CN109620205B - Electrocardiogram data classification method and device, computer equipment and storage medium - Google Patents
Electrocardiogram data classification method and device, computer equipment and storage medium Download PDFInfo
- Publication number
- CN109620205B CN109620205B CN201811597800.0A CN201811597800A CN109620205B CN 109620205 B CN109620205 B CN 109620205B CN 201811597800 A CN201811597800 A CN 201811597800A CN 109620205 B CN109620205 B CN 109620205B
- Authority
- CN
- China
- Prior art keywords
- data
- network
- attention
- electrocardiographic
- electrocardiogram
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
Images
Classifications
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
- A61B5/00—Measuring for diagnostic purposes; Identification of persons
- A61B5/24—Detecting, measuring or recording bioelectric or biomagnetic signals of the body or parts thereof
- A61B5/316—Modalities, i.e. specific diagnostic methods
- A61B5/318—Heart-related electrical modalities, e.g. electrocardiography [ECG]
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
- A61B5/00—Measuring for diagnostic purposes; Identification of persons
- A61B5/24—Detecting, measuring or recording bioelectric or biomagnetic signals of the body or parts thereof
- A61B5/316—Modalities, i.e. specific diagnostic methods
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
- A61B5/00—Measuring for diagnostic purposes; Identification of persons
- A61B5/72—Signal processing specially adapted for physiological signals or for diagnostic purposes
- A61B5/7235—Details of waveform analysis
- A61B5/7264—Classification of physiological signals or data, e.g. using neural networks, statistical classifiers, expert systems or fuzzy systems
- A61B5/7267—Classification of physiological signals or data, e.g. using neural networks, statistical classifiers, expert systems or fuzzy systems involving training the classification device
Landscapes
- Health & Medical Sciences (AREA)
- Life Sciences & Earth Sciences (AREA)
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Animal Behavior & Ethology (AREA)
- Veterinary Medicine (AREA)
- Biophysics (AREA)
- Biomedical Technology (AREA)
- Heart & Thoracic Surgery (AREA)
- Medical Informatics (AREA)
- Molecular Biology (AREA)
- Surgery (AREA)
- Artificial Intelligence (AREA)
- General Health & Medical Sciences (AREA)
- Public Health (AREA)
- Pathology (AREA)
- Cardiology (AREA)
- Evolutionary Computation (AREA)
- Fuzzy Systems (AREA)
- Mathematical Physics (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Physiology (AREA)
- Psychiatry (AREA)
- Signal Processing (AREA)
- Measurement And Recording Of Electrical Phenomena And Electrical Characteristics Of The Living Body (AREA)
Abstract
The application relates to an electrocardiogram data classification method, an electrocardiogram data classification device, a computer device and a storage medium. The method comprises the following steps: acquiring the electrocardiogram data to be classified, and classifying the heart beats in the electrocardiogram data according to the electrocardiogram data and a preset classification model to obtain the heart beat category. By adopting the method, the classification model comprises the LSTM network and the attention network, so that the attention mechanism is introduced to determine the heart beat type, the accuracy of heart beat classification is greatly improved, and the accuracy of the heart disease type determined according to the heart beat type is greatly improved.
Description
Technical Field
The present application relates to the field of computer technologies, and in particular, to a method and an apparatus for classifying electrocardiographic data, a computer device, and a storage medium.
Background
With the rapid development of computer technology, computers have replaced the labor of people more and more, so that the work and the life of people are more and more convenient.
In the medical field, the electrocardiographic waveform is the most important information for doctors to analyze and diagnose heart diseases, and is also an important index for monitoring the physiological conditions of patients with various diseases in real time. Generally, a patient is in a hospital or a physical examination institution by using special electrocardiograph equipment to acquire electrocardiograph data by a professional doctor, and then the category of the electrocardiograph data of the patient is obtained by analyzing the whole oscillogram and information of each wave band of the electrocardiograph data. In order to improve the accuracy and efficiency of classification of electrocardiographic data, a computer-aided classification technology is applied to classification of electrocardiographic data, and the technology can classify the electrocardiographic data based on an existing neural network model, for example, classification based on a convolutional neural network.
However, the accuracy of classifying the electrocardiographic data by using the existing neural network model is not high.
Disclosure of Invention
In view of the above, it is necessary to provide an electrocardiographic data classification method, apparatus, computer device, and storage medium capable of improving the accuracy of classification.
