Estimating Left Ventricle Ejection Fraction Levels Using Circadian Heart Rate Variability Features and Support Vector...

Area of Science:

• Cardiology
• Biomedical Engineering
• Data Science

Background:

• Left Ventricular Ejection Fraction (LVEF) is a crucial indicator of cardiac function.
• Accurate LVEF assessment is vital for diagnosing and managing cardiovascular diseases.
• Current methods for LVEF estimation have limitations in continuous monitoring.

Purpose of the Study:

• To determine the optimal hourly fit for estimating LVEF from 24-hour ECG recordings.
• To compare the accuracy of estimated LVEF with gold-standard guidelines.
• To explore the potential of Heart Rate Variability (HRV) features for LVEF prediction.

Main Methods:

• Support Vector Regression (SVR) models were employed to estimate LVEF.
• ECG-derived HRV data from 24-hour recordings were utilized.
• A step-wise feature selection approach was implemented for precise LVEF estimation.
• Data from the Intercity Digital ECG Alliance (IDEAL) study was used, including patients with varying LVEF levels.

Main Results:

• The lowest Root Mean Square Error (RMSE) between original and estimated LVEF occurred during specific hourly intervals: 3-4 am, 5-6 am, and 6-7 pm.
• These identified intervals suggest optimal times for LVEF assessment.
• LVEF classification aligned with ACCF/AHA guidelines demonstrated improved accuracy for mid-range LVEF.

Conclusions:

• The study successfully established optimal hourly intervals for LVEF estimation from 24-hour ECGs.
• The findings suggest potential for improved intervention timing and treatment outcomes.
• This research supports the use of HRV features for predicting LVEF, potentially leading to automated classification for conditions like Coronary Artery Disease (CAD).