Construction of genetic classification model for coronary atherosclerosis heart disease using three machine learning...

Area of Science:

• Genomics
• Bioinformatics
• Cardiovascular Disease Research

Background:

• Early-stage coronary artery disease (CAD) often presents asymptomatically, leading to missed diagnoses.
• Gene expression levels fluctuate during disease development, offering potential for diagnostic biomarkers.
• Current diagnostic methods for CAD require innovation to improve early detection.

Purpose of the Study:

• To construct genetic classification models for CAD using gene expression data.
• To identify key genes and gene modules associated with CAD pathogenesis.
• To provide new insights into the molecular mechanisms underlying CAD.

Main Methods:

• Utilized R software and downloaded three CAD-related gene expression datasets (GSE12288, GSE7638, GSE66360) from the Gene Expression Omnibus database.
• Employed Limma package for identifying differentially expressed genes (DEGs) and WGCNA for recognizing CAD-related gene modules and hub genes.
• Applied recursive feature elimination to select optimal feature genes (OFGs) and constructed classification models using Support Vector Machine (SVM), Random Forest (RF), and Logistic Regression (LR), validated with ROC curve analysis.

Main Results:

• Identified 374 DEGs, eight gene modules, 33 hub genes, and 12 OFGs (including HTR4, KISS1, CA12, CAMK2B, KLK2, DDC, CNGB1, DERL1, BCL6, LILRA2, HCK, MTF2).
• The Support Vector Machine (SVM) model achieved the highest accuracy (75.58%) and an Area Under the Curve (AUC) of 0.813.
• Random Forest (RF) and Logistic Regression (LR) models showed accuracies of 63.57% and 63.95%, respectively, with AUCs of 0.727 and 0.783.

Conclusions:

• The study identified 12 gene signatures crucial to the pathogenic mechanisms of CAD.
• The SVM model exhibited superior performance in classifying CAD compared to RF and LR models.
• These findings suggest the potential of gene expression-based classifiers for early CAD diagnosis and understanding its pathogenesis.