In a first aspect, an embodiment of the present application provides an electrocardiographic data classification method, where the method includes:
acquiring electrocardiogram data to be classified;
classifying the heart beats in the electrocardiogram data according to the electrocardiogram data and a preset classification model to obtain the heart beat category, wherein the classification model comprises a Long Short-Term Memory (LSTM) network and an attention network.
In one embodiment, before classifying the heart beats in the electrocardiographic data according to the electrocardiographic data and a preset classification model, the method further includes:
and preprocessing the electrocardio data to obtain standard electrocardio data, wherein the preprocessing comprises heart beat voltage value normalization and time length clipping.
In one embodiment, the classification model includes: the method comprises the following steps that an encoding network, an attention network and a decoding network are adopted, the encoding network and the decoding network are a convolutional neural network or a cyclic neural network, and the heart beats in the electrocardiogram data are classified according to the electrocardiogram data and a preset classification model, and the method comprises the following steps:
inputting the electrocardio data into the coding network to obtain an electrocardio data characteristic matrix;
inputting the electrocardio data characteristic matrix into the attention network, and performing characteristic weighting on the electrocardio data characteristic matrix through the attention network to obtain an electrocardio data attention weighted characteristic matrix;
and inputting the electrocardio data weighted feature matrix into the decoding network to obtain a classification result corresponding to the electrocardio data.
In one embodiment, before classifying the cardiac beats in the electrocardiographic data according to the electrocardiographic data and a preset classification model, the method includes:
acquiring a plurality of initial electrocardiogram data, and preprocessing the plurality of initial electrocardiogram data to obtain a plurality of electrocardiogram training data, wherein the preprocessing comprises heart beat voltage value normalization and time length clipping;
and inputting the plurality of electrocardiogram training data into a preset initial classification model for training to obtain the classification model.
In one embodiment, the inputting the plurality of electrocardiographic training data into a preset initial classification model for training to obtain the classification model includes:
inputting a plurality of electrocardiogram training data into the coding network to obtain a plurality of electrocardiogram training data characteristic matrixes output through a hidden layer of the coding network;
inputting a plurality of the ECG training data feature matrixes into the attention network, and performing feature weighting on the ECG training data feature matrixes through the attention network to obtain a plurality of ECG training data weighted feature matrixes;
inputting a plurality of the ECG training data weighted feature matrixes into the decoding network to obtain a plurality of classification results corresponding to the ECG training data;
and determining the classification model according to the classification results corresponding to the plurality of electrocardiogram training data.
In one embodiment, after determining the classification model according to the classification result corresponding to the plurality of pieces of electrocardiographic training data, the method further includes:
inputting electrocardiographic test data into the classification model;
and determining the accuracy of the classification model according to the result output by the classification model.
In one embodiment, the method further comprises:
and if the heartbeat category is the abnormal heartbeat category, a statement or an alarm is sent out.
In a second aspect, an embodiment of the present application provides an electrocardiographic data classification apparatus, where the apparatus includes: the device comprises an acquisition module and a first processing module;
the acquisition module is used for acquiring the electrocardiogram data to be classified;
the first processing module is used for classifying heartbeats in the electrocardiogram data according to the electrocardiogram data and a preset classification model to obtain the category of the heartbeats, wherein the classification model comprises a long-term and short-term memory network and an attention network.
In a third aspect, an embodiment of the present application provides a computer device, including a memory and a processor, where the memory stores a computer program, and the processor implements the following steps when executing the computer program:
acquiring electrocardiogram data to be classified;
classifying the heart beats in the electrocardiogram data according to the electrocardiogram data and a preset classification model to obtain the heart beat category, wherein the classification model comprises a long-term and short-term memory network and an attention network.
In a fourth aspect, an embodiment of the present application provides a computer-readable storage medium, on which a computer program is stored, where the computer program, when executed by a processor, implements the following steps:
acquiring electrocardiogram data to be classified;
classifying the heart beats in the electrocardiogram data according to the electrocardiogram data and a preset classification model to obtain the heart beat category, wherein the classification model comprises a long-term and short-term memory network and an attention network.
