Machine learning-driven set of peripheral blood microRNAs as diagnostic biomarkers for myocardial infarction

Cardiovascular disease is the leading cause of mortality worldwide and myocardial Infarction (MI) isresponsible for ۸۵% of cardiovascular disease mortality. Since the survival rate in MI cases strongly dependson fast diagnosis and treatment, discovering novel biomarkers for rapid and accurate diagnosis is of greatimportance. MicroRNAs are significant regulators of adaptive and maladaptive responses in cardiovasculardiseases. Hence, microRNAs have undergone intensive research as possible therapeutically and diagnostictargets. However, their role as novel biomarkers for diagnosing MI needs to be more investigated. Themicroarray GSE۶۱۷۴۱ dataset has been downloaded from the Gene Expression Omnibus (GEO) databaseincluding ۸۶۳ microRNAs expression profile in peripheral blood. The selected samples included ۹۴ healthy(as the control) and ۶۲ samples with MI (as the case). At the first, differentially expressed microRNAs hasbeen identified using the limma package with the adjusted P-value <۰.۰۵ and -۱ > log۲ FC > ۱ criteria. Then,sequential forward and backward selection algorithms has been applied for feature selection. Finally, thesupport vector machine (SVM) algorithm has been performed on selected microRNAs to classify sampleswith ۱۰-fold cross-validation. ۱۰۰ differentially expressed microRNAs has been identified in samples withMI compared to healthy samples. ۳۵ microRNAs with the greatest importance has been selected using featureselection algorithms. Among them, a unique signature of five microRNAs (including hsa-miR-۱۲۴۶, hsamiR-۱۲۵۸, hsa-miR-۱۲۷۹, hsa-miR-۱۳۲*, and hsa-miR-۱۴۲-۳p) have been chosen and an SVM model hasbeen trained with their expression values. The trained model predictive values are ۰.۹۲, ۰.۸۴, and ۰.۹۱, forAUC, sensitivity, and specificity, respectively. Based on our findings, the multi-marker approach increasespredictive values in comparison to single microRNAs. Therefore, microRNA signatures derived fromperipheral blood could be valuable novel biomarkers for more accurate diagnosis of MI.

Keywords:

Myocardial infarction , MicroRNA , Machine Learning , MicroRNA Signature , Biomarker