Figure 1. Flow chart of the study.

Figure 2. GSE66360 dataset pre-processing. Note. Figures 2A–B is the schematic diagram of the GSE66360 dataset before and after correction, and the corrected dataset has a good consistency. Figure 2C is the differential gene volcano map of GSE66360; Figure 2D is the differential gene heat map of GSE66360.

Figure 3. GO/KEGG/DO enrichment analysis of differential genes. Note. Figure 3A–C shows the GO, KEGG, and DO enrichment analysis of the differential genes in order. In GO string plot, the gene name is on the left side, and the GO entry is on the right side. The more GO entries corresponding to a gene indicate that the gene mediates AMI through multiple mechanisms, and the legend in the lower-left corner represents the logFC value of the gene. In KEGG plot, the inner concentric circles represent the gene, the outer concentric circles represent the signaling pathways enriched by the gene, and the legend in the lower-left corner represents the logFC value of the gene. In DO bar plot, the vertical coordinate is the disease in which the gene is enriched, and the horizontal coordinate is the number of enriched genes corresponding to the disease.

Figure 4. SVM and random forest tree to filter feature genes. Note. Figure 4A shows the schematic diagram of SVM. The horizontal coordinate is the number of genes, and the vertical coordinate is the error of cross-validation. When the number of genes is 10, the error is the smallest, so 10 feature genes are derived. Figure 4B shows the schematic diagram of a random forest tree. The horizontal coordinate is the gene importance score, and the vertical coordinate is the gene name. The gene ranked in top 30 is selected as the feature gene. Figure 4C shows the feature genes screened by SVM and random forest tree gene intersection Venn diagram.

Figure 5. Differential gene validation and ROC curve. Note. Figure 5A–F indicates the variability of target genes in the validation set, the horizontal coordinates indicate the normal group and the control group, and the vertical coordinates indicate the gene expression. Figure 5G–L shows the ROC curves of GSE66360, the horizontal coordinates indicate the False Positive Rate (FPR), and the vertical coordinates indicate the True Positive Rate (TPR). The larger the AUC, the higher the accuracy of the feature gene. Figure 5M–R shows the ROC curve of GSE61145.

Figure 6. "Feature gene-drug component-Chinese medicine" network diagram and Chinese medicine collation. Note. Figure 6A is a network diagram of the "signature gene-drug component-Chinese medicine" for the top five Chinese herbal medicines, in which the red line represents the up-regulation relationship, the green line represents the down-regulation relationship, and the purple line represents the bidirectional regulation. The left inverted triangle represents the five feature genes, the middle concentric circles represent the components with regulatory effects on the feature genes, where the outer green circle represents the components with up-regulation, the red represents the components with down-regulation, the blue represents the components with bidirectional regulation, and the right diamond represents the Chinese herbal medicines mapped to the components (this paper only showed the mapping process of the top five Chinese herbal medicines). Figure 6B–D: radar diagrams of the four qi, the five flavors, and the attribute of potential Chinese medicines are shown in order, and the numbers represent the frequencies of their respective occurrences.