Predicting Severe Cardiac Arrhythmias in the Perioperative Period Using AI-ECG

Overview

During the preoperative examination, in addition to the standard examination recommended before high-risk cardiac surgeries, a digital single-lead ECG is recorded using portable CardioQUARK devices. This ECG is analyzed for time, amplitude, and frequency parameters-more than 200 parameters per cardiac cycle. Additionally, the available parameters of a standard 12-lead ECG recording are determined. All parameters of both single-lead and 12-lead ECGs will be included in a multivariate regression analysis to determine the optimal method for predicting atrial fibrillation, life-threatening cardiac arrhythmias, and conduction abnormalities in the perioperative period after high-risk cardiac procedures. The study will include patients undergoing high-risk cardiac surgeries, such as those on the heart, aorta, esophagus, pancreas, liver, and lungs, as well as high-risk oncological surgeries. A partial analysis will also be performed depending on the type of intervention. Rhythm and conduction disturbances will be recorded during surgery, as well as in the postoperative period, using ECG monitors, 24-hour ECG monitoring, and repeated recordings of a standard ECG when complaints or signs of unstable hemodynamics arise

In study, the investigators plan to identify single-channel and 12-lead ECG parameters that correlate significantly with the occurrence of atrial fibrillation and life-threatening cardiac arrhythmias and conduction disturbances in the perioperative period after high-risk cardiac interventions. A prospective, non-randomized, observational, controlled, single-center study consisting of two consecutive stages is being conducted. The first stage involves identifying single-channel and 12-lead ECG parameters that correlate significantly with the onset of atrial fibrillation and life-threatening cardiac arrhythmias and conduction disorders in the perioperative period after interventions with high cardiac risk. Participants who meet the inclusion criteria will undergo the following during preoperative preparation: recording of a digital single-channel ECG using the CardioQVARK software and hardware complex (CardioQVARK medical device (Electrocardiograph) issued by Roszdravnadzor (Federal Service for Surveillance in Healthcare) P3H 2024/22855, issued on June 5, 2024); a digital 12-lead ECG; an extended echocardiographic examination; and 24-hour Holter ECG monitoring. The single-channel ECG will be recorded for 3 minutes at rest in a sitting position. The participants leans back against the back of the chair, legs uncrossed, arms resting on the knees or on the table. Standard lead I is recorded using the CardioQVARK software and hardware complex. Next, the parameters of single-channel and 12-channel ECG are identified that have a reliable correlation with the occurrence of atrial fibrillation and life-threatening arrhythmias and conduction disorders in the perioperative period after interventions with high cardiac risk, followed by the creation of a machine learning model and/or or neural network analysis with deep learning support for predicting the occurrence of atrial fibrillation and life-threatening arrhythmias and conduction disorders. The second stage will involve testing using a test sample of the algorithm, taking into account the ECG parameters which, according to the statistical analysis performed in the first stage, had the highest sensitivity and specificity in predicting the onset of atrial fibrillation and life-threatening cardiac arrhythmias and conductions disorders.