Investigating automated regression models for estimating left ventricular ejection fraction levels in heart failure patients using circadian ECG features

• DOI: 10.1371/journal.pone.0295653
• Publication Date: 2023-12-11T18:30:49Z
• Authors: Sona M. Al Younis; Leontios J. Hadji...; Aamna M. Al Shehhi; Cesare Stefanini; Mohanad Alkhodari; Stergios Soulaido...; Petros Arsenos; Ioannis Doundoulakis; Konstantinos A. G...; Konstantinos Tsio...; Ahsan H. Khandoker
• Keywords: Hyperparameter

Heart Failure (HF) significantly impacts approximately 26 million people worldwide, causing disruptions in the normal functioning of their hearts. The estimation left ventricular ejection fraction (LVEF) plays a crucial role diagnosis, risk stratification, treatment selection, and monitoring heart failure. However, achieving definitive assessment is challenging, necessitating use echocardiography. Electrocardiogram (ECG) relatively simple, quick to obtain, provides continuous patient's cardiac rhythm, cost-effective procedure compared In this study, we compare several regression models (support vector machine (SVM), extreme gradient boosting (XGBOOST), gaussian process (GPR) decision tree) for LVEF three groups HF patients at hourly intervals using 24-hour ECG recordings. Data from 303 with preserved, mid-range, or reduced were obtained multicentre cohort (American Greek). extracted features used train different one-hour intervals. To enhance best possible level estimations, hyperparameters tuning nested loop approach was implemented (the outer divides data into training testing sets, while inner further set smaller sets cross-validation). levels estimated rational quadratic GPR fine tree an average root mean square error (RMSE) 3.83% 3.42%, correlation coefficients 0.92 (p<0.01) 0.91 (p<0.01), respectively. Furthermore, according experimental findings, time periods midnight-1 am, 8-9 10-11 pm demonstrated be lowest RMSE values between actual predicted levels. findings could potentially lead development automated screening system coronary artery disease (CAD) by measurement timings during circadian cycles.