A Pulse Correlation Model for PPG Signal Quality Assessment and Key Pulse Extraction

Photoplethysmograph (PPG) signals can be utilized to measure various physiological parameters such as heart rate and oxygen saturation. Also, PPG signals are a versatile tool for monitoring cardiovascular health and assessing the risk of cardiovascular disease including blood pressure, and arterial stiffness. A PPG pulse is a periodic signal that captures the blood flow pattern during a single heartbeat cycle. However, in dynamic measurements, there are often multiple similar pulses, and the quality of the pulses may be compromised by motion artifacts, making them difficult to analyze effectively. In this paper, we propose a pulse correlation model for conducting signal quality assessment (SQA), which primarily focuses on the morphological features. Additionally, we present a key pulse extraction framework aimed at reducing pulse redundancy and obtaining a representative pulse. Experiments were conducted using a public MIMIC-II waveform database. The results demonstrated that SQA based on PPG pulse morphology is feasible, and the extracted key pulses are not influenced by abnormal pulses. By utilizing the proposed model, there is a notable reduction in transmission consumption and computational complexity related to feature engineering.