Modeling the diagnosis of coronary artery disease by discriminant analysis and logistic regression: a cross-sectional study

Risk factors for high CAD-RADS scoring in CAD patients revealed by machine learning methods: a retrospective study

OBJECTIVE

This study aimed to investigate a variety of machine learning (ML) methods to predict the association between cardiovascular risk factors and coronary artery disease-reporting and data system (CAD-RADS) scores.

METHODS

This is a retrospective cohort study. Demographical, cardiovascular risk factors and coronary CT angiography (CCTA) characteristics of the patients were obtained. Coronary artery disease (CAD) was evaluated using CAD-RADS score. The stenosis severity component of the CAD-RADS was stratified into two groups: CAD-RADS score 0-2 group and CAD-RADS score 3-5 group. CAD-RADS scores were predicted with random forest (RF), k-nearest neighbors (KNN), support vector machines (SVM), neural network (NN), decision tree classification (DTC) and linear discriminant analysis (LDA). Prediction sensitivity, specificity, accuracy and area under the curve (AUC) were calculated. Feature importance analysis was utilized to find the most important predictors.

RESULTS

A total of 442 CAD patients with CCTA examinations were included in this study. 234 (52.9%) subjects were CAD-RADS score 0-2 group and 208 (47.1%) were CAD-RADS score 3-5 group. CAD-RADS score 3-5 group had a high prevalence of hypertension (66.8%), hyperlipidemia (50%) and diabetes mellitus (DM) (35.1%). Age, systolic blood pressure (SBP), mean arterial pressure, pulse pressure, pulse pressure index, plasma fibrinogen, uric acid and blood urea nitrogen were significantly higher (p < 0.001), and high-density lipoprotein (HDL-C) lower (p < 0.001) in CAD-RADS score 3-5 group compared to the CAD-RADS score 0-2 group. Nineteen features were chosen to train the models. RF (AUC = 0.832) and LDA (AUC = 0.81) outperformed SVM (AUC = 0.772), NN (AUC = 0.773), DTC (AUC = 0.682), KNN (AUC = 0.707). Feature importance analysis indicated that plasma fibrinogen, age and DM contributed most to CAD-RADS scores.

CONCLUSION

ML algorithms are capable of predicting the correlation between cardiovascular risk factors and CAD-RADS scores with high accuracy.

Intelligent prediction of major adverse cardiovascular events (MACCE) following percutaneous coronary intervention using ANFIS-PSO model

Background

This study aimed to use the hybrid method based on an adaptive neuro-fuzzy inference system (ANFIS) and particle swarm optimization (PSO) to predict the long term occurrence of major adverse cardiac and cerebrovascular events (MACCE) of patients underwent percutaneous coronary intervention (PCI) with stent implantation.

Method

This retrospective cohort study included a total of 220 patients (69 women and 151 men) who underwent PCI in Ekbatan medical center in Hamadan city, Iran, from March 2009 to March 2012. The occurrence and non-occurrence of MACCE, (including death, CABG, stroke, repeat revascularization) were considered as a binary outcome. The predictive performance of ANFIS model for predicting MACCE was compared with ANFIS-PSO and logistic regression.

Results

During ten years of follow-up, ninety-six patients (43.6%) experienced the MACCE event. By applying multivariate logistic regression, the traditional predictors such as age (OR = 1.05, 95%CI: 1.02–1.09), smoking (OR = 3.53, 95%CI: 1.61–7.75), diabetes (OR = 2.17, 95%CI: 2.05–16.20) and stent length (OR = 3.12, 95%CI: 1.48–6.57) was significantly predicable to MACCE. The ANFIS-PSO model had higher accuracy (89%) compared to the ANFIS (81%) and logistic regression (72%) in the prediction of MACCE.

Conclusion

The predictive performance of ANFIS-PSO is more efficient than the other models in the prediction of MACCE. It is recommended to use this model for intelligent monitoring, classification of high-risk patients and allocation of necessary medical and health resources based on the needs of these patients. However, the clinical value of these findings should be tested in a larger dataset.