a Foundational Model for Cardiovascular Disease Diagnosis and Prediction

Fudan University

Status: Not yet enrolling

Conditions: Valvular Heart Disease; Arrhythmias; Heart Failure; Coronary Heart Disease (CHD)

Treatments: Other: No Interventions

Study type: Observational

Funder types: Other

Identifiers

NCT06591923

CardioVerse

Details and patient eligibility

About

The goal of this observational study is to develop and evaluate the efficacy of a foundational model that integrates multimodal medical data to improve the diagnosis and prediction of cardiovascular diseases in patients aged 18 and older, including those with various heart conditions such as coronary artery disease, heart failure, and arrhythmias. The main questions it aims to answer are:

Can a multimodal data-based diagnostic model match or exceed the accuracy of traditional gold-standard methods like coronary angiography, MRI, and echocardiography? Does integrating different types of data (ECG, imaging, biochemical tests) improve diagnostic accuracy and prediction of cardiovascular disease outcomes? Researchers will compare the foundational model with traditional diagnostic methods to see if the model offers better sensitivity, specificity, and prediction accuracy across different heart disease types.

Participants will:

Provide data from past medical records, including ECG, echocardiography, cardiac MRI, and biochemical tests.

Undergo further data collection if necessary, in line with standard clinical procedures for cardiovascular disease management.

Full description

This study aims to develop and validate a foundational model that uses multimodal medical data for the diagnosis and prediction of cardiovascular diseases. By integrating data from ECG, echocardiography, cardiac MRI, CTA, nuclear imaging (SPECT/PET), and biochemical tests, the model seeks to improve diagnostic accuracy and predict disease outcomes.

Study Design Study Type: Retrospective, multicenter, observational study Study Population: Adults aged 18 and older, including patients with coronary artery disease (CAD), heart failure, arrhythmias, and valvular heart disease (VHD).

Objectives Primary Objective: To create a model that improves the diagnosis and prediction of cardiovascular diseases using multimodal data.

Secondary Objective: To compare the performance of the model against traditional diagnostic methods like coronary angiography, echocardiography, and MRI.

Methodology Data from 2009 to 2023 will be collected from multiple hospitals. The model will use deep learning techniques to integrate the data for more accurate diagnosis and prediction.

The performance of the model will be compared with current gold-standard methods.

Expected Outcomes Improved diagnostic accuracy and early disease detection. Enhanced prediction of long-term outcomes, allowing for better treatment planning.

Enrollment: 1,000,000 estimated patients

Sex: All

Ages: 18+ years old

Volunteers: Accepts Healthy Volunteers

Inclusion criteria

Age ≥ 18 years: Patients who are 18 years of age or older. Time period: Patients who were treated or diagnosed between January 1, 2009, and December 31, 2023.

Complete medical records: Patients with comprehensive medical records, including ECG, echocardiography, MRI, CTA, nuclear imaging (SPECT/PET), and biochemical test results.

Cardiovascular diseases: Patients with diagnosed cardiovascular conditions, such as coronary artery disease (CAD), heart failure, arrhythmias, and valvular heart disease (VHD), as well as healthy individuals for comparison.

Willingness to participate: Patients who are able to provide informed consent or their legal representatives.

Exclusion criteria

Participation in other clinical trials: Patients who are currently participating in other clinical trials that may affect the study outcomes.

Incomplete medical records: Patients whose medical records lack essential data, such as ECG, echocardiography, MRI, CTA, nuclear imaging (SPECT/PET), or biochemical test results.

Data quality issues: Patients with records that have significant errors, inconsistencies, or incomplete data that cannot be reasonably corrected.

Ethical or legal concerns: Patients whose data cannot be used due to a lack of necessary consent or legal/ethical restrictions.