Machine learning-based mRNA signature in early acute myocardial infarction patients: the perspective toward...

Patients diagnosed with stable coronary artery disease (CAD) are at continued risk of experiencing acute myocardial infarction (AMI). This study aims to unravel the pivotal biomarkers and dynamic immune cell changes, from an immunological, predictive, and personalized viewpoint, by implementing a machine-learning approach and a composite bioinformatics strategy. Peripheral blood mRNA data from different datasets were analyzed, and CIBERSORT was used for deconvoluting human immune cell subtype expression matrices. Weighted gene co-expression network analysis (WGCNA) in single-cell and bulk transcriptome levels was conducted to explore possible biomarkers for AMI, with a particular emphasis on examining monocytes and their involvement in cell-cell communication. Unsupervised cluster analysis was performed to categorize AMI patients into different subtypes, and machine learning methods were employed to construct a comprehensive diagnostic model to predict the occurrence of early AMI. Finally, RT-qPCR on peripheral blood samples collected from patients validated the clinical utility of the machine learning-based mRNA signature and hub biomarkers. The study identified potential biomarkers for early AMI, including CLEC2D, TCN2, and CCR1, and found that monocytes may play a vital role in AMI samples. Differential analysis revealed that CCR1 and TCN2 exhibited elevated expression levels in early AMI compared to stable CAD. Machine learning methods showed that the glmBoost+Enet [alpha=0.9] model achieved high predictive accuracy in the training set, external validation sets, and clinical samples in our hospital. The study provided comprehensive insights into potential biomarkers and immune cell populations involved in the pathogenesis of early AMI. The identified biomarkers and the constructed comprehensive diagnostic model hold great promise for predicting the occurrence of early AMI and can serve as auxiliary diagnostic or predictive biomarkers.