Machine learning-derived identification of an obesity and lipid metabolism-related genes signature for the diagnosis and molecular typing of acute myocardial infarction

Figures

• 

  Figure 1. Differentially expressed obesity and lipid metabolism-related genes (DE-OLMRGs) and function enrichment analysis in acute myocardial infarction (AMI) and control samples. (A) The volcano map of differentially expressed genes (DEGs) between AMI and control samples. (B) Identification of 16 DE-OLMRGs by intersecting the DEGs with LMRGs and ORGs. (C,D) Bubble plots (C) and circle plots (D) of the GO enrichment analysis of the 16 DE-OLMRGs. (E) KEGG pathway analysis of the 16 DE-OLMRGs. DEGs, differentially expressed genes; LMRGs, lipid metabolism-related genes; ORGs, obesity-related genes; GO, Gene Ontology; KEGG, Kyoto Encyclopedia of Genes and Genomes.

• 

  Figure 2. Screening process of diagnostic signature genes for AMI. (A) Forest blot of univariate logistic regression analysis. (B) The outcomes of the least absolute shrinkage and selection operator (LASSO) analysis. (C,D) A plot of biomarkers selection utilizing the support vector machine-recursive feature elimination (SVM-RFE) algorithm. (E) Identification of the AMI-specific genes using the random forest (RF) approach. (F) The most importance genes selected by the RF with the x-axis representing the importance index and the y-axis listing the respective genes. (G) A Venn diagram illustrating seven candidate diagnostic markers that are shared among the logistic regression, LASSO, SVM-RFE, and RF analyses. (H) Validation of the expression levels of candidate diagnostic markers using the GSE59867 dataset. (I) Candidate diagnostic markers were validated in peripheral blood samples from AMI patients and healthy controls through RT-qPCR. (J) A Venn diagram depicting the distribution of diagnostic signature genes. LMRGs, lipid metabolism-related genes; ORGs, obesity-related genes. *p < 0.05, ***p < 0.001.

• 

  Figure 3. Assessment of signature genes and the diagnostic model for AMI diagnosis. (A,B) ROC curves for evaluating the diagnostic performance of signature genes in differentiating AMI patients from control subjects based on the training cohort. (C,D) ROC curves for evaluating the diagnostic performance of signature genes in distinguishing AMI patients from controls based on the testing cohort. (E) Nomogram model. (F,G) Calibration curve (F) and decision curve analysis (DCA) (G) to evaluate the predictive value and the clinical value in the training cohort. (H,I) Calibration curve (H) and DCA (I) to evaluate the predictive value and the clinical value in the testing cohort. ROC, receiver operating characteristic; AUC, area under the curve.

• 

  Figure 4. Analysis of the relationship between risk score and indicated immune cells. (A) Correlation between immune cells and five signature genes alongside the risk score. (B–G) The correlation analysis between the risk score and the proportion of mast cells (B), macrophages (C), neutrophils (D), activated dendritic cells (E), activated CD8T cells (F), and effector memory CD4T cells (G) in AMI. *p < 0.05, **p < 0.01, ***p < 0.001.

• 

  Figure 5. Comparison of the expression of immune cells and immune functions between low-risk and high-risk groups. (A,B) Differential analysis of immune cell infiltration (A) and 29 immune functions (B) across differential risk groups. (C) The results of gene set variation analysis (GSVA) for pathways based on the merged dataset including GSE62646, GSE66360, GSE60993, and GSE48060 databases. *p < 0.05, **p < 0.01, ***p < 0.001.