AI-Enhanced ECG Interpretation for Structural Heart Disease

(HEART-AI Trial)

Age: 18+

Sex: Any

Trial Phase: Academic

Sponsor: Montreal Heart Institute

Disqualifiers: Arrhythmia, Heart failure, Coronary artery disease, others

No Placebo Group All trial participants will receive the active study treatment (no placebo)

Compensation: Varies

Number of Visits: Unknown

16160 Participants Needed

Trial Summary

Do I need to stop my current medications for this trial?

The trial information does not specify whether participants need to stop taking their current medications.

What data supports the effectiveness of the treatment ECHONEXT for structural heart disease?

Research shows that using AI with ECGs can effectively identify heart conditions like coronary artery disease and left ventricular dysfunction, which are related to structural heart disease. AI-enhanced ECGs have been shown to accurately detect these conditions, suggesting that similar AI-based approaches like ECHONEXT could be effective in identifying structural heart disease. ^{1 2 3 4 5}

Is AI-enhanced ECG interpretation safe for humans?

The research on AI-enhanced ECG interpretation, like the SEER tool and other AI algorithms, primarily focuses on improving diagnostic accuracy and risk prediction for heart conditions. While these studies do not directly address safety, they suggest that AI tools are used as non-invasive, supportive diagnostic aids, which generally implies a low risk to human safety. ^{2 5 6 7 8}

How is the treatment ECHONEXT different from other treatments for structural heart disease?

ECHONEXT is unique because it uses artificial intelligence to enhance the interpretation of ECGs (electrocardiograms), allowing for the early detection of structural heart diseases like left ventricular dysfunction. This AI-enhanced approach makes it a powerful, non-invasive, and cost-effective screening tool compared to traditional methods, which may not detect asymptomatic conditions as effectively. ^{1 2 6 9 10}

What is the purpose of this trial?

The HEART-AI (Harnessing ECG Artificial Intelligence for Rapid Treatment and Accurate Interpretation) is an open-label, single-center, randomized controlled trial, that aims to deploy a platform called DeepECG at point-of-care for AI-analysis of 12-lead ECGs. The platform will be tested among healthcare professionals (medical students, residents, doctors, nurse practitioners) who read 12-lead ECGs. In the intervention group, the platform will display the ECHONeXT structural heart disease (SHD) scores in randomized patients to help doctors prioritize transthoracic echocardiography (TTEs) and reduce the time to diagnosis of structural heart disease. Also, this platform will display the DeepECG-AI interpretation which detects problems such as ischemic conditions, arrhythmias or chamber enlargements and acts an improved alternative to commercially available ECG interpretation systems such as MUSE. Our primary objective is to assess the impact of displaying the ECHONeXT interpretation on 12-lead ECGs on the time to diagnosis of Structural Heart Disease (SHD) among newly referred patients at MHI. We will compare the time interval from the initial ECG to SHD diagnosis by transthoracic echocardiogram (TTE) between patients in the intervention arm (where ECHONeXT prediction of SHD and TTE priority recommendation are displayed) and patients in the control arm (where ECHONeXT prediction and recommendation are hidden). The main secondary objective is to evaluate the rate of SHD detection on TTE among newly referred patients. We also aim to assess the delay between the time of the first ECG opened in the platform and the TTE evaluation among newly referred patients at high or intermediate risk of SHD. By integrating an AI-analysis platform at the point of care and evaluating its impact on ECG interpretation accuracy and prioritization of incremental tests, the HEART-AI study aims to provide valuable insights into the potential of AI in improving cardiac care and patient outcomes.

Eligibility Criteria

This trial is for healthcare professionals at the Montreal Heart Institute who read ECGs, and patients aged 18 or older with single ventricle or structural heart disease. Participants must have recorded a high-quality 12-lead ECG during the study period and given informed consent.

Inclusion Criteria

ECGs of adequate technical quality for interpretation, as determined by the recording software and visual inspection

I visited the outpatient, cardiology clinic, or emergency room for my ECG.

Users providing clinical care and reading ECGs as part of their practice

Exclusion Criteria

Users who are unable to commit to the duration of the study (approximately 1 month minimum) or adhere to the study protocol

ECG with too many artefacts or without any QRS visible as interpreted by the MUSE GE algorithm

Timeline

Screening

Participants are screened for eligibility to participate in the trial

2-4 weeks

Intervention

Participants in the intervention group receive the ECHONeXT interpretation on 12-lead ECGs to aid in prioritizing transthoracic echocardiography (TTE) and reducing time to diagnosis of structural heart disease.

