Discrimination power of short-term heart rate variability measures for CHF assessment

Area of Science:

• Cardiology and Medical Informatics
• Biomedical Signal Processing

Background:

• Distinguishing between normal subjects and chronic heart failure (CHF) patients is crucial for timely intervention.
• Heart rate variability (HRV) analysis offers a non-invasive method for assessing cardiac autonomic function.
• Existing HRV analysis methods may lack interpretability for clinical decision-making.

Purpose of the Study:

• To evaluate the discrimination power of short-term HRV features for differentiating normal subjects from CHF patients.
• To develop and validate a classification model using HRV for CHF detection.
• To enhance the interpretability of the classification process for clinical application.

Main Methods:

• Analysis of 1914.40 hours of ECG data from 83 subjects (54 normal, 29 CHF NYHA I-III) from public databases.
• Time and frequency domain analysis to extract standard HRV features (e.g., RMSSD, Total Power, HF, LF/HF).
• Development of a Classification and Regression Tree (CART) model for subject classification, incorporating novel features (ΔAVNN, ΔLF/HF).

Main Results:

• The CART classifier using standard HRV features achieved 79.3% sensitivity and 100% specificity.
• Incorporating non-standard features (ΔAVNN, ΔLF/HF) improved classification performance to 89.7% sensitivity and 100% specificity.
• The developed CART model provides easily understandable 'if...then...' rules for classification.

Conclusions:

• Short-term HRV analysis, particularly with enhanced features, demonstrates significant potential for discriminating CHF patients.
• The CART method offers a robust and interpretable approach for data mining in medical applications.
• This methodology provides a clinically valuable tool for objective assessment and diagnosis of chronic heart failure.