ECG data classification method based on continuous deep learning

Abstract

The invention discloses an ECG data classification method based on continuous deep learning, and the method specifically comprises the steps: obtaining ECG data with the sampling rates of 128 Hz and 360 Hz, and carrying out the preprocessing of the ECG data; constructing a convolutional neural network model, and setting hyper-parameters of the model; training and testing a heart beat sample with a sampling rate of 128 Hz by using the constructed convolutional neural network model and combining an EWC method to obtain a first trained convolutional neural network model; training and testing the heart beat sample with the sampling rate of 360 Hz by utilizing the convolutional neural network model trained for the first time in combination with an EWC method to obtain an ECG data classification model; testing the ECG data classification model by using a heart beat sample with a sampling rate of 128 Hz; and classifying the ECG data to be classified by adopting the ECG data classification model to obtain a classification result. According to the method, heart beat types can be automatically, efficiently and accurately classified for the electrocardio data with two different sampling rates, the classification precision is further improved, and the generalization ability of the model is enhanced.

Description

technical field [0001] The invention relates to an ECG data classification method based on continuous deep learning, and belongs to the technical field of ECG data classification. Background technique [0002] Deep neural network is a deep architecture neural network with local feature extraction and learning capabilities built by imitating the human brain mechanism. In recent years, with the continuous development and maturity of artificial intelligence technology, new breakthroughs have been made in implementing ECG data analysis engines to automatically extract patient ECG features and realize automatic classification of ECG data using deep learning algorithms. For example, support vector machine (SVM), Gemini SVM, sequence-to-sequence deep learning method, attention-based LSTM-CNN hybrid model and other methods. Although these deep learning models can be used to accurately and objectively automatically classify heartbeat types; however, these models are currently only a...

AI Technical Summary

Problems solved by technology

Although these deep learning models can be used to accurately and objectively automatically classify heartbeat types; however, these models are currently only aimed at a sampling rate of ECG data; however, when trained with a sampling rate of ECG data When the model is used to classify another ECG data with a different sampling rate, the classification effect will be poor, the accuracy rate will be low, and the generalization ability and migration ability of the model will be weak.

[0003] The current deep learning model does not adapt well to new tasks; when the model is trained to learn new tasks, the effect is not very good; it cannot achieve a balance between learning new knowledge and protecting old knowledge; therefore, Deep learning models are currently facing the same problem: how to enhance the ability of model transfer and overcome the occurrence of catastrophic forgetting

[0004] 1) Not efficient

When classifying ECG signals, if it takes too long to analyze ECG signals, it will not show the advantages of computer analysis signals, and the longer it takes to diagnose, the more harmful it is to the patient's health, and the more hidden dangers there will be ,low efficiency

[0005] 2) Only for data with one sampling rate

[0006] 3) Existing data is less

Images