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

Background Acute myocardial infarction (AMI) ranks among the leading causes of death globally and is linked to obesity and the metabolism of lipids. The objective of this research was to develop an innovative predictive model utilizing obesity and lipid metabolism-related genes (OLMRGs) to facilitate the diagnosis and molecular typing of AMI. Methods Microarray data were obtained from the Gene Expression Omnibus (GEO) repository, while OLMRGs were extracted from the GeneCards and GSEA databases. Important signature genes were pinpointed utilizing univariate regression analysis, LASSO regression, Random Forest, and SVM algorithms. A diagnostic model was then developed using logistic regression. The model's diagnostic efficacy was subsequently confirmed in the validation set GSE59867. Immune infiltration levels were assessed via ssGSEA, and the key genes were validated using RT-qPCR. Results An obesity and lipid metabolism-related genes signature, consisting of five genes (IL1RN, SERPINA1, CEBPB, NFKBIA, and VNN1), was developed as a diagnostic biomarker for AMI (AUC = 0.827) and corroborated in the GSE59876 dataset (AUC = 0.870). The diagnostic model revealed comparisons between groups at high and low risk, identifying twenty-four unique immune cell types alongside nineteen distinct immune functions. Additionally, validation with RT-qPCR confirmed the differential expression of these five signature genes in both AMI and control samples. Conclusion The novel five-gene signature may act as a new biomarker indicating the presence of AMI, providing fresh perspectives for AMI diagnosis and molecular classification.