Identification of Blood Oxygen Saturation State Based on Analysis of Single-lead ECG Signal Features

Monitoring and identifying the occurrence of hypoxemia has positive practical significance for early detection、treatment、prognosis of the cardiovascular diseases. The development of wearable devices makes electrocardiogram(ECG) collection more convenient. At the same time, there is a certain correlation between ECG and blood oxygen saturation. In view of this, a hypoxemia recognition model was established based on a 1-minute single channel electrocardiogram to achieve real-time monitoring of hypoxemia.Using electrocardiogram (ECG) signals segmented by sliding windows, a random forest model is developed here to identify the hypoxemia with the following features.We also made a SHAP explanatory analysis of the established model, and obtained the cutoff values of the important parameters of the model, which provides a reference for doctors in evaluating the risk of hypoxemia.The accuracy and generalization performance of this work is highlighted at the end of this article.