Feature extraction and classification method for automatically extracted electrocardio beats

Abstract

The present invention discloses a feature extraction and classification method for automatically extracted electrocardio beats. The feature extraction and classification method comprises the followingsteps: preprocessing electrocardio signals collected by a wearable device, automatically positioning the pre-processed electrocardio signals, intercepting 1,000 point electrocardio data, conducting nonlinear feature extraction and frequency domain statistical feature extraction of the 1,000 point electrocardio data, conducting feature fusion of the extracted nonlinear features and frequency domain statistical features and conducting classification of the fusion features by using a support vector machine optimized by a grid search method. The feature extraction and classification method avoidsa problem that clutter signals affect final classification results and avoids a problem that a single feature affects an accuracy rate of the feature extraction; and the optimized support vector machine is used as a classifier and improves feasibility of accurate classification of the electrocardio signals in small sample data. The feature extraction and classification method improves the accuracy of the electrocardio signal extraction in the wearable device and avoids a problem that a classification accuracy rate is reduced due to non-obvious electrocardio features.

Description

technical field [0001] The invention relates to the extraction of electrocardiographic beat features. In particular, it involves feature extraction and classification methods for automated extraction of ECG beats. Background technique [0002] Irregular life in modern society and greater work pressure have greatly increased the incidence of cardiovascular disease. Cardiovascular disease has always been a disease with a high incidence, and its prevention and diagnosis are the focus of people's attention. Since cardiovascular diseases are sudden and unpredictable, and early cardiovascular diseases have no obvious characteristics and auras, and wearable ECG monitoring devices can monitor ECG signals for a long time, wearable ECG monitoring The device can reduce the damage caused by sudden cardiovascular disease to a certain extent. Electrocardiogram is the most intuitive response of human heart activity, which provides a basis for the diagnosis of cardiovascular diseases. In...

AI Technical Summary

Problems solved by technology

[0004] The ECG signal is a low-frequency signal with small amplitude, low frequency, and strong noise background, and the ECG signal collected by the wearable ECG monitoring device is doped with a large amount of noise signal. Therefore, when classifying the ECG signal If the ECG segments with more obvious features are not extracted, it will lead to unsatisfactory classification results, and may even lead to misclassification, which greatly reduces the accuracy and reliability of wearable ECG monitoring equipment, making its actual usability difficult. Unable to meet daily testing needs

Images