Clinically-Linked ECG and Arrhythmia Repository (CLEAR)
Invention Summary
Duke inventors have developed a database of digital 12- lead ECG samples that represent key rhythms of cardiac disorders. Each patient’s disorder was validated by a licensed and board-certified cardiovascular medicine physician to ensure each ECG sample accurately represents the rhythm of the disease. This is intended to be utilized by companies developing medical devices that detect and analyze heart rhythms. Specifically, the detected rhythms of novel devices can be compared to the database to validate their diagnostic, prognostic, or rhythm identification algorithm and satisfy the FDA’s requirements for investigational device exemption (IDE) and premarket approval. These ECGs are linked at the patient level to several other clinical databases (see below) to support data science inquiries into both cross-sectional and longitudinal outcomes.
Materials and Results
Over 4.5 million de-identified electrocardiograms (ECGs) acquired through routine clinical care have been catalogued in this database. Data can be exported as digital waveforms in .csv or .xml format to support waveform level analyses. These waveforms can be cross-referenced to demographic information, validated electrocardiographic diagnoses/intervals, and/or elements contained in linked datasets. These linked datasets include:
• Longitudinal echocardiographic data
• Clinical outcomes as ascertained by billing codes
• Cardiac Implantable Electronic Device (CIED) interrogations
• Ambulatory electrocardiographic information
• Cardiac procedure information (heart catheterization, electrophysiology procedures, surgery etc.)