Willem AI ECG assessment: There is no study intervention. The Willem AI platform will assess all study ECGs for the identification of cardiac patterns, arrhythmias, and/or cardiac diseases. Regardless of retrospective or prospective enrollment, Willem output will not be provided to the healthcare professional user for clinical evaluation, and therefore routine practice will not be impacted nor altered.
Study summary
The WILLEM Registry is a large-scale, single-group, observational, registry study to collect continuous clinical evidence of Willem in real-world settings. Cardiovascular diseases are a major problem for public health and healthcare systems. Electrocardiograms (ECGs) are simple tests which increase diagnostic performance and early detection of cardiovascular diseases. However, its interpretation is complex, time consuming for cardiology experts, and entails high costs for healthcare systems. Willem allows AI-based automatic interpretation and its performance has been examined in previous clinical trials, but additional clinical evidence is needed for its integration in real-world clinical settings. This study will collect clinical evidence of Willem performance to detect cardiac abnormalities in ECGs from high-risk cardiac patients admitted to cardiovascular units.
Eligibility
Sex
ALL
Min age
18 Years
Max age
—
Healthy volunteers
Accepted
Inclusion Criteria:
* EC/IRB approval of ICF waiver prior to recruitment; otherwise, signed informed consent form by subject and investigator
* Age \> 18 years-old, with no upper limit
* Subjects undergoing standard of care electrocardiogram (ECG) of any duration from any hardware device
* All available, but at least one, legible ECG tracings in raw data format (e.g. DICOM, XML, EDF, JSON, HL7, SCP, WFDB, CSV, etc.)
* Available subject clinical data associated with the ECG
* For 12-lead ECGs, a minimum length of 10 seconds at a minimum sample frequency of 250 Hz
* For ECGs from Holters, wearables, patches, insertable cardiac monitors, telemetries, etc., a minimum length of 30 seconds at a minimum sample frequency of 200 Hz with a lead I / II or its MCL-DII lead approximation
* For prospective eligibility only:
* Signed informed consent form, unless previously waived by the EC/IRB
* Site technical viability for ECG and subject clinical data transfer (e.g. end-to-end integration following interoperability standards such as FHIR, HL7 or DICOM)
Exclusion Criteria:
* Unavailable or suboptimal quality of the raw data from the ECG signal
* Age \< 18 years-old
Primary outcome measure(s)
Primary endpoint analysis: Willem performance — From enrollment to any standard of care timepoint when the patient underwent (retrospective) or will undergo within the next 10 years (prospective) an eligible electrocardiogram ECG data will be categorized according to SOC-defined cardiopathies, arrhythmic events, and cardiac diseases. If SOC diagnosis is unavailable or inconsistent, an independent committee of expert cardiologists will review and provide their diagnosis according to a cardiac defined ontology which extends values defined in HL7-aECG data store. Then, the performance of Willem to detect cardiac patterns, arrhythmias, and cardiac disease from ECGs will be assessed. In order to define True Positive, True Negative, False Positive, and False Negative classifications, the ground truth for comparison will be Standard Of Care (SOC) manually performed cardiologist diagnosis.
Performance metrics such as diagnostic accuracy, sensitivity, specificity, predictive positive value (PPV), negative predictive value (NPV), F1-Score and Area Under the Receiver Operating Characteristic Curve (AUROC) will be obtained.
This page summarises publicly available registry data for informational purposes — not medical advice. Eligibility is determined by each study team; patients should discuss participation with their clinician.
We use cookies to analyse site traffic and improve your experience. With your consent, we may also use cookies for advertising. You can change your choice at any time on our Cookie Policy page. See also our Privacy Policy.