From Machine Learning to the Bedside: An Explainable and Clinically Interpretable Mortality Prediction Framework in Aluminum Phosphide Poisoning

Résumé fourni par la source

# Explainable mortality-prediction pipeline for acute aluminium phosphide poisoning Reproducible analysis code accompanying the manuscript*"From Machine Learning to the Bedside: An Explainable and Clinically InterpretableMortality Prediction Framework in Aluminum Phosphide Poisoning."* ## Contents- `alp_reproducible_pipeline.py` — the complete, self-contained analysis script.- `requirements_zenodo.txt` — pinned dependencies. ## DataThe script expects a single CSV of admission variables and the in-hospitalmortality outcome (`Overal_outcome`, 1 = death, 0 = survival). Column roles can beset in the `Config` block at the top of the script. ## How to run```pip install -r requirements_zenodo.txtpython alp_reproducible_pipeline.py path/to/alppreimputation.csv```All tables and figures are written to the `alp_outputs/` directory. ## What it doesStratified train/hold-out split; training-only random-forest imputation;two-stage logistic-regression feature selection within the training set; tuningand internal-hold-out evaluation of five classifiers (Random Forest, XGBoost,Logistic Regression, SVM, ANN) with bootstrap 95% confidence intervals; SHAPexplainability; calibration; decision-curve analysis; probability-based riskstratification; and a simplified integer bedside score with nomogram. A fixedrandom seed (42) is used throughout. ## LicenseSee the `LICENSE` file (e.g. MIT or CC BY 4.0).

Ce résumé expose les affirmations des auteurs. BNTIC ne l’interprète pas comme une validation indépendante des résultats.