An interventional study of ECHONEXT in Structural Heart Abnormality and Structural Heart Disease, sponsored by Montreal Heart Institute. Enrolling by invitation at 1 site in Canada. Open to participants aged 18 Years and older. Per ClinicalTrials.gov, last updated 2026-10-13.
Sponsored by Montreal Heart Institute · Not applicable, Interventional, and Diagnostic
Help build ClinicalTrials.gg — short surveys, remote interviews or testing, in your language. Take part →
HEART-AI is a pragmatic randomized trial currently active at Montreal Heart Institute, with planned multicenter expansion testing whether an artificial intelligence (AI) score from a routine 12-lead electrocardiogram (ECG) helps clinicians identify structural heart disease sooner. Structural heart disease includes reduced heart pumping function, valve disease, pulmonary hypertension, thickened heart muscle and fluid around the heart. The study plans to enroll 5,700 adults seen in outpatient clinics or ambulatory emergency departments who have no known structural heart disease.
Patients are assigned equally to standard care with or without the EchoNext AI risk score and echocardiography recommendation. For high-risk patients, EchoNext recommends a transthoracic echocardiogram (TTE, an ultrasound of the heart) within 2 to 4 weeks, no later than 30 days. Low-risk patients receive routine care, with TTE when clinically indicated. Clinicians retain responsibility for all care decisions. The score is calculated in both groups but shown only in the AI group.
The main outcome is the proportion of patients with structural heart disease documented on a clinical TTE within 30 days of their first study ECG. Other outcomes include diagnoses within 12 months, time to TTE and diagnosis, imaging use, management changes and model performance.
Design and population HEART-AI is a pragmatic, parallel-group trial currently active at one site, Montreal Heart Institute, with planned multicenter expansion with patient-level 1:1 randomization through DeepECG.ai at review of the first eligible clinically indicated ECG. Randomization is stratified by center, biological sex (administrative sex as proxy), and setting (outpatient versus ambulatory emergency department), using permuted blocks of varying size. Allocation remains fixed for subsequent ECGs; only the first ECG is analyzed. The planned expansion includes Montreal Heart Institute (MHI), McGill University Health Centre, University of Ottawa Heart Institute, Jewish General Hospital, University Health Network, Marienkrankenhaus Hamburg, Sorel and Joliette. Each new site activates only after ethics approval and an executed data-processing agreement. Study completion is anticipated in 2027. MHI completed setup in 2025 and continues randomizing. The target is 5,700 patients.
Intervention and workflow Providers confirm eligibility before randomization and record their TTE ordering intent and intended priority before any AI result is displayed. The intervention arm receives standard care plus the EchoNext SHD risk score and TTE recommendation: high risk, TTE within 2 to 4 weeks (≤30 days); low risk, routine care and TTE only if clinically indicated. The control arm receives standard ECG interpretation and clinical judgment without displayed AI scores or recommendations. Scores are computed in both arms. Providers manage both arms and retain full clinical autonomy. The post-ECG form records TTE referral, intended priority (≤30 or >30 days), and AI use in the intervention arm. Model weights remain frozen throughout the trial.
SHD definition and outcome ascertainment SHD is any of: LVEF ≤45% measured by Simpson biplane; moderate or greater RV systolic dysfunction (TAPSE ≤17 mm, fractional area change ≤35%, or moderate-or-worse impairment on visual assessment); systolic pulmonary artery pressure ≥45 mm Hg; moderate or greater mitral, tricuspid, pulmonic or aortic regurgitation or aortic stenosis; septal or posterior LV wall thickness ≥1.3 cm; or moderate or large pericardial effusion. Two trained abstractors blinded to arm assess clinical TTE reports. Disagreements are adjudicated by a study cardiologist not involved in the patient's care.
