Representative R Code for: Development and External Validation of a Machine Learning Model for Predicting Liver Injury in Children With Mycoplasma Pneumoniae Pneumonia

Le résumé fourni par la source

This repository contains a representative R implementation of the analyticalpipeline described in the Methods section of the associated manuscript"Development and External Validation of a Machine Learning Model forPredicting Liver Injury in Children With Mycoplasma Pneumoniae Pneumonia"(Yu, Jing, Tang, Liu, Wei; Scientific Reports, in press). The pipeline covers: data cleaning, correlation analysis, Boruta featureselection, data preprocessing (Yeo-Johnson transformation, Z-scorestandardisation, SMOTE oversampling), training and evaluation of tensupervised machine-learning algorithms (catboost, gbm, kknn, lightgbm,naive_bayes, nnet, ranger, rpart, svm, xgboost), SHAP-based interpretabilityanalysis, and an interactive Shiny web calculator. IMPORTANT: The original model development in this study was performed usingthe Free Statistics analysis platform (v2.4), a graphical point-and-clickinterface built on R that does not export its underlying scripts. Thisrepository was therefore written afterwards, directly from the manuscript'sMethods section, to document and illustrate the analytical workflow formethodological transparency. It is not the literal code that produced thepublished results, and running it is not expected to exactly reproduce thepublished tables/figures. See README.md for full details.

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