Research Paper Volume 16, Issue 9 pp 8361—8377
Comprehensive bioinformatics analytics and in vivo validation reveal SLC31A1 as an emerging diagnostic biomarker for acute myocardial infarction
Figure 2. Homogenization of different samples from various datasets and selection of the contributor genes for the construction of AI diagnostic predictor. (A, B) UMAP plot visualizes the sample distribution. As shown, originally the samples were fairly separated (A), but they were very well homogenous after normalization (B). (C–F) Determination of suitable contributor genes by SVM-RFE, XGBoost-RFE, Boruta, and LASSO, respectively. (G) Venn diagram demonstrated the overlapping genes. 4 genes (i.e., PDHB, CDKN2A, GLS, and SLC31A1) were finally selected.