducible Code and Data Repository: Design-Aware Predictive and Causal Modeling of Cardiovascular Risk in Chronic Kidney Disease Using Penalized and Double Machine Learning Approaches

Description

This repository contains the data and code necessary to reproduce the main results of the study. The paper proposes a unified framework integrating survey-weighted penalized prediction (LASSO), Double/Debiased Machine Learning (DML), Two-Stage Residual Inclusion (2SRI), and design-consistent bootstrap inference.

Repository Structure

/data
  ckd_survey_sim.csv
  simulation_metrics.csv
  metrics_all_sim.csv
  stability_sel_prob.csv

/code
  SimulacionGeneral.R
  Empirical Study.R

/docs
  README.docx

Requirements

R version 4.3 or higher with the following packages:

survey, glmnet, dplyr, tidyr, ggplot2, haven, purrr, gridExtra

Reproducibility Guide

Step 1 – Simulation:

source('code/SimulacionGeneral.R')

Step 2 – Empirical analysis:

source('code/Empirical Study.R')

Data Availability

Simulated datasets are included. The ENS 2016–2017 dataset is not redistributed and must be obtained from official sources (MINSAL).

Methodological Notes

Survey design uses the survey package. Penalized models use glmnet. DML uses cross-fitting at PSU level. 2SRI addresses endogeneity. Metrics include weighted AUC and Brier score.

Disclaimer

Simulated data are synthetic. ENS variable names may vary. Adjustments may be required.

Contact

Fernando Rojas
fernando.rojas@uv.cl

License

Creative Commons Attribution (CC BY)