Deep Learning for Intelligent Identification of Arrhythmias

Not yet recruiting

• Conditions: Arrhythmia
• Interventions: Other: Observational

• Registration Number: NCT05967546
• Lead Sponsor: First Affiliated Hospital Xi'an Jiaotong University

Brief Summary

This study aims to design and train a deep learning model for the diagnosis of different arrhythmias.

Detailed Description

This study aims to retrospectively and prospectively collect routine clinical data such as electrocardiograms from patients with arrhythmias who meet the inclusion and exclusion criteria. Then we will design and train a deep learning model to analyse the electrocardiographic features of the arrhythmias, and identify the types of arrhythmias and evaluate the value of the model for the diagnosis of different arrhythmias.

Study Timeline

• Study First Submitted: 2023-07-06
• Study First Submitted QC: 2023-07-21
• Study First Posted: 2023-08-01
• Status Verified: 2024-04
• Last Update Submitted: 2024-04-02
• Last Update Posted: 2024-04-04
• Study Start: 2024-12-30
• Primary Completion: 2028-08-31
• Study Completion: 2028-12-31

Recruitment & Eligibility

• Status: NOT_YET_RECRUITING
• Sex: All
• Target Recruitment: 4000

Inclusion Criteria

• For retrospective study: 1.Patients with arrhythmia diagnosed by routine surface 12-lead electrocardiogram or Holter; 2.The type of arrhythmia is diagnosed by intracardiac electrophysiological examination.
• For prospective study: 1.Patients with arrhythmia diagnosed by routine surface 12-lead electrocardiogram or Holter; 2.Intracardiac electrophysiological examination is planned.

Exclusion Criteria

• Lack of routine surface 12-lead electrocardiogram or holter data;
• Lack of intracardiac electrophysiological examination;
• Patients refused to sign informed consent and refused to participate in the study.

Study & Design

• Study Type: OBSERVATIONAL
• Study Design: Not specified

Arm && Interventions

Group	Intervention	Description
Experimental Group	Observational	ECG data and clinical data from this group of arrhythmia patients will be used to build a deep learning model.

Primary Outcome Measures

Name	Time	Method
A deep learning model designed to intelligently identify the types of arrhythmia.	1 day after the enrollment.	The model is trained on the training set, the best model and hyperparameters are selected through the verification set, and finally the model results are tested on the test set.

Secondary Outcome Measures

Name	Time	Method
The sensitivity, specificity and accuracy of the deep learning model	1 day after the enrollment.	The sensitivity, specificity and accuracy of a deep learning model designed were evaluated by intracardiac electrophysiological examination results to identify patients with arrhythmia from various centers.

Trial Locations

Locations (1)

First Affiliated Hospital of Xi'an Jiantong University; 🇨🇳 Xi'an, Shaanxi, China