Feature extraction and heart rate estimation of SCG based on hybrid EMD decomposition

Heart rate estimation is crucial in cardiac health monitoring, especially in real-time monitoring and non-invasive detection, and can provide important support for early disease warning and personalized medicine. With the development of health monitoring technology, seismic cardiography (SCG) signals have received widespread attention for their ability to reflect cardiac motion and haemodynamic characteristics. However, SCG are often interfered by various noises in practical applications, such as motion artefacts, respiration, and body movements, which not only affect the signal quality but also may lead to inaccuracy in heart rate estimation. Therefore, the implementation of an effective decomposition strategy is an important prerequisite for improving the accuracy of heart rate estimation. In this paper, a decomposition algorithm based on the combination of CEEMDAN and EEMD is proposed for extracting information related to heart rate variability (HRV) and frequency domain features from SCG. Firstly, CEEMDAN and EEMD decompositions of SCG were performed to obtain the corresponding intrinsic modal functions (IMFs), respectively; secondly, the effective IMF components were selected through the frequency domain analysis as a means of capturing the important frequency components related to cardiac vibration and autonomic nervous system regulation; and the decomposed IMFs demonstrated clear HRV features in the time domain. Combining these two decomposition methods can effectively suppress noise and reduce modal aliasing, while comprehensively extracting SCG features related to heart rate estimation. The experimental results show that the method significantly improves the accuracy and robustness of heart rate estimation, providing strong support for feature extraction of SCG and non-invasive heart rate monitoring.