County-Level Opioid Cost Modeling (PBM + SDOH): Data and Code for Replication

This repository contains all data, code, and replication materials for the study: “Predicting and Simulating County-Level Opioid Cost Burden Using Integrated PBM and SDOH Determinants”. Included files: panel_out_with_predictions.csv — county-level baseline & scenario predictions feature_engineered_panel.csv — processed Medicare Part D & SDOH predictors policy_simulation_results.csv — national scenario outputs 08_modelsimulation_v3.ipynb — end-to-end code for prediction, simulation, and SHAP analysis final_model.pkl — trained Random Forest model final_feature_list.json — model features used at inference These materials support full replication of the machine-learning pipeline, scenario evaluation, and figure generation.All data sources are de-identified aggregates derived from publicly available Medicare Part D and County Health Rankings datasets.