AI-powered ECG Analysis for Deadly Arrhythmias and ICI Myocarditis (ELDORA)

Efficient Deep Learning Approaches for the Rapid and Interpretable Detection of Deadly Arrhythmias in ECG Data

ELDORA is a non-interventional observational data-science study aiming to develop and validate clinical-grade artificial intelligence tools applied to electrocardiogram (ECG) data. The project will standardize heterogeneous ECGs, create the ECGInsight harmonized database, and train interpretable models for life-threatening arrhythmia risk prediction, especially Torsades-de-Pointes/long QT syndrome and immune checkpoint inhibitor (ICI)-induced myocarditis. The project uses existing and ongoing national and international ECG cohorts with de-identified clinical metadata; AI outputs are intended for research/model development and are not used to drive patient care during the study.

Study Overview

• Status: Recruiting
• Conditions: Arrhythmia; Immune Checkpoint Inhibitor-Related Myocarditis; Long QT Syndrome

Detailed Description

"ELDORA (Efficient Deep Learning Approaches for the Rapid and Interpretable Detection of Deadly Arrhythmias in ECG Data) is an observational, non-interventional project focused on ECG-based artificial intelligence. Its overarching objective is to develop and optimize clinical-grade AI-powered tools for: (1) digitizing, standardizing and analyzing heterogeneous ECG signals, including real-life analog/paper-derived and digital recordings; and (2) supporting clinical decision research for two sudden-cardiac-arrest-prone conditions: Torsades-de-Pointes (TdP) risk prediction in established long QT syndrome, whether congenital or drug-induced, and diagnosis, prognosis and risk prediction for immune checkpoint inhibitor-induced myocarditis.

The project will consolidate diverse ECG and clinical datasets into ECGInsight, a harmonized database planned to include approximately 49 national and international ECG cohorts, around 127,000 subjects and up to about 10 million 10-second ECG equivalents. Cohorts cover a broad spectrum of health states and cardiovascular conditions, including healthy volunteers, congenital and drug-induced long QT/TdP populations, cancer patients treated with immune checkpoint inhibitors with or without myocarditis, heart transplant, diabetes, obesity and hormonal phenotyping cohorts. Data include raw ECG waveforms, automatic and expert annotations, scanned paper ECGs where applicable, demographics, clinical characteristics, laboratory results, drug exposure and hormono-metabolic assessments near the time of ECG acquisition.

Data curation will include mapping of cohort variables and clinical concepts into an ELDORA glossary, using controlled terminologies where appropriate, including ICD-10, MedDRA, OMOP and ATC for drug exposure. ECGs will be standardized using the project toolkit and integrated in a secure, GDPR-compliant infrastructure. Access is intended to be controlled and limited to approved researchers/clinicians under the project governance. The study involves no treatment allocation, no investigational medicinal product and no direct AI-driven change to patient care. Model performance will be evaluated using standard classification and regression metrics, including AUC, sensitivity, specificity, F1 score, accuracy, MAE, RMSE, R2 and Bland-Altman analyses, as appropriate to each task."

• Study Type: Observational
• Enrollment (Estimated): 127000

Contacts and Locations

• Study Contact: Name: Joe-Elie Salem, MD-PhD; Phone Number: 0033142178535; Email: joe-elie.salem@aphp.fr
• Study Contact Backup: Name: Edi Prifti, PhD; Phone Number: +33 1 48 02 55 20; Email: edi.prifti@ird.fr

Study Locations

• France
  • Paris, France, 75013
    • Recruiting
    • CIC-2503

Participation Criteria

Eligibility Criteria

• Ages Eligible for Study: Child; Adult; Older Adult
• Accepts Healthy Volunteers: Yes

• Sampling Method: Non-Probability Sample

Study Population

Subjects from existing and ongoing ECG cohorts contributing to ECGInsight, including healthy volunteers and patients with cardiovascular diseases, cancer/ICI exposure, LQT/TdP and ICI-myocarditis-relevant phenotypes.

Description

Inclusion Criteria:

• subjects included in participating existing or ongoing ECG cohorts made available to ECGInsight
• availability of ECG data (digital waveform or scanned/paper ECG suitable for digitization) and relevant clinical/demographic metadata
• data use permitted by applicable ethical, regulatory, contractual and GDPR requirements.

Exclusion Criteria:

• datasets or individual records for which required approvals, data-sharing agreements, de-identification/anonymization, or minimum ECG/metadata quality requirements are not met. No interventional study treatment is assigned.

Study Plan

How is the study designed?

Number of groups / cohorts

1

Cohorts and Interventions

Group / Cohort
A unified dataset (ECGinsight) comprising at least 10 millions ECG; A unified dataset (ECGinsight) comprising at least 10 millions ECG spanning from multiple international setting and including healthy volunteers, LQT/TdP, cancer/ICI myocarditis, heart transplant, diabetes, obesity, hormonal and patients with cardiovascular comorbidities and events

What is the study measuring?

