2026-10-15T17:21:52Z https://www.scidb.cn/oai

10.57760/sciencedb.30932 2025-11-07T18:29:20Z

2025-11-07 Construct a prognostic mode doi:10.57760/sciencedb.30932 en In this study, 150 host genes associated with the Firmicutes phylum were selected to construct a prognostic model. Univariate Cox regression analysis (with a significance level of P< 0.05) was employed to identify Firmicutes-related host genes significantly correlated with overall survival . A total of 516 curated lung adenocarcinoma (LUAD) tumor samples were divided into a training set (LUAD-train) and a validation set (LUAD-test) using a 7:3 ratio via stratified random sampling, implemented with the machine learning caretpackage in R . The prognostic model was subsequently developed using the integrated machine learning toolkit Mime1, which incorporates ten distinct algorithms: "RSF" (Random Survival Forest), "Enet" (Elastic Net), "StepCox" (Stepwise Cox Regression), "CoxBoost", "PlsRcox" (Partial Least Squares Regression for Cox Models), "Superpc" (Supervised Principal Components), "GBM" (Gradient Boosting Machine), "Survivalsvm" (Support Vector Machines for Survival Analysis), "Ridge" (Ridge Regression), and "Lasso" (Lasso Regression).The predictive performance of the prognostic model was evaluated using the consistency index (C-index)and the area under the receiver operating characteristic curve (AUC). Based on these evaluations, the model combining StepCox[forward] and GBMwas ultimately selected as the final prognostic model. LUAD; prognostic model; Machine Learning Algorithm MatsuJun GUO YUQI PUBLIC https://creativecommons.org/licenses/by/4.0/ dataset Science Data Bank