KDE-GR: A Density-Aware Graph-Guided Resampling Framework for Imbalanced Brain Stroke Prediction — Source Code

Complete source code for the paper "KDE-GR: A Density-AwareGraph-Guided Resampling Framework for Imbalanced Brain StrokePrediction" (Ibrahim & El-Ghaish, Scientific Reports, 2026). Includes: the kdegr Python package implementing the KDE-GRalgorithm, RF-ET-MLP stacking ensemble, all 10 resamplingcomparison methods, nested cross-validation pipeline withfold-level checkpointing, ablation study, sensitivity analyses(bandwidth, threshold), feature importance, probabilitycalibration, decision-curve analysis, and figure generationscripts. Evaluated on BRFSS 2015 (CDC) and NHANES 2021-2023. Requirements: Python 3.9+, scikit-learn 1.3.0,imbalanced-learn 0.11.0, LightGBM 4.0.0. See README.mdinside the zip for full installation and usage instructions.