Triage and Recognition of Acute Aortic Dissection in Chest Pain by Electrocardiogram-Artificial Intelligence (TRACE)

A Multicenter Prospective Study to Develop and Validate an Artificial Intelligence-Based Electrocardiogram Model for the Diagnosis of Acute Type A Aortic Dissection in Patients Presenting With Chest Pain

The goal of this prospective multicenter observational study is to learn whether an artificial intelligence model based on electrocardiograms (ECGs) can help diagnose acute type A aortic dissection (TAAD) in adults who come to the emergency department with chest pain or related symptoms. The main question it aims to answer is:

Can the AI-ECG model accurately distinguish TAAD from other causes of chest pain in a real-world emergency setting? Researchers will compare the AI model's ECG-based predictions with the final diagnosis confirmed by computed tomographic angiography (CTA), which is the reference standard. Participants will undergo routine emergency ECG testing and subsequent diagnostic evaluation as part of standard care. Clinical and ECG data will be collected from five tertiary hospitals, and the model's diagnostic performance will be assessed across centers.

Study Overview

• Status: Not yet recruiting
• Conditions: Chest Pain; Aortic Dissection Type A

• Study Type: Observational
• Enrollment (Estimated): 10000

Participation Criteria

Eligibility Criteria

• Ages Eligible for Study: Adult; Older Adult
• Accepts Healthy Volunteers: No

• Sampling Method: Non-Probability Sample

Study Population

Adult male and female emergency department patients aged 18 to 80 years who present with clear chest pain or related chest/back pain symptoms at five tertiary hospitals, undergo standard 12-lead electrocardiography within 24 hours of symptom onset, and subsequently receive definitive diagnostic evaluation confirming acute type A aortic dissection or another final diagnosis.

Description

Inclusion Criteria:

• Male or female emergency department patients aged 18-80 years;
• Clear presentation of chest pain or related chest/back pain;
• Completion of standard 12-lead electrocardiography (ECG) within 24 hours after onset of chest pain;
• ECG signal quality meeting the following criteria: QRS amplitude ≥ 0.1 mV and noise proportion < 20%;
• Availability of subsequent diagnostic workup confirming whether the patient had acute type A aortic dissection (TAAD) or another definitive diagnosis.

Exclusion Criteria:

• Poor-quality ECG recordings, defined as missing leads in ≥ 3 leads or severe baseline instability;
• Indeterminate final diagnosis;
• History of prior surgery involving the aortic valve, aortic root, or ascending aorta.

Study Plan

How is the study designed?

Number of groups / cohorts

2

Cohorts and Interventions

Group / Cohort
Acute Type A Aortic Dissection (TAAD); Participants presenting with chest pain or related symptoms who are ultimately diagnosed with acute type A aortic dissection based on computed tomographic angiography (CTA) or other definitive diagnostic modalities. All participants undergo electrocardiogram (ECG) acquisition and standard clinical evaluation in the emergency setting, and their data are used to assess the diagnostic performance of the artificial intelligence-based ECG model.
Non-TAAD Chest Pain; Participants presenting with chest pain or related symptoms who are determined not to have acute type A aortic dissection after complete diagnostic evaluation. Final diagnoses may include other cardiovascular or non-cardiovascular causes of chest pain. All participants undergo electrocardiogram (ECG) acquisition and standard clinical evaluation in the emergency setting, and their data are used to assess the diagnostic performance of the artificial intelligence-based ECG model.

What is the study measuring?

Primary Outcome Measures

Outcome Measure	Measure Description	Time Frame
Diagnostic performance of the AI-based electrocardiogram model for acute type A aortic dissection	Diagnostic performance of the artificial intelligence model based on electrocardiograms for identifying acute type A aortic dissection among patients presenting with chest pain or related symptoms, using CTA-confirmed final diagnosis as the reference standard. Primary performance will be summarized by the area under the receiver operating characteristic curve (AUROC).	From emergency department presentation to completion of CTA and final diagnostic confirmation during the index visit, up to 24 hours

Secondary Outcome Measures

Outcome Measure	Measure Description	Time Frame
Sensitivity of the AI-based electrocardiogram model for acute type A aortic dissection	Sensitivity of the artificial intelligence model based on electrocardiograms for identifying acute type A aortic dissection among patients presenting with chest pain or related symptoms, using CTA-confirmed final diagnosis as the reference standard.	From emergency department presentation to completion of CTA and final diagnostic confirmation during the index visit, up to 24 hours
Specificity of the AI-based electrocardiogram model for acute type A aortic dissection	Specificity of the artificial intelligence model based on electrocardiograms for correctly identifying participants who do not have acute type A aortic dissection, using CTA-confirmed final diagnosis as the reference standard.	From emergency department presentation to completion of CTA and final diagnostic confirmation during the index visit, up to 24 hours
Positive predictive value of the AI-based electrocardiogram model for acute type A aortic dissection	Positive predictive value of the artificial intelligence model based on electrocardiograms for acute type A aortic dissection among participants classified as positive by the model, using CTA-confirmed final diagnosis as the reference standard.	From emergency department presentation to completion of CTA and final diagnostic confirmation during the index visit, up to 24 hours
Negative predictive value of the AI-based electrocardiogram model for acute type A aortic dissection	Negative predictive value of the artificial intelligence model based on electrocardiograms for acute type A aortic dissection among participants classified as negative by the model, using CTA-confirmed final diagnosis as the reference standard.	From emergency department presentation to completion of CTA and final diagnostic confirmation during the index visit, up to 24 hours
Diagnostic time from emergency department presentation to AI model output	Elapsed time from emergency department presentation to generation of the artificial intelligence model output after electrocardiogram acquisition.	At the index visit, up to 24 hours
Diagnostic time reduction associated with the AI-based electrocardiogram workflow compared with standard care	Difference in diagnostic time between the AI-based electrocardiogram workflow and the conventional diagnostic process. This outcome will be calculated as the time from emergency department presentation to final diagnostic confirmation under standard care minus the time from emergency department presentation to AI model output.	At the index visit, up to 24 hours

Collaborators and Investigators

• Sponsor: Shanghai Zhongshan Hospital
• Collaborators: Guangdong Provincial People's Hospital; Mianyang Central Hospital; Taian City Central Hospital; Yan'an Hospital of Kunming City

Study record dates

Study Major Dates

• Study Start (Estimated): April 1, 2026
• Primary Completion (Estimated): December 1, 2026
• Study Completion (Estimated): December 1, 2026

Study Registration Dates

• First Submitted: April 11, 2026
• First Submitted That Met QC Criteria: April 11, 2026
• First Posted (Actual): April 17, 2026

Study Record Updates

• Last Update Posted (Actual): April 17, 2026
• Last Update Submitted That Met QC Criteria: April 11, 2026
• Last Verified: April 1, 2026

More Information

Terms related to this study

• Additional Relevant MeSH Terms: Pain; Neurologic Manifestations; Pathological Conditions, Signs and Symptoms; Signs and Symptoms; Chest Pain
• Other Study ID Numbers: B2026-106

Drug and device information, study documents

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