Clinical Validation of SMD-RVECG for Predicting Atrial Fibrillation With Rapid Ventricular Response Within 2 Hours

Overview

The purpose of this retrospective study is to evaluate the clinical performance of SMD-RVECG, an artificial intelligence-based medical device software that analyzes single-lead electrocardiogram data to predict the risk of atrial fibrillation with rapid ventricular response occurring within 2 hours. A total of 348 eligible electrocardiogram datasets collected through VitalDB at Seoul National University Hospital will be included. Atrial fibrillation with rapid ventricular response is defined in this study as atrial fibrillation accompanied by an average heart rate of 110 beats per minute or greater for at least 30 seconds. Eligible datasets will be classified as positive or negative for atrial fibrillation with rapid ventricular response. Two qualified physicians blinded to the software results will review the electrocardiogram data and relevant medical records to establish the reference-standard classification. The blinded electrocardiogram datasets will then be analyzed using SMD-RVECG, and the software-generated predictions will be compared with the reference standard to evaluate clinical performance.

This clinical study is designed as a retrospective, single-center, single-arm, superiority confirmatory clinical performance study conducted at Seoul National University Hospital. Pre-screening Electronic medical records and the VitalDB registry will be reviewed to identify potentially eligible data. The study will use continuous single-lead electrocardiogram data collected through VitalDB between September 1, 2022, and September 30, 2025, from patients who were 19 years of age or older at the time of data collection. Screening and Data Collection A screening number will be assigned sequentially to each potentially eligible dataset. Eligibility will be determined by reviewing the predefined inclusion and exclusion criteria. For eligible cases, continuous single-lead electrocardiogram data, signal loss, heart rate, the occurrence and timing of atrial fibrillation with rapid ventricular response episodes, the electrocardiogram data collection period, demographic information, previous history of atrial fibrillation, and pregnancy status will be collected. Directly identifying personal information will not be collected. Test Dataset Creation Eligible electrocardiogram datasets will be classified into positive and negative groups for atrial fibrillation with rapid ventricular response. The positive group will include electrocardiogram datasets in which atrial fibrillation is followed by rapid ventricular response, defined as an average heart rate of 110 beats per minute or greater for at least 30 seconds. The dataset will include electrocardiogram data preceding the onset of rapid ventricular response and data following the episode, as specified in the study protocol. The negative group will include electrocardiogram datasets with atrial fibrillation that is not accompanied by rapid ventricular response during the predefined assessment period. The planned study dataset consists of 348 cases, including 58 positive cases and 290 negative cases. Reference Standard Establishment Two anesthesiology and pain medicine specialists with at least 5 years of relevant clinical experience in electrocardiogram-based arrhythmia interpretation will independently review the electrocardiogram data and relevant medical records. The reviewers will assess electrocardiogram data quality, confirm the presence or absence and timing of atrial fibrillation with rapid ventricular response, and establish the reference-standard classification. The reviewers will be blinded to each other's initial assessments and to the results generated by SMD-RVECG. If the two reviewers disagree, the classification will be determined through consensus and the basis for the consensus will be documented. If consensus cannot be reached, the dataset will be excluded from the study. Application of the Investigational Device The medical device operator will receive the test electrocardiogram datasets identified only by screening numbers and will be blinded to the positive or negative reference-standard classifications. The datasets will be retrospectively analyzed using SMD-RVECG to generate risk estimates for atrial fibrillation with rapid ventricular response occurring within 2 hours. The software-generated peak risk values and corresponding times will be recorded. Statistical Analysis The software-generated predictions will be compared with the reference-standard classifications. The primary performance measure is the area under the receiver operating characteristic curve for predicting atrial fibrillation with rapid ventricular response within 2 hours. A two-sided 95% confidence interval will be calculated using Newcombe's Wald method. The study will be considered successful if the lower bound of the 95% confidence interval is greater than 0.8. Secondary performance measures include the area under the precision-recall curve, threshold-specific precision, recall, F1 score, sensitivity, specificity, positive predictive value, negative predictive value, confusion-matrix results, and prediction time horizon. Because this retrospective study uses previously collected data, there will be no direct participant contact, additional examination, treatment, or change in clinical care.