Electrocardiosignal automatic analysis method based on deep learning

Abstract

An electrocardiosignal automatic analysis method based on deep learning comprises the steps: downloading labeled electrocardiosignal data from a public data set, processing the electrocardiosignal data to obtain a data set, and dividing the data set into a training set, a verification set and a test set; constructing a deep learning model according to the DLA structure, and performing training toobtain a trained deep learning model; adjusting hyper-parameters, and selecting a model with the best classification effect on the verification set and the test set; and processing 12-lead electrocardiogram data to be classified to obtain a data set, and inputting the data of the data set into the model with the best classification effect to obtain the classification to which the electrocardiogramsignals of the electrocardiogram data belong. According to the method, low-level waveform structure features are extracted through one-dimensional convolution, shallow and deep layers are aggregated,the space and semantic features of the electrocardiosignal are obtained, morphological analysis is completed, correlation between morphologies is obtained, and the method can be applied to classification of electrocardiograms or one-dimensional time series electrocardiograms.

Description

technical field [0001] The invention relates to an automatic analysis method of electrocardiographic signals based on deep learning. Background technique [0002] The electrocardiogram is the most direct response to the electrical activity of the human heart, and it is one of the important basis for doctors to diagnose and treat heart disease. Usually, the collection and classification of electrocardiogram waveform data is carried out in hospitals or physical examination centers, which has disadvantages such as inconvenient detection and low detection frequency. In recent years, with the popularization of the Internet and mobile smart phones, it is possible to introduce portable ECG monitors and family personal ECG monitors, so the automatic identification and classification of ECG signals has high practical significance. [0003] The traditional ECG measurement classification method is mainly divided into the following steps, signal preprocessing, waveform detection, featur...

AI Technical Summary

Problems solved by technology

[0004] The defect of the traditional ECG classification algorithm is that it is necessary to design a feature extraction method to obtain useful information. The selection of features is usually through trial and error or experience, which is highly dependent on the selected features. When the designed features cannot reflect the intrinsic properties of the data , the accuracy of the classification algorithm will drop significantly

In this process, the nonlinear fitting ability of principal component analysis, wavelet transform and other methods is limited, and some information will be lost, which is prone to misclassification and false positives in practical applications.

Images