fig6

Figure 6. Machine learning construction of the prognostic model MOMLS. (A) Each model’s concordance index (C-index) derived from 101 machine-learning algorithms in the TCGA-LUAD and meta cohorts, ranked by the average C-index from the validation set; (B) The gene framework for the model constructed using the RSF algorithm. Each color represents each gene; (C) Univariate Cox regression analysis results for hub genes in the training and validation cohorts; (D and E) KM curves for the high and low MOMLS score groups in the meta-LUAD and meta cohorts. MOMLS: Multi-omics-driven machine learning signature; TCGA: The Cancer Genome Atlas; LUAD: landscape of lung adenocarcinoma; RSF: random survival forest; KM: Kaplan-Meier; Enet: elastic net; GBM: generalized boosted regression model; SuperPC: supervised principal components; plsRcox: partial least Cox.