Dynamic electrocardiogram heart beat classification method based on gradient boosting decision tree

A dynamic electrocardiogram and classification method technology, applied in medical science, sensors, diagnostic recording/measurement, etc., can solve the problem that the classification method cannot adapt to the diversity of ECG signal forms, and achieve the effect of avoiding influence

CN109303559AActive Publication Date: 2019-02-05杭州质子科技有限公司
6 Cites 13 Cited by

Patent Information

Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Publication Date
2019-02-05

Smart Images

  • Figure 1
    Figure 1
  • Figure 2
    Figure 2
Patent Text Reader

Abstract

The invention relates to a dynamic electrocardiogram heart beat classification method based on a gradient boosting decision tree. The method comprises the steps that in actual dynamic electrocardiogram, classification is conducted on single heart beats in an electrocardiogram signal according to whether or not arrhythmia exists and the types of arrhythmia, specific classification categories comprise normal heart beats, supraventricular ectopic beat heart beats, ventricular ectopic beats, ventricular beat and normal beat fusion heart beats and pacemaker heart beats; the method comprises the following steps that 1, training data is obtained; 2, heart beat interception and feature extraction are conducted; 3, feature selection and classification model training are conducted; 4, classificationmodel application is conducted, wherein a tree-model-based feature selection method is adopted in step 3 to select features, and the classification model is trained through a gradient boosting decision tree classification method. The method is suitable for arrhythmia classification training of dynamic electrocardiogram and classification identification of different types of heart beats, and a doctor can be assisted in accurately reading and analyzing the electrocardiogram.
Need to check novelty before this filing date? Find Prior Art

Description

technical field

[0001] The invention relates to the technical field of automatic auxiliary detection of a dynamic electrocardiogram, in particular to a method for classifying heartbeats of a dynamic electrocardiogram based on a gradient lifting decision tree. Background technique

[0002] With the continuous acceleration of the pace of human life, heart disease has become an important disease that threatens human health, and most heart patients are accompanied by arrhythmia. Therefore, accurate detection and diagnosis of arrhythmia is very important for the prevention, monitoring, treatment and assistance It is of great significance for doctors to diagnose and improve the efficiency of doctors to read ECG.

[0003] There are many kinds of arrhythmia, supraventricular ectopic beat, ventricular ectopic beat, fusion of ventricular beat and normal heartbeat, and pacemaker heartbeat are not only common in heart disease patients, but also supraventricular ectopic beat, ventricular...

Examples

Embodiment Construction

[0025] The present invention will be further described below in conjunction with the accompanying drawings.

[0026] refer to figure 1 , a dynamic electrocardiogram beat classification method based on a gradient boosting decision tree. This method obtains five types of heartbeat data from the existing heartbeat-marked database, intercepts the heartbeat and obtains the multidimensional features of each heartbeat. After feature selection, training and classification model, and finally output the classification results of the test data according to the classification model.

[0027] In this embodiment, mainly aiming at the problem of automatic recognition of cardiac arrhythmia in dynamic electrocardiogram, a kind of dynamic electrocardiogram cardiac beat classification method based on gradient lifting decision tree is provided, including the following steps:

[0028] (1) Obtain training data: Select the ECG signal data from the ECG signal database with existing heartbeat type la...