Abstract

Ventricular arrhythmias (VA) are still the leading cause of sudden cardiac death. Therefore the identification of the predictors of high-grade VA and risk stratification of fatal events is important for clinical cardiology. We analyzed the clinical and Holter monitoring data of 993 patients with and without high-grade VA, referred for the coronary angiography. Patients with VA were older (57.2±8.4 years vs. 54.9±8.3 years, р=0.001), had lower left ventricular (LV) ejection fraction (51.6±11.2% vs. 58.4±7.6%, p<0.001), and the larger extent of LV wall motion abnormalities (10.8±15.7% vs. 5.5±10.9, р=0.001). In patients with VA, LV end diastolic diameter was significantly greater (54.3±7.4 mm vs. 49.9±4.7 mm, p<0.001), and severe functional class of chronic heart failure (NYHA) was more common (28.1% vs. 15.5%, p<0.001). For VA prediction, we used mathematical model, artificial neuronal network (ANN), and multilayer perceptron (3 neurons in input layer; 11 neurons in hidden layer, and 2 neurons in output layer). Sensitivity and specificity rates of this model were 83.58% and 53.8%, respectively. Model of ANN demonstrated high diagnostic accuracy in prediction of high-grade VA development in all three samples: learning, testing, and control. For prediction of high-grade VA, computer software “The Diagnostic Calculator” was proposed.

Full Text
Published version (Free)

Talk to us

Join us for a 30 min session where you can share your feedback and ask us any queries you have

Schedule a call