The primary numerator is any qualifying clinical TTE documenting SHD within 30 days of the index ECG; the denominator is all randomized patients. The earliest SHD-positive clinical TTE is used. Patients without an accessible qualifying result at 30 days, including deaths, have no documented 30-day SHD diagnosis. Each SHD component is reported separately by arm. Primary analysis follows intention-to-treat and uses GEE logistic regression adjusted for center, sex and setting, clustered on provider. Secondary time-to-TTE and time-to-SHD analyses censor at 12 months and treat death as a competing event. Clinical and process outcomes are also reported by biological sex.
Monitoring A blinded interim at approximately 2,850 patients re-estimates the pooled 30-day SHD rate. There is no early stopping for efficacy or futility. No DSMB is constituted. An independent safety officer with no role in trial conduct or platform development reviews safety every 6 months, focusing on low-risk under-diagnosis and TTEs performed beyond the intended priority window. Seriant conducts independent analyses.
Outcome amendment rationale, October 4, 2026 The primary outcome is revised from time to SHD diagnosis to the proportion of randomized patients with clinical TTE-documented SHD within 30 days, aligning the registry with the revised multicenter grant's primary objective and sample-size calculation. This measures whether AI-guided prioritization closes the gap in timely SHD detection. Time to SHD diagnosis is retained as a secondary outcome with follow-up through 12 months.
This outcome amendment precedes any unblinded comparison of randomized study arms. There is no verification TTE substudy.
Montreal Heart Institute is the lead sponsor of 101 studies on the registry; 23 are open to participants now.
Counted across the registry records on this site, refreshed daily.
The criteria were revised on the registry on Oct 13, 2026. See what changed
Go through the criteria one by one — yes, no or not sure — and take the list to your doctor or the study team. Kept in this browser only.
Providers:
- Licensed healthcare professionals (physicians, nurse practitioners, physician assistants, and residents) who interpret ECGs in ambulatory cardiology clinics or emergency departments and have authority to order TTEs.
Patients:
ECGs:
- Standard 12-lead ECGs in XML format recorded after study initiation and successfully transmitted to DeepECG.ai.
Exclusion Criteria:
Providers:
- None.
Patients:
- Inpatient or hospitalized patients at the time of . Patients randomized in the ambulatory emergency department who are subsequently hospitalized remain eligible.
ECGs:
- Failed automated quality checks, including poor technical quality or indiscernible QRS waveforms.
Underlined terms in the criteria open these too. Explained by clinicaltrials.gg, not part of the registry record. The study team decides who can take part.
A recording of the heart’s electrical activity, taken with stickers on the chest, arms and legs. It takes a few minutes and does not hurt.
ECG (electrocardiogram)
A recording of the heart’s electrical activity, taken with stickers on the chest, arms and legs. It takes a few minutes and does not hurt.
Explained by clinicaltrials.gg, not part of the registry record. The study team decides who can take part.
Standard care plus the EchoNext structural heart disease risk score and TTE priority recommendation displayed through DeepECG.ai at review of the index 12-lead ECG. High risk: recommend TTE within 2 to 4 weeks (≤30 days). Low risk: routine care, TTE only if clinically indicated, and symptom monitoring. Scores are computed in both arms but displayed only in this arm. Model weights remain frozen for the trial. Before display, providers record TTE ordering intent and intended priority; after ECG review they record final referral and intended priority (≤30 or \>30 days). Providers retain full clinical autonomy.
Other: ECHONEXT
Standard ECG interpretation with clinician judgment and physical examination for TTE referral. EchoNext SHD risk scores are computed but neither the score nor the TTE recommendation is displayed to providers. Before ECG interpretation, providers record TTE ordering intent and intended priority; final referral and intended priority (≤30 or \>30 days) are recorded afterward. Providers retain full access to medical records and full clinical autonomy.
EchoNext is an ECG-based artificial intelligence algorithm that predicts structural heart disease from a clinically indicated 12-lead ECG. Through DeepECG.ai, intervention-arm providers receive the SHD risk score and a TTE priority recommendation: high risk, TTE within 2 to 4 weeks (≤30 days); low risk, routine care and TTE only if clinically indicated. Scores are computed in both randomized arms but are displayed only in the intervention arm. Model weights remain frozen throughout the trial. This is single-encounter clinical decision support; providers retain full autonomy over referral and management.
