Reproducibility package (code + synthetic dataset) for: Assessing mortality risk in pulmonary tuberculosis and severe malnutrition: development of the IIR marker via artificial intelligence

This Zenodo record (record: 18042276; DOI: 10.5281/zenodo.18042276) contains the code and a fully synthetic dataset sufficient to reproduce the analyses reported in the manuscript “Assessing mortality risk in pulmonary tuberculosis and severe malnutrition: development of the IIR marker via artificial intelligence” (submitted to Scientific Reports). Contents: Google Colab notebook reproducing model training and evaluation (Random Forest and Logistic Regression), LIME explanations, calibration, and an IIR-only validation (in-sample AUC, repeated CV out-of-fold AUC, bootstrap 95% CI). Synthetic dataset pro1.csv (fully synthetic; no real patient data). Outputs exported by the notebook (figures and CSV tables), including beta coefficients (univariate and multivariate) and cross-validation metrics. How to run (Colab): Upload the provided synthetic pro1.csv to Google Drive under: My Drive/pro1.csv Open and run the notebook from top to bottom. All outputs will be shown in the notebook. Notes: The dataset is fully synthetic and is provided only to enable reproducibility without sharing the original clinical dataset. The original clinical dataset is not included due to participant confidentiality and GDPR constraints. Library versions are printed at the start of the notebook for reproducibility. License: The code in this record is released under the MIT License (see the attached LICENSE file).