ECG database of short fragments with arrhythmias classification according to the degree of danger to the patient’s life

Studying is devoted to the development of a new database of short ECG fragments designed to solve the problems of recognizing dangerous arrhythmias during monitoring cardiological control. The database contains various types of rhythm disturbances, including extremely dangerous arrhythmias. It includes one thousand sixteen 2-second ECG fragments. This database allows to classify cardiac arrhythmias and determine the degree of danger to the patient’s life. The database contains six classes of ECG signals, the first three classes including the most dangerous forms of disorders: 1) ventricular flutter and ventricular fibrillation, 2) ventricular tachycardia torsade de pointes, 3) high-rate ventricular tachycardia. The remaining classes represent arrhythmias that are not fatal. Ventricular and supraventricular disorders were classified into different classes according to the degree of danger to the patient’s life. In addition to 2-second fragments, this database contains the full spectrum of ECG fragments (0-180 Hz) and smoothed spectrums (0-15 Hz). This representation allows to study the frequency properties of certain types of ECG signals and build algorithms for arrhythmia recognition in the spectral region. The description of the database structure is given, the content of each of the considered classes is considered. The proposed database of ECG signals can evaluate the effectiveness of your own or existing algorithms for recognizing life-threatening arrhythmias.