Proportion of randomized patients with TTE-documented structural heart disease within 30 days
Proportion of all randomized patients with at least one clinical TTE demonstrating prespecified structural heart disease (SHD) within 30 days of the index ECG, analyzed by intention-to-treat. SHD is any of: LVEF ≤45% by Simpson biplane; moderate or greater RV dysfunction (TAPSE ≤17 mm, fractional area change ≤35%, or moderate-or-worse visual impairment); systolic pulmonary artery pressure ≥45 mm Hg; moderate or greater mitral, tricuspid, pulmonic or aortic regurgitation or aortic stenosis; septal or posterior LV wall thickness ≥1.3 cm; or moderate or large pericardial effusion. Clinical TTE reports are abstracted by two assessors blinded to arm, with disagreements adjudicated. The earliest SHD-positive clinical TTE is used. Patients without an accessible qualifying clinical TTE result by day 30, including deaths, have no documented 30-day SHD diagnosis.
Time frame: 30 days from the index ECG
Proportion of randomized patients with TTE-documented structural heart disease within 12 months
Proportion of all randomized patients with echocardiographically-confirmed SHD on clinical TTE within 12 months of the index ECG, compared between arms using the prespecified SHD composite defined in the primary outcome. Outcome assessment is blinded to randomized arm. Analyses are reported overall and by biological sex.
Time frame: 12 months from the index ECG
Time from index ECG to clinical TTE
Time from the index ECG to the first completed clinical TTE, compared between randomized arms overall and by EchoNext risk class. Time-to-event analyses include all randomized patients, censor at 12 months, and treat death as a competing event. Analyses are also reported by biological sex.
Time frame: 12 months from the index ECG
Intended TTE priority at the index ECG
Proportion of patients assigned intended TTE priority ≤30 days versus \>30 days on the post-ECG questionnaire, compared between randomized arms overall and by EchoNext risk class and biological sex.
Time frame: At baseline
Provider concordance with EchoNext TTE recommendations
Proportion of intervention-arm patients for whom a TTE is ordered within the EchoNext recommended window, overall and stratified by EchoNext risk class. High-risk patients are recommended TTE within 2 to 4 weeks (≤30 days); low-risk patients receive routine care with TTE when clinically indicated.
Time frame: At baseline
Time from index ECG to SHD diagnosis
Time from the index ECG to the first clinical TTE documenting the prespecified SHD composite defined in the primary outcome. Analyze all randomized patients, censor at 12 months, and treat death as a competing event. Compare arms overall and by EchoNext risk class and biological sex. This outcome was retained as secondary when the primary outcome was changed to documented SHD within 30 days.
Time frame: 12 months from the index ECG
Diagnostic yield of clinical TTE
Number of patients with SHD diagnosed on completed clinical TTE divided by the number of completed clinical TTEs, compared between arms overall and by EchoNext risk class. In high-risk patients, also report incremental yield as the arm difference in 30-day SHD diagnoses divided by the arm difference in completed 30-day clinical TTEs.
Time frame: 30 days from the index ECG
Net clinical TTE volume per 100 randomized patients
Number of completed clinical TTEs per 100 randomized patients and the between-arm difference, overall and by EchoNext risk class. Also report clinical TTEs per SHD diagnosis to distinguish added imaging from improved targeting.
Time frame: 30 days from the index ECG
Documented management change after SHD diagnosis
Proportion of patients diagnosed with SHD who have a documented specialty referral, initiation or uptitration of guideline-directed therapy, structural intervention, or right heart catheterization within 12 months. Compare randomized arms overall and by biological sex.
Time frame: 12 months from the index ECG
SHD diagnosis in low-risk patients by randomized arm
Proportion of low-risk randomized patients with clinical TTE-documented SHD within 30 days, compared between arms as a safety and falsification outcome. A low-risk deficit may indicate under-diagnosis from deprioritization; an excess is interpreted as contamination or chance.
