Towards Diagnostic Aided Systems in Coronary Artery Disease Detection: A Comprehensive Multiview Survey of the State of the Art

Abstract

Introduction. Coronary artery disease (CAD) is one of the main causes of death all over the world. One way to reduce the mortality rate from CAD is to predict its risk and take effective interventions. The use of machine learning- (ML-) based methods is an effective method for predicting CAD-induced death, which is why many studies in this field have been conducted in recent years. Thus, this study aimed to review published studies on artificial intelligence classification algorithms in CAD detection and diagnosis. Methods. This study systematically reviewed the most cutting-edge techniques for analyzing clinical and paraclinical data to quickly diagnose CAD. We searched PubMed, Scopus, and Web of Science databases using a combination of related keywords. A data extraction form was used to collect data after selecting the articles based on inclusion and exclusion criteria. The content analysis method was used to analyze the data, and based on the study’s objectives, the results are presented in tables and figures. Results. Our search in three prevalent databases resulted in 15689 studies, of which 54 were included to be reviewed for data analysis. Most studies used laboratory and demographic data classification and have shown desirable results. In general, three ML methods (traditional ML, DL/NN, and ensemble) were used. Among the algorithms used, random forest (RF), linear regression (LR), neural networks (NNs), support vector machine (SVM), and K-nearest networks (KNNs) have the most applications in the field of code recognition. Conclusion. The findings of this study show that these models based on different ML methods were successful despite the lack of a benchmark for comparing and analyzing ML features, methods, and algorithms in CAD diagnosis. Many of these models performed better in their analyses of CAD features as a result of a closer look. In the near future, clinical specialists can use ML-based models as a powerful tool for diagnosing CAD more quickly and precisely by looking at its design’s technical facets. Among its incredible outcomes are decreased diagnostic errors, diagnostic time, and needless invasive diagnostic tests, all of which typically result in decreases in diagnostic expenses for healthcare systems.