Home-based Outpatient Multicenter Evaluation Using Electrocardiogram (HOME-ECG)

Prospective Multicenter Validation of a Home-Based Artificial Intelligence Enabled Single-Lead Electrocardiogram for Detecting Low Ejection Fraction and Structural Heart Disease

Status

Recruiting

Phases

Unknown

Study type

Observational

Source

ClinicalTrials.gov

Registry ID

NCT07488052

Acronym

HOME-ECG

Enrollment

5000

Registered

2026-03-23

Start date

2026-03-01

Completion date

2026-09-30

Last updated

2026-03-25

For informational purposes only — not medical advice. Sourced from public registries and may not reflect the latest updates. Terms

Conditions

Heart Failure

Brief summary

This prospective multicenter observational study will validate the accuracy of a previously developed artificial intelligence-enabled single-lead electrocardiogram (AI-ECG) model for identifying low ejection fraction and other structural heart disease phenotypes. Adult participants receiving a model-compatible single-lead electrocardiogram (ECG) (Apple Watch and QOCA ECG102D) and transthoracic echocardiography at five hospitals in Taiwan will be enrolled between March 1, 2026 and June 30, 2026. Model predictions will be compared with echocardiographic reference standards obtained within 30 days after the index ECG.

Interventions

DIAGNOSTIC_TESTSingle-Lead ECG Acquisition [Apple Watch and QOCA ECG102D]

A single-lead Lead-I ECG recorded using a device compatible with the prespecified AI-ECG pipeline and analyzed offline by the locked AI model.

Study design

Observational model

COHORT

Time perspective

PROSPECTIVE

Eligibility

Sex/Gender

ALL

Age

20 Years to 90 Years

Healthy volunteers

Yes

Inclusion criteria

* Age 20 years or older * Single-lead ECG recorded using a study-compatible device * Ability to comply with study procedures and, when applicable, provide informed consent according to local institutional review board (IRB) requirements * No transthoracic echocardiography performed within 90 days before the index ECG

Exclusion criteria

* ECG signal quality insufficient for prespecified AI analysis * No transthoracic echocardiography available within 30 days after the index ECG * Echocardiography unavailable or technically inadequate for determining left ventricular ejection fraction (EF)

Design outcomes

Primary

Measure	Time frame	Description
Area Under the Receiver Operating Characteristic Curve for Detection of Low Left Ventricular Ejection Fraction	30 days	AUC with 95% confidence interval for detection of low EF, defined as echocardiographic LVEF ≤40%, by the pre-specified AI-enabled single-lead ECG model.
Sensitivity, Specificity, Positive Predictive Value, and Negative Predictive Value for Low EF Detection at the Pre-specified Operating Threshold	30 days	Diagnostic operating characteristics of the locked AI model for low EF at the pre-specified threshold.

Secondary

Measure	Time frame	Description
Positive Predictive Value and Negative Predictive Value for Structural Heart Disease at the Pre-specified Operating Threshold	30 days	Prevalence of structural heart disease in patients identified as positive and negative by AI-ECG for detecting low EF, including left atrial enlargement, valvular heart disease, pulmonary hypertension, left ventricular hypertrophy, etc.
Difference in All-Cause Mortality Risk at the Pre-specified Operating Threshold	90 days	All-cause mortality risk in patients identified as positive and negative by AI-ECG for detecting low EF.

Countries

Taiwan

Contacts

CONTACTCHIN LIN, PhD

xup6fup0629@gmail.com+886958259269

PRINCIPAL_INVESTIGATORCHIN LIN, PhD

National Defense Medical Center

Outcome results

None listed

Source: ClinicalTrials.gov · Data processed: Mar 26, 2026