According to the method and the device for classifying the electrocardiogram data, the computer equipment and the storage medium, the electrocardiogram data to be classified are obtained through the computer equipment, and the heart beats in the electrocardiogram data are classified according to the electrocardiogram data and the preset classification model to obtain the heart beat category. The classification model comprises an LSTM network and an attention network, so that the attention mechanism is introduced to determine the type of the heart beat, the problem that the classification accuracy is low due to the fact that specific relevant waveband information of the electrocardiogram data cannot be considered when the existing neural network model is adopted to classify the heart beat in the prior art is solved, the classification model can be combined with the relevant waveband information of the electrocardiogram data, the accuracy of classification of the heart beat in the electrocardiogram data is greatly improved, and the accuracy of the type of the heart beat determined according to the type of the heart beat is greatly improved.
Drawings
FIG. 1 is a diagram of the internal structure of a computer device in one embodiment;
FIG. 2 is a schematic flowchart of a method for classifying electrocardiographic data according to an embodiment;
FIG. 2a is a diagram illustrating a time-length clipping of a heart beat according to an embodiment;
FIG. 3 is a schematic flowchart of a method for classifying electrocardiographic data according to another embodiment;
FIG. 3a is a network structure of an LSTM subunit;
FIG. 4 is a schematic flowchart of a method for classifying electrocardiographic data according to another embodiment;
FIG. 5 is a schematic flowchart of a method for classifying electrocardiographic data according to another embodiment;
FIG. 5a is a schematic diagram of a classification model according to an embodiment;
FIG. 6 is a block diagram illustrating an exemplary embodiment of an apparatus for classifying electrocardiographic data;
FIG. 7 is a block diagram of an apparatus for classifying electrocardiographic data according to another embodiment;
fig. 8 is a schematic structural block diagram of an electrocardiographic data sorting device according to yet another embodiment.
Detailed Description
In order to make the objects, technical solutions and advantages of the present application more apparent, the present application is described in further detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative of and not restrictive on the broad application.
The electrocardiogram data classification method provided by the embodiment of the application can be applied to the computer device shown in fig. 1, and the computer device comprises a processor, a memory, a network interface, a display screen and an input device which are connected through a system bus. Wherein the processor of the computer device is configured to provide computing and control capabilities. The memory of the computer device comprises a nonvolatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program, and a database. The internal memory provides an environment for the operation of an operating system and computer programs in the non-volatile storage medium. The database of the computer device is used for storing a classification model preset in the following embodiments and electrocardiographic data to be classified, and optionally, may also store an encoding network, an attention network, and a decoding network, and the description about the classification model and each network may refer to the contents of the following method embodiments. The network interface of the computer device may be used to communicate with other external devices via a network connection. Optionally, the computer device may be a server, a desktop, a personal digital assistant, other terminal devices such as a tablet computer, a mobile phone, and the like, or a cloud or a remote server, and the embodiment of the present application does not limit the specific form of the computer device. The display screen of the computer equipment can be a liquid crystal display screen or an electronic ink display screen, and the input device of the computer equipment can be a touch layer covered on the display screen, a key, a track ball or a touch pad arranged on the shell of the computer equipment, an external keyboard, a touch pad or a mouse and the like. Of course, the input device and the display screen may not be part of the computer device, and may be external devices of the computer device.
Those skilled in the art will appreciate that the architecture shown in fig. 1 is merely a block diagram of some of the structures associated with the disclosed aspects and is not intended to limit the computing devices to which the disclosed aspects apply, as particular computing devices may include more or less components than those shown, or may combine certain components, or have a different arrangement of components.
According to the method and the device for classifying the electrocardiogram data, the computer equipment and the storage medium, the electrocardiogram data are automatically classified by the computer equipment and the neural network model combined with the attention mechanism, and the classification accuracy is improved.
The following describes the technical solutions of the present application and how to solve the above technical problems with specific examples. The following several specific embodiments may be combined with each other, and details of the same or similar concepts or processes may not be repeated in some embodiments. Embodiments of the present application will be described below with reference to the accompanying drawings.
It should be noted that the execution subject of the method embodiments described below may be an electrocardiographic data sorting apparatus, which may be implemented as part or all of the computer device in a software, hardware, or a combination of software and hardware. The following method embodiments are described by taking the execution subject as the computer device as an example.
Fig. 2 is a schematic flow chart of the electrocardiographic data classification method according to an embodiment. The embodiment relates to a specific process for automatically classifying electrocardiogram data by computer equipment. As shown in fig. 2, the method includes:
s101, obtaining the electrocardiogram data to be classified.
Specifically, the computer device may obtain electrocardiographic data to be classified, which may be dual-lead electrocardiographic data. Optionally, the computer device may receive pre-detected electrocardiographic data to be classified sent by the external device, obtain classified electrocardiographic data by reading the database, and detect electrocardiographic data in real time by other detection devices.