Time frame: 30 days from the index ECG
EchoNext diagnostic performance and algorithmic fairness
Sensitivity, specificity, positive and negative predictive values of EchoNext against clinical TTE-confirmed SHD within 30 days in the control arm; intervention-arm estimates are reported separately. Assess equalized odds, calibration slope and predictive parity across age (\<65, 65-84, ≥85 years) and biological sex subgroups. Trial model weights remain frozen; subgroup recalibration is reported as a secondary analysis without changing the deployed model.
Time frame: 30 days from the index ECG
Pre-display TTE intent versus final referral decision
Compare the clinician's TTE-ordering intent and intended priority recorded before any AI result is displayed with the final post-ECG referral decision and intended priority (≤30 versus \>30 days), by randomized arm and EchoNext risk class.
Time frame: At baseline
Reduced left ventricular ejection fraction diagnosed on clinical TTE
Proportion of randomized patients with LVEF ≤45% measured by Simpson biplane on clinical TTE, reported separately by arm and biological sex as a component of the prespecified SHD composite.
Time frame: 30 days from the index ECG
Reduced right ventricular systolic function diagnosed on clinical TTE
Proportion of randomized patients with moderate or greater RV dysfunction: TAPSE ≤17 mm, fractional area change ≤35%, or moderate-or-worse visual impairment. Report separately by arm and biological sex.
Time frame: 30 days from the index ECG
Pulmonary hypertension diagnosed on clinical TTE
Proportion of randomized patients with systolic pulmonary artery pressure ≥45 mm Hg on clinical TTE, reported separately by arm and biological sex as a component of the SHD composite.
Time frame: 30 days from the index ECG
Valvular heart disease diagnosed on clinical TTE
Proportion of randomized patients with moderate or greater mitral, tricuspid, pulmonic or aortic regurgitation, or aortic stenosis, on clinical TTE. Each valve lesion is reported separately by arm and biological sex as a component of the prespecified SHD composite.
Time frame: 30 days from the index ECG
Left ventricular hypertrophy diagnosed on clinical TTE
Proportion of randomized patients with septal or posterior LV wall thickness ≥1.3 cm on clinical TTE, reported separately by arm and biological sex as a component of the SHD composite.
Time frame: 30 days from the index ECG
Pericardial effusion diagnosed on clinical TTE
Proportion of randomized patients with moderate or large pericardial effusion on clinical TTE, reported separately by arm and biological sex as a component of the SHD composite.
Time frame: 30 days from the index ECG
Describe the engagement of users and the overall utilization of the DeepECG platform algorithm in the clinical setting
Number of ECGs accessed per user Number of days per user with at least 1 ECG accessed using the platform over total number of days the user is in the study (i.e. has access to the platform).
Time frame: 18 months
Intended TTE priority by ambulatory care setting
Post-ECG intended TTE priority (≤30 days versus \>30 days), compared by randomized arm within outpatient and ambulatory emergency department settings.
Time frame: At baseline
Review additional qualitative feedback and insights captured after reading an ECG
Narrative description put in the "other" field of the post-ECG questionnaire
Time frame: 18 months
Provider experience with EchoNext and DeepECG
End-of-study provider survey assessing usability with the System Usability Scale, trust in automation and workflow impact. Analyses are stratified by technological literacy and provider gender, with optional GENESIS-PRAXY characterization.
Time frame: At the end of the 27-month randomized main trial phase
Assess the agreement of the user with the ECG-AI algorithm's interpretations according to practice type, number of years in practice, age of user and familiarity with AI tools (based on end of study questionnaire)
Agreement (Yes/No) of the user with the ECG-AI algorithm's interpretation. Agreement is defined as the user clicking on "thumbs up" on the platform.
Time frame: 18 months
30-day SHD diagnosis by prior TTE status and biological sex
Prespecified subgroup analyses of the primary endpoint by new patient status (no prior TTE) versus surveillance patient status (most recent normal TTE at least 24 months previously), and separately by patient biological sex. Report arm-specific proportions, adjusted odds ratios and arm-by-subgroup interactions. These comparisons are hypothesis-generating.