Primary Outcome Measures

Outcome Measure	Measure Description	Time Frame
Performance of AI models for ECG-based prediction/diagnosis of life-threatening arrhythmia conditions: AUC	Model discrimination performance assessed using the Area Under the Receiver Operating Characteristic Curve (AUC) for prediction of torsade de pointes (TdP)/long QT risk and immune checkpoint inhibitor (ICI)-myocarditis diagnosis, prognosis, and risk.	Up to study completion (anticipated 48 months)

Secondary Outcome Measures

Outcome Measure	Measure Description	Time Frame
Creation and harmonization of the ECG Insight database across participating ECG cohorts	Consolidation, anonymization/de-identification, standardization and secure integration of ECG waveforms, annotations and clinical metadata from participating cohorts into ECGInsight.	Up to study completion (anticipated 48 months)
Performance of ECG digitization/standardization toolkit for heterogeneous ECG data : Accuracy	Accuracy of ECG digitization and standardization tools for conversion of analog/paper-derived and digital ECG data into analysis-ready formats, assessed by comparison with reference ECG signals.	Up to study completion (anticipated 48 months)
Performance of AI models for ECG-based prediction/diagnosis of life-threatening arrhythmia conditions: Sensitivity	Sensitivity of the prediction models for TdP/long QT risk and ICI-myocarditis diagnosis, prognosis, and risk prediction.	Up to study completion (anticipated 48 months)
Performance of AI models for ECG-based prediction/diagnosis of life-threatening arrhythmia conditions: Specificity	Specificity of the prediction models for TdP/long QT risk and ICI-myocarditis diagnosis, prognosis, and risk prediction.	Up to study completion (anticipated 48 months
Performance of AI models for ECG-based prediction/diagnosis of life-threatening arrhythmia conditions: F1 Score	F1 score of the prediction models for TdP/long QT risk and ICI-myocarditis diagnosis, prognosis, and risk prediction.	Up to study completion (anticipated 48 months)
Performance of AI models for ECG-based prediction/diagnosis of life-threatening arrhythmia conditions: Accuracy	Accuracy of the prediction models for TdP/long QT risk and ICI-myocarditis diagnosis, prognosis, and risk prediction.	Up to study completion (anticipated 48 months)
Performance of AI models for ECG-based prediction/diagnosis of life-threatening arrhythmia conditions: Regression / Agreement metrics	Regression / Agreement metrics of the prediction models for TdP/long QT risk and ICI-myocarditis diagnosis, prognosis, and risk prediction.	Up to study completion (anticipated 48 months)

Collaborators and Investigators

• Sponsor: Groupe Hospitalier Pitie-Salpetriere
• Collaborators: Assistance Publique - Hôpitaux de Paris; Institut National de la Santé Et de la Recherche Médicale, France; University of California, San Francisco; Vanderbilt University Medical Center; University Hospital, Bordeaux; Institut de Recherche pour le Developpement; Banook Group

Publications and helpful links

General Publications

• Prifti E, Fall A, Davogustto G, Pulini A, Denjoy I, Funck-Brentano C, Khan Y, Durand-Salmon A, Badilini F, Wells QS, Leenhardt A, Zucker JD, Roden DM, Extramiana F, Salem JE. Deep learning analysis of electrocardiogram for risk prediction of drug-induced arrhythmias and diagnosis of long QT syndrome. Eur Heart J. 2021 Oct 7;42(38):3948-3961. doi: 10.1093/eurheartj/ehab588.

Study record dates

Study Major Dates

• Study Start (Actual): January 1, 2026
• Primary Completion (Estimated): December 31, 2029
• Study Completion (Estimated): December 31, 2029

Study Registration Dates

• First Submitted: June 2, 2026
• First Submitted That Met QC Criteria: June 8, 2026
• First Posted (Actual): June 12, 2026

Study Record Updates

• Last Update Posted (Actual): June 12, 2026
• Last Update Submitted That Met QC Criteria: June 8, 2026
• Last Verified: May 1, 2026

More Information

Terms related to this study

• Keywords: AI; ECG; Immune Checkpoint Inhibitors-Related Myocarditis; long QT
• Additional Relevant MeSH Terms: Cardiac Conduction System Disease; Cardiovascular Diseases; Pathologic Processes; Heart Diseases; Congenital Abnormalities; Cardiovascular Abnormalities; Heart Defects, Congenital; Congenital, Hereditary, and Neonatal Diseases and Abnormalities; Pathological Conditions, Signs and Symptoms; Arrhythmias, Cardiac; Long QT Syndrome
• Other Study ID Numbers: CIC2503-26-05

Plan for Individual participant data (IPD)

• Plan to Share Individual Participant Data (IPD)?: NO

Drug and device information, study documents

• Studies a U.S. FDA-regulated drug product: No
• Studies a U.S. FDA-regulated device product: No