S102, classifying the heart beats in the electrocardiogram data according to the electrocardiogram data and a preset classification model to obtain the heart beat category, wherein the classification model comprises an LSTM network and an attention network.
Specifically, the computer device may input the acquired electrocardiographic data into a preset classification model, and the classification model processes the input electrocardiographic data, so as to output a type of heart beat in the electrocardiographic data, thereby classifying the heart beat in the electrocardiographic data.
It should be noted that the classification model includes an LSTM network and an attention network, wherein the LSTM network may be a neural network model, and the LSTM network is a model obtained by training a plurality of pieces of electrocardiographic data with heartbeat-type labels; the attention network may be a neural network model, and the attention network may include a plurality of layers of attention coefficients, which may be multiplied by hidden layer vectors of the LSTM network as weighting factors, so as to obtain a weighted result of the LSTM network.
For example, the computer device obtains the electrocardiographic data to be classified, inputs the electrocardiographic data into a preset classification model, realizes classification of heart beats of the electrocardiographic data, and outputs heart beat categories in the electrocardiographic data. Optionally, the heartbeat category may include: normal heart beat (N), ventricular Premature beat (V), isolated QRS-like artifact (I), and Left bundle branch block (L)
In this embodiment, the computer device obtains the electrocardiographic data to be classified, and classifies the cardiac beats in the electrocardiographic data according to the electrocardiographic data and a preset classification model to obtain a cardiac beat category. The classification model comprises an LSTM network and an attention network, so that the attention mechanism is introduced to determine the type of the heart beat, the problem that the classification accuracy is not high due to the fact that specific relevant waveband information of the electrocardiogram data cannot be considered when the existing neural network model is adopted to classify the heart beat in the prior art is solved, the classification model can be combined with the relevant waveband information of the electrocardiogram data, the accuracy of classification of the heart beat in the electrocardiogram data is greatly improved, and the accuracy of the type of the heart beat determined according to the type of the heart beat is greatly improved.
Optionally, on the basis of the foregoing embodiment, before step S102 in fig. 2, the method may further include: and preprocessing the electrocardio data to obtain standard electrocardio data, wherein the preprocessing comprises heart beat voltage value normalization and time length clipping. Specifically, after obtaining the electrocardiographic data to be classified, the computer device may perform preprocessing on the electrocardiographic data, where the preprocessing operation may include performing voltage value normalization operation on the electrocardiographic data, and performing time length clipping to obtain the processed standard electrocardiographic data. The standard electrocardiographic data is electrocardiographic data having a normalized voltage value and a unit for dividing the standard electrocardiographic data by the same time length. Alternatively, the voltage normalization operation described above mayTo pass through the formula
Or the formula is normalized by the variation, wherein X is the voltage amplitude of each heart beat sample in the electrocardio data, E (X) is the mean value of X, D (X) is the variance of X, and X is * Is a normalized voltage value of X. Optionally, the computer device performs time length clipping on the normalized electrocardiographic data, for example, dividing the continuous electrocardiographic data into heart beats, each heart beat has a length of N, dividing each heart beat by a unit length k, the unit length of the divided heart beats forming a matrix of k (N/k), and using the matrix of k (N/k) as the standard electrocardiographic data. Alternatively, the continuous electrocardiographic data may be divided into heartbeats, each heart beat has a length of 360, each heart beat is divided by a unit length of 10, the unit length of the divided heart beats forms a matrix of 10 × 36, and the matrix of 10 × 36 is used as the standard electrocardiographic data, which may be specifically shown in fig. 2 a. In this embodiment, the computer device obtains the standard electrocardiographic data by performing preprocessing including voltage value normalization and time length clipping on the electrocardiographic data, so that the calculation is more convenient and efficient when the standard electrocardiographic data is used for cardiac beat classification processing.
Fig. 3 is a schematic flow chart of a method for classifying electrocardiographic data according to another embodiment. The classification model includes: the Network coding and decoding method comprises a coding Network, an attention Network and a decoding Network, wherein the coding Network and the decoding Network are a Convolutional Neural Network (CNN) or a Recurrent Neural Network (RNN). The embodiment relates to a specific process of classifying the heart beats in the electrocardiogram data by computer equipment according to the electrocardiogram data and a preset classification model. As shown in fig. 3, the method includes:
s201, inputting the standard electrocardiogram data into the coding network to obtain an electrocardiogram data characteristic matrix.