Time frame: 30 days from the index ECG
Describe the agreement of the user with the ECG-AI algorithm's interpretation according to the diagnosis category of the ECG-AI (ischemic, arrythmia, chamber enlargement, structural heart disease, other)
Agreement (Yes/No) of the user with the ECG-AI algorithm's interpretation. Agreement is defined as the user clicking on "thumbs up" on the platform.
Time frame: 18 months
Describe the agreement of the user with the ECG-AI algorithm's interpretation in the two subgroups.
Agreement (Yes/No) of the user with the ECG-AI algorithm's interpretation Agreement is defined as the user clicking on "thumbs up" on the platform.
Time frame: 18 months
Evaluate the effect of displaying the ECHONeXT interpretation on the delay between the ECG and the TTE evaluation, by subgroups defined by the risk level of SHD (low/ intermediate/ high)
Delay between the time of the first ECG opened in the platform and the TTE calculated as: Date of TTE evaluation - Date of access of the first ECG where an ECHONeXT interpretation was available, and a user consulted the ECG
Time frame: 18 months
Evaluate the model performance after applying continual learning
We will re-train the DeepECG models using examples that were downvoted by users or that users " bookmarked " in addition to the previous ECGs that were used for training the model. Model performance will be compared using the DeLong test for the AUC and AUPRC before and after retraining the model. Users will not be exposed ot this new model.
Time frame: 18 months
Major adverse cardiovascular events
Exploratory 12-month composite of all-cause death, myocardial infarction, stroke and any hospitalization. Obtain aggregated event counts by randomized arm and jurisdiction from ICES, RAMQ/MED-ÉCHO and German statutory health-insurance claims data, following linkage within each custodian's secure environment. No patient-level records are transferred to the study team. If a jurisdiction does not release counts, report available jurisdictions.
Time frame: 12 months from the index ECG
Resource utilization and cost-effectiveness
Exploratory assessment of TTE prescriptions, completed volumes, priority mix and completion within 30 days. Economic evaluation uses a Canadian payer perspective over 12 months and reports cost per day earlier SHD diagnosis, incremental cost per additional 30-day SHD diagnosis, and 12-month budget impact. Costs include AI licensing and compute, coordinator time, TTE capacity and MACE costs; development costs are excluded from the primary ICER.
Time frame: 12 months from the index ECG
Patient experience with TTE referral and AI-supported care
Exploratory satisfaction survey in 200 referred patients (100 per arm), selected through the platform and consented before survey administration. Assess wait times, communication clarity, AI comprehension and perceived care quality using a questionnaire adapted from the Patient Satisfaction Questionnaire. Report overall and by biological sex.
Time frame: Through 12 months from the index ECG
Plan to share: No
Started Apr 2025 (actual). Primary completion is estimated for Dec 2027, 32 months after the start.
Since we started tracking on Sep 25, 2026, the dates moved once, on Oct 13, 2026: primary completion Jan 2026 → Dec 2027 and completion Jan 2027 → Dec 2027.
Earlier changes are on the record history on ClinicalTrials.gov ↗
How these figures are worked out, and which studies are left out of them.
From the registry record's own update history. This site started tracking changes on Sep 25, 2026; for anything earlier, see the record history on ClinicalTrials.gov ↗
Eligibility is decided by the study team. Share this record with your doctor or contact the team directly.
No contact was published for this record. The registry link below has the sponsor’s details.
Get an email when the registry record changes — status, dates, results — or when someone posts here.
Sign in to followQuestions and observations about this study, from anyone following it. Not medical advice, and not a channel to the study team — their contact details are on the registry record.
Sign in to join the discussion. Reading takes no account; posting does. You choose a display name, and a pseudonym is the default.
Nothing here yet. If you are running this trial, taking part in it, or weighing whether to, this is the place to say so.