The Important Use Of Cardiac Conductance Lines

Cardiovascular disease is the number one cause of death in China, with more than 544,000 cases of sudden death from central origin every year, and its incidence is on the rise with the change of lifestyle and the aging of the population.

Sudden cardiac death (SCD) is an unexpected death from cardiac causes, usually occurring within one hour of the onset of heart disease symptoms, characterized by sudden loss of consciousness, sudden cessation of heart activity, and hemodynamic failure, usually due to persistent ventricular tachycardia or ventricular fibrillation.

Therefore, identifying people at high risk of SCD can go a long way in preventing the disease.

An electrocardiogram (ECG) is a widely used method for detecting the electrical activity of the heart, and the usual clinical standard uses a 12-lead ECG to assess a patient's heart health.As a key component connecting the patient's body with the electrocardiogram instrument, the cardiac conductance wire plays a vital role in the quality of the electrocardiogram signal.

According to a new study published in Communications Medicine, one of the world's leading scientific journals,The ECG-based Deep Learning model for assessment of SCD risk can more accurately distinguish SCD cases from control groups than the traditional ECG risk model.This will help clinicians screen and identify people at higher risk for SCD, so they can be tested regularly to prevent SCD from occurring.

The team used data from two geographically independent, prospective, community-based ongoing out-of-hospital SCD studies, with a total sample of 2,510 SCD cases. Deep learning (DL) models were trained, validated, and tested on 1796 SCD Sudden Accidental Death Study data from Oregon, USA (Oregon SUDS), and externally validated on 714 Sudden Death Prediction Study data from Multi-ethnic Communities in Ventura County, California (Ventura PRESTO). All cases of resting 12-lead electrocardiograms available for analysis, which were recorded prior to SCD and were not associated with SCD events, were included in the study, excluding ECG with pacing rhythm, atrial fibrillation, or atrial flutter a priori to create a DL model that could be applied to sinus rhythm electrocardiograms.

The team developed a convolutional neural network model capable of identifying SCD cases using 12-lead ECG waveforms. The researchers obtained two independent control samples from 1,342 electrocardiograms from 1,325 individuals who had at least 50 percent of coronary artery disease. They trained the model using 1076 SCD cases from Oregon SUDS and 1,101 12-lead ECGs before cardiac arrest, 597 SCD cases from the control group and 613 12-lead ECGs. A separate validation cohort, consisting of 366 pre-cardiac arrest ECGs and 200 control ECGs, was used to determine when to stop model training. The study samples were divided by patient level so that multiple electrocardiograms from the same patient could be included in the same cohort.

Compared with traditional ECG models, the DL model based on 12-lead ECG waveform was more accurate in identifying SCD cases, with an internal cohort AUROC (a neural network evaluation measure) of 0.889 and an externally validated AUROC of 0.820, and was better than the traditional ECG risk score. This is the first report that an ECG-based DL model outperforms traditional ECG risk models in predicting out-of-hospital SCD at the community level.