It should be noted that the classification model includes: an encoding network, an attention network, and a decoding network, wherein the encoding network and the decoding network are each part of an LSTM network. Wherein, the encoding network and the decoding network can be both CNN or RNN.
Specifically, the computer device inputs the standard electrocardiographic data into an encoding network in the LSTM network, a hidden layer of the encoding network outputs an electrocardiographic data feature matrix, and the electrocardiographic data feature matrix can represent the characterization electrocardiographic data in a digitized manner. Wherein the coding network comprises a plurality of LSTM subunits, and the network structure of the LSTM subunits can be seen in fig. 3a, where X t Is the input of the LSTM subunit, which can be standard ECG data, h t σ represents a sigmoid function and Tanh is a hyperbolic tangent function, which is the output of the LSTM subunit.
S202, inputting the electrocardiogram data characteristic matrix into the attention network, and performing characteristic weighting on the electrocardiogram data characteristic matrix through the attention network to obtain an electrocardiogram data attention weighted characteristic matrix.
Specifically, the computer device inputs the electrocardiogram data feature matrix output by the encoding network into the attention network, and performs feature weighting on the electrocardiogram data feature matrix multiplied by a corresponding attention coefficient through the attention network, thereby obtaining the electrocardiogram data attention weighted feature matrix. Wherein, the attention network is in the form of a network matrix with hierarchy, and the attention network can perform the following operations: and multiplying each attention coefficient in the attention network by the vector output by each hidden layer of the coding network to obtain a weighting result, and using the weighting result as an electrocardiogram data attention weighting characteristic matrix.
It should be noted that the attention network can be expressed by the following formula:
wherein, a i In order to be aware of the power factor,
the vector output by the hidden layer of the coding network is N represents the heart beat length, k represents the unit length of heart beat cutting, and N/k represents the heart beat after cuttingThe number of fragments.
S203, inputting the electrocardio data weighting characteristic matrix into the decoding network to obtain a classification result corresponding to the electrocardio data.
Specifically, the computer device inputs the weighted feature matrix of the electrocardiographic data to a decoding network in the LSTM network, so as to output a classification result corresponding to the electrocardiographic data, thereby obtaining a cardiac beat category. Wherein, the decoding network comprises a plurality of LSTM subunits, and the network structure of the LSTM subunits can be seen in fig. 3 a.
In this embodiment, the structure of the classification model is coding network-attention network-decoding network. After preprocessing the electrocardiographic data to obtain standard electrocardiographic data, the computer device inputs the standard electrocardiographic data into the classification model, and the standard electrocardiographic data sequentially passes through the coding network, the attention network and the decoding network, so that a classification result of the electrocardiographic data, namely a category of the heart beat to which the heart beat in the electrocardiographic data belongs, is obtained.
In this embodiment, the computer device inputs the standard electrocardiographic data into the encoding network to obtain an electrocardiographic data feature matrix, inputs the electrocardiographic data feature matrix into the attention network, performs feature weighting on the electrocardiographic data feature matrix through the attention network to obtain an electrocardiographic data attention weighting feature matrix, and inputs the electrocardiographic data weighting feature matrix into the decoding network to obtain a classification result corresponding to the electrocardiographic data. By adopting the method, the computer equipment can combine an attention mechanism on the basis of the LSTM network, and the condition of low classification accuracy caused by failing to consider specific relevant waveband information of the electrocardiogram data when the existing neural network model is adopted to classify the heart beats in the prior art is avoided, so that the operation is more convenient and the efficiency is higher when standard electrocardiogram data is used for classifying the heart beats.
Fig. 4 is a schematic flowchart of a method for classifying electrocardiographic data according to yet another embodiment. The embodiment relates to a specific process of training a preset initial classification model by computer equipment according to initial electrocardiogram data to obtain the classification model. As shown in fig. 4, before S102 in fig. 2, the method may further include:
s301, acquiring a plurality of initial electrocardiogram data, and preprocessing the plurality of initial electrocardiogram data to obtain a plurality of electrocardiogram training data, wherein the preprocessing comprises cardiocapture voltage value normalization and time length clipping.
