SCG Processing and Heart Rate Estimation Algorithm Based on Frequency-Domain Non-Uniform Sampling and First-Order Derivative

This study presents a specialized algorithm designed to estimate heart rates (HR) from seismocardiograms (SCG) by employing frequency-domain non-uniform sampling and first-order differentiation techniques. The proposed algorithm is structured around three pivotal steps: first, a uniform downsampling process is applied in the time domain to reduce the overall data load while preserving the essential characteristics of the signal. Next, non-uniform sampling is conducted in the frequency domain, allowing for a more tailored approach that focuses on the frequency components most relevant to heart rate estimation. Finally, first-order differentiation is utilized to enhance the signal, making it more suitable for accurate heart rate calculations. In this preliminary study, the effectiveness of the algorithm is rigorously evaluated using a combination of recordings sourced from the PhysioNet CEBS dataset. This dataset includes diverse physiological measurements such as electrocardiograms (ECG), respiration data and seismocardiograms, providing a comprehensive basis for testing the performance of the algorithm. The evaluation of the heart rate detection algorithm employs metrics such as the normalized mean square error, which is calculated to be 0.1343, and a correlation coefficient exceeding 0.94 (R > 0.94), indicating a robust relationship between the estimated heart rates and the reference measurements. The results of this study underscore the high accuracy and reliability of the algorithm in heart rate estimation. suggesting that it can serve as a valuable tool in both clinical and research settings for analyzing cardiovascular health through seismocardiographic data. By leveraging innovative sampling techniques and advanced signal processing methods, this algorithm holds promise for enhancing the precision of heart rate monitoring and potentially improving patient outcomes in various healthcare applications.