Echocardiography-based machine learning algorithm for distinguishing ischemic cardiomyopathy from dilated...

Area of Science:

• Cardiology
• Medical Informatics
• Machine Learning

Background:

• Heart failure is often caused by cardiomyopathy, necessitating distinct treatments based on etiology.
• Differentiating between ischemic cardiomyopathy (ICM) and dilated cardiomyopathy (DCM) is crucial for effective patient management.
• Machine learning (ML) offers potential for analyzing complex data to predict and diagnose cardiovascular conditions.

Purpose of the Study:

• To evaluate the diagnostic performance of an ML algorithm in distinguishing ICM from DCM.
• To assess the utility of combining echocardiographic data for automated cardiomyopathy classification.
• To explore the potential of ML in improving etiological diagnosis for heart failure.

Main Methods:

• Retrospective collection of echocardiographic data from 200 DCM and 199 ICM patients.
• Data split into training and test sets using 10-fold cross-validation.
• Comparison of four ML algorithms (random forest, logistic regression, neural network, XGBoost) for classification accuracy.

Main Results:

• The XGBoost model achieved the highest diagnostic performance with an AUC of 0.934.
• XGBoost demonstrated an average sensitivity of 72% and specificity of 78% on the test set.
• External validation showed an AUC of 0.804, indicating generalizability.

Conclusions:

• Advanced ML algorithms can accurately differentiate ICM from DCM using echocardiographic parameters.
• This approach offers precision in etiological diagnosis for heart failure.
• ML-driven insights can support individualized treatment strategies for cardiomyopathy patients.