Related Experiment Video
Updated: Jul 8, 2025
Lumped-Parameter and Finite Element Modeling of Heart Failure with Preserved Ejection Fraction
Published on: February 13, 2021
Investigating automated regression models for estimating left ventricular ejection fraction levels in heart failure
Sona M Al Younis1, Leontios J Hadjileontiadis1,2, Aamna M Al Shehhi1
1Department of Biomedical Engineering, Healthcare Engineering Innovation Centre (HEIC), Khalifa University, Abu Dhabi, United Arab Emirates.
Insights
Electrocardiogram (ECG) can estimate left ventricular ejection fraction (LVEF) in heart failure (HF) patients. Machine learning models, particularly Gaussian Process Regression and Decision Trees, show high accuracy in predicting LVEF from ECG data.
Area of Science:
- Cardiology
- Biomedical Engineering
- Machine Learning
Background:
- Heart Failure (HF) affects millions globally, necessitating accurate Left Ventricular Ejection Fraction (LVEF) assessment for diagnosis and management.
- Echocardiography is standard for LVEF estimation, but Electrocardiogram (ECG) offers a simpler, cost-effective, and continuous monitoring alternative.
- Developing non-invasive methods to estimate LVEF from ECG is crucial for widespread HF monitoring.
Purpose of the Study:
- To compare the efficacy of various regression models (SVM, XGBOOST, GPR, Decision Tree) for estimating LVEF using 24-hour ECG recordings in HF patients.
- To identify the optimal machine learning models and time intervals for accurate LVEF prediction from ECG data.
- To explore the potential for an automated screening system for coronary artery disease (CAD) based on ECG-derived LVEF.
Main Methods:
- Utilized 24-hour ECG recordings from 303 HF patients across a multicentre cohort (American and Greek).
- Extracted ECG features to train regression models (SVM, XGBOOST, GPR, Decision Tree) at hourly intervals.
- Implemented hyperparameter tuning using a nested loop approach for model optimization and cross-validation.
Main Results:
- Rational Quadratic Gaussian Process Regression (GPR) and Fine Decision Tree models achieved the best LVEF estimations.
- Achieved average Root Mean Square Errors (RMSE) of 3.83% for GPR and 3.42% for Decision Trees, with correlation coefficients of 0.92 and 0.91, respectively (p<0.01).
- Identified specific circadian time periods (midnight-1 am, 8-9 am, 10-11 pm) with the lowest RMSE between actual and predicted LVEF.
Conclusions:
- Machine learning models, particularly GPR and Decision Trees, can accurately estimate LVEF from ECG data in heart failure patients.
- The findings support the development of automated systems for HF monitoring and potentially CAD screening using ECG.
- Optimizing measurement timings based on circadian rhythms can enhance the accuracy of ECG-based LVEF estimation.
Abstract:
Heart Failure (HF) significantly impacts approximately 26 million people worldwide, causing disruptions in the normal functioning of their hearts. The estimation of left ventricular ejection fraction (LVEF) plays a crucial role in the diagnosis, risk stratification, treatment selection, and monitoring of heart failure. However, achieving a definitive assessment is challenging, necessitating the use of echocardiography. Electrocardiogram (ECG) is a relatively simple, quick to obtain, provides continuous monitoring of patient's cardiac rhythm, and cost-effective procedure compared to echocardiography. In this study, we compare several regression models (support vector machine (SVM), extreme gradient boosting (XGBOOST), gaussian process regression (GPR) and decision tree) for the estimation of LVEF for three groups of HF patients at hourly intervals using 24-hour ECG recordings. Data from 303 HF patients with preserved, mid-range, or reduced LVEF were obtained from a multicentre cohort (American and Greek). ECG extracted features were used to train the different regression models in one-hour intervals. To enhance the best possible LVEF level estimations, hyperparameters tuning in nested loop approach was implemented (the outer loop divides the data into training and testing sets, while the inner loop further divides the training set into smaller sets for cross-validation). LVEF levels were best estimated using rational quadratic GPR and fine decision tree regression models with an average root mean square error (RMSE) of 3.83% and 3.42%, and correlation coefficients of 0.92 (p<0.01) and 0.91 (p<0.01), respectively. Furthermore, according to the experimental findings, the time periods of midnight-1 am, 8-9 am, and 10-11 pm demonstrated to be the lowest RMSE values between the actual and predicted LVEF levels. The findings could potentially lead to the development of an automated screening system for patients with coronary artery disease (CAD) by using the best measurement timings during their circadian cycles.
