Recurrence Plot-based Classification of Ischemic and Dilated Cardiomyopathy Patients.

Area of Science:

• Cardiology
• Biomedical Engineering
• Data Science

Background:

• Cardiovascular diseases significantly impact the elderly, with early cardiomyopathy prognosis remaining a challenge.
• Distinguishing between different types of cardiomyopathy, such as ischemic (ICM) and dilated (DCM), is crucial for effective treatment.
• Recurrence plot analysis offers a novel approach to characterizing complex physiological signals.

Purpose of the Study:

• To classify cardiomyopathy patients based on etiology using significant indexes from recurrence plot characterization.
• To differentiate between ischemic cardiomyopathy (ICM) and dilated cardiomyopathy (DCM) patients.
• To assess the utility of recurrence plot analysis in distinguishing cardiomyopathy patients from healthy controls.

Main Methods:

• Extracted beat-to-beat (BBI), systolic blood pressure (SBP), diastolic blood pressure (DBP), and respiratory flow (FLW) time series from 39 cardiomyopathy patients (24 ICM, 15 DCM) and 39 controls.
• Calculated recurrence plots for each signal and characterized them using 12 indexes.
• Developed support vector machine (SVM) models using the best classifiers for patient classification.

Main Results:

• The optimal SVM model achieved 92.3% accuracy, 95.8% sensitivity, and 86.6% specificity in classifying ICM versus DCM patients.
• When comparing all cardiomyopathy patients (CMP) against control subjects (CON), the best model demonstrated 85.8% accuracy, 92.3% sensitivity, and 80.1% specificity.
• Analysis suggested a more deterministic signal behavior in DCM patients compared to ICM patients.

Conclusions:

• Recurrence plot analysis, utilizing physiological time series, is a promising tool for classifying cardiomyopathy subtypes.
• The developed SVM models show high accuracy in differentiating between ICM and DCM, and between patients and controls.
• These findings support the clinical relevance of recurrence plot characterization for improving cardiomyopathy prognosis and patient stratification.