18 months

Regular visits as per clinical need

Control

Participants in the control group do not receive the ECHONeXT interpretation on 12-lead ECGs.

18 months

Regular visits as per clinical need

Follow-up

Participants are monitored for safety and effectiveness after the intervention period.

4 weeks

Treatment Details

Interventions

• ECHONEXT

Trial Overview The HEART-AI trial tests an AI platform called DeepECG that analyzes ECGs to help diagnose structural heart disease faster. It compares time to diagnosis between those using the AI tool (ECHONeXT scores displayed) and standard care without AI assistance.

Participant Groups

Group I: ECHONEXT interpretation Experimental Treatment 1 Intervention

The ECHONeXT algorithm was trained to predict the presence of SHD on TTE using a single 12-lead ECG. It was developed at Columbia hospital, released as open-weights and validated at the MHI. It was trained on 800,000 ECG and TTE pairs.

Group II: No ECHONEXT interpretation Active Control 1 Intervention

Not displaying the ECHONEXT algorithm interpretation.

Who Is Running the Clinical Trial?

Montreal Heart Institute

Lead Sponsor

Trials

125

Recruited

85,400+

Findings from Research

A study involving 44,959 patients demonstrated that an AI model applied to ECG data can effectively identify asymptomatic left ventricular dysfunction (ALVD), achieving high accuracy (85.7%) and sensitivity (86.3%).

Patients who screened positive for ALVD using the AI model were found to be four times more likely to develop future ventricular dysfunction, highlighting the potential of AI-enhanced ECG as a proactive screening tool for early intervention.

NCBI (Pubmed)

pubmed.ncbi.nlm.nih.gov

Screening for cardiac contractile dysfunction using an artificial intelligence-enabled electrocardiogram. Attia, ZI., Kapa, S., Lopez-Jimenez, F., et al.[2022]

The AI-ECG algorithm demonstrated strong performance in detecting left ventricular systolic dysfunction (LVSD) in an external population of 4277 adults, achieving an area under the receiver operating curve of 0.82, indicating good accuracy.

While the AI-ECG showed high specificity (97.4%) and accuracy (97.0%), its sensitivity was lower at 26.9%, suggesting that population-specific cut-offs may be needed for optimal clinical use, especially given the differences in patient characteristics compared to the original study.

NCBI (Pubmed)

pubmed.ncbi.nlm.nih.gov

External validation of a deep learning electrocardiogram algorithm to detect ventricular dysfunction. Attia, IZ., Tseng, AS., Benavente, ED., et al.[2021]

References

1. Switzerland pubmed.ncbi.nlm.nih.gov

An Artificial Intelligence-Enabled ECG Algorithm for the Prediction and Localization of Angiography-Proven Coronary Artery Disease. [2022]

2. United States pubmed.ncbi.nlm.nih.gov

Screening for cardiac contractile dysfunction using an artificial intelligence-enabled electrocardiogram. [2022]

3. United States pubmed.ncbi.nlm.nih.gov

rECHOmmend: An ECG-Based Machine Learning Approach for Identifying Patients at Increased Risk of Undiagnosed Structural Heart Disease Detectable by Echocardiography. [2022]

4. England pubmed.ncbi.nlm.nih.gov

Deep learning analysis of resting electrocardiograms for the detection of myocardial dysfunction, hypertrophy, and ischaemia: a systematic review. [2023]

5. England pubmed.ncbi.nlm.nih.gov

A deep learning-based electrocardiogram risk score for long term cardiovascular death and disease. [2023]

6. Netherlands pubmed.ncbi.nlm.nih.gov

External validation of a deep learning electrocardiogram algorithm to detect ventricular dysfunction. [2021]

7. United States pubmed.ncbi.nlm.nih.gov

Performance of Off-the-Shelf Machine Learning Architectures and Biases in Detection of Low Left Ventricular Ejection Fraction. [2023]

8. Switzerland pubmed.ncbi.nlm.nih.gov

Current and Future Use of Artificial Intelligence in Electrocardiography. [2023]

9. England pubmed.ncbi.nlm.nih.gov

Artificial Intelligence-Augmented Electrocardiogram Detection of Left Ventricular Systolic Dysfunction in the General Population. [2023]

10. Italy pubmed.ncbi.nlm.nih.gov

[Artificial intelligence applied to the electrocardiogram, or is there really a needle in a haystack?] [2023]