Linear classification of healthy people and patients with valvular heart diseases based on heart rate variability indices derived from electrocardiograms

Heart rate variability (HRV) is an important marker in various cardiovascular and non-cardiovascular conditions. This study aimed to evaluate the effectiveness of three linear models (Logistic Regression, Ridge Regression, Support Vector Machine) in distinguishing between healthy individuals and those with valvular heart diseases (VHD) using time domain and frequency domain HRV indices derived from electrocardiographic (ECG) signals. We analyzed 59 recordings taken from two public datasets containing electrocardiographic, seismocardiographic, and gyrocardiographic signals from “Mechanocardiograms with ECG reference” and “An Open-access Database for the Evaluation of Cardio-mechanical Signals from Patients with Valvular Heart Diseases” that contain 29 and 30 recordings, respectively. HRV analysis included time and frequency domain indices and the linear models were evaluated using 5-fold stratified cross-validation. The highest sensitivity, PPV, accuracy and F1 score were observed for Logistic Regression (0.8810, 0.8819, 0.8814, 0.8812), followed by Ridge Regression (0.8805, 0.8858, 0.8814, 0.8808), and the lowest were observed for linear SVM (0.8310, 0.8318, 0.8305, 0.8305). The results showed that it is possible to distinguish healthy volunteers and patients with linear classifiers and time domain and frequency domain HRV indices obtained from ECG signals with decent performance.