A Retrospective Cohort Study of Multimodal Electrocardiography and Chest Radiography for Prediction of Progression to Moderate or Severe Regurgitant Valvular Heart Disease
A Retrospective Cohort Study of Multimodal Electrocardiography and Chest Radiography for Prediction of Progression to Moderate or Severe Regurgitant Valvular Heart Disease
This retrospective multicenter cohort study aims to develop and validate an artificial intelligence model integrating electrocardiography (ECG) and chest radiography (CXR) to predict future progression of regurgitant valvular heart disease (rVHD), including aortic, mitral, and tricuspid regurgitation. Adult patients with ECG, CXR, and echocardiography obtained within 60 days, together with follow-up echocardiographic data, are included. The primary objective is to determine whether multimodal ECG+CXR modeling improves prediction of progression to moderate or severe regurgitation beyond ECG-only or CXR-only models. Secondary objectives include evaluation of clinical utility, risk stratification, and model interpretability. This study is intended to assess whether routinely acquired ECG and CXR can be used to support surveillance echocardiography and risk-directed management in patients at risk of future rVHD progression.
Inclusion Criteria:
Adult patients aged ≥18 years. Underwent routine electrocardiography (ECG), chest radiography (CXR), and echocardiography within 60 days.
Had subsequent follow-up echocardiographic data available for outcome ascertainment.
Patients were identified from routine clinical practice in inpatient admission or outpatient evaluation settings.
Exclusion Criteria:
Age below the adult threshold. Missing any of the required baseline examinations (ECG, CXR, or echocardiography within 60 days).
No follow-up echocardiographic data available. For future-risk evaluation in the test cohort, samples with moderate or severe regurgitation at baseline were excluded.