A Multi-Center Study of Detection of Low Ventricular Ejection Fraction

Overview

This is a multi-site, retrospective study to evaluate the performance of a locked AI-based algorithm for detection of left ventricular systolic dysfunction. A prerequisite for inclusion of subjects from each institution will be the availability of at least one digital 12-lead ECG paired with an echocardiogram with LVEF information within 30 days of the date of the ECG. The AI-ECG LVSD algorithm will be applied on all ECGs and diagnostic performance features for the detection of LVSD will be estimated using the provided paired LVEF value (Low LVEF as the reference label). Performance will also be assessed in subgroups of subjects determined by demographic and clinical factors.

Following institutional review board approval, 12,000 12-lead ECG's paired with an echocardiogram with LVEF information within 30 days of the date of the ECG will be collected across three enrolled sites. Each site will provide data from up to 4000 enrolled subjects that meet the inclusion criteria. No other demographic characteristics or enrichment will be considered in the selection of subjects in order to best represent the general population for that site. Sites will securely transfer the data to a centralized repository for processing. Once data is collected, the device will be used to analyze the ECG data for all enrolled subjects without reference or access to the echocardiogram data. The device will display a binary 36 prediction of the likelihood of LVEF less than or equal to 40%. Results will be compared to the echocardiogram reference standard in accordance with the statistical analysis plan.

Conditions

Interventions