Specifically, the computer device may obtain a plurality of initial electrocardiographic data, and then preprocess the initial electrocardiographic data to obtain a plurality of electrocardiographic training data. The preprocessing may include cardioversion voltage value normalization and time length clipping, and the specific process may refer to the description of preprocessing the electrocardiographic data in the above embodiments, which is not described herein again. Optionally, the plurality of initial electrocardiographic data may be data in an acquired MIT-BIH dataset comprising 30-minute dual lead data of 48 fully labeled peak interval classes; or may be data in an acquired European ST-T data set, where the European ST-T data set includes 90 twin lead electrocardiographic data with peak class labels, or may be acquired data in other data sets or clinically acquired data, and then the acquired data is labeled to obtain initial electrocardiographic data.
S302, inputting the plurality of electrocardiogram training data into a preset initial classification model for training to obtain the classification model.
Specifically, the computer device inputs the plurality of electrocardiographic training data obtained through the preprocessing into a preset initial classification model for training, so as to obtain the classification model.
Optionally, a possible implementation manner of this step S302 may be as shown in fig. 5, and specifically may include:
s401, inputting a plurality of pieces of the electrocardiogram training data into the coding network to obtain a plurality of electrocardiogram training data characteristic matrixes output through a hidden layer of the coding network.
Specifically, the computer device may input the plurality of cardiac training data to an encoding network of the initial classification network, and output a plurality of cardiac training data feature matrices through a hidden layer of the encoding network.
S402, inputting the plurality of electrocardiogram training data feature matrixes into the attention network, and performing feature weighting on the electrocardiogram training data feature matrixes through the attention network to obtain a plurality of electrocardiogram training data weighted feature matrixes.
Specifically, the computer device inputs the feature matrix of the electrocardiographic training data output by the coding network into the attention network, and performs feature weighting on the feature matrix of the electrocardiographic training data through the attention network, that is, the feature matrix of the electrocardiographic training data is multiplied by a corresponding attention coefficient, so as to obtain a plurality of weighted feature matrices of the electrocardiographic training data.
And S403, inputting the plurality of electrocardio training data weighting characteristic matrixes into the decoding network to obtain a plurality of classification results corresponding to the electrocardio training data.
Specifically, the computer device inputs the cardiac training data weighted feature matrix output by the attention network into the decoding network, so as to obtain a classification result corresponding to a plurality of cardiac training data.
S404, determining the classification model according to the classification results corresponding to the plurality of electrocardiogram training data.
Specifically, the computer device may iteratively perform the operations of S401 to S403, and adjust parameters in the initial classification model, so that after multiple iterations, the loss functions of the classification results corresponding to the obtained multiple pieces of electrocardiographic training data satisfy a preset convergence requirement, thereby determining that the classification model is obtained by the training.
For example, the computer device may perform iterative processing operations that include: inputting a probability matrix of a classification result of the electrocardiogram training data and an original category matrix of the electrocardiogram training data into a preset loss function to obtain a result value of the loss function, and feeding the result value back to an initial classification network to adjust initial classification parameters in the classification network, wherein the loss function is used for representing the convergence degree of the distance between the probability matrix of the classification result of the electrocardiogram training data and the original category matrix of the electrocardiogram training data; judging whether the result value of the loss function meets a preset convergence condition or not; if yes, the adjusted classification parameters are usedReplacing the initial classification parameters to obtain the classification model; and if not, replacing the initial classification parameters with the adjusted classification parameters to obtain a new classification network, executing the iterative processing operation again until the result value output by the loss function meets the preset convergence condition, and replacing the original classification parameters in the initial classification network with the classification parameters corresponding to the loss function meeting the convergence condition to obtain a classification model. See in particular fig. 5 a. Alternatively, it may be that the loss value calculated by the loss function is terminated when it no longer changes during 5 consecutive rounds of training. Wherein the loss function can be formulated
Or a variation of this equation. Where N is the number of training samples input in each batch, for example, 64, y may be selected to represent the real output of the training samples,
representing the output of the classification network.
Optionally, the implementation manner may further include: and (3) randomly adding noise, such as white noise, to the electrocardio training data, so that data amplification is realized, and the trained classification model has wider applicability and stronger robustness.
In this implementation, the computer device can input a plurality of pieces of electrocardiographic training data into the encoding network to obtain a plurality of electrocardiographic training data feature matrices output through a hidden layer of the encoding network, then input the plurality of electrocardiographic training data feature matrices into the attention network, perform feature weighting on the electrocardiographic training data feature matrices through the attention network to obtain a plurality of electrocardiographic training data weighted feature matrices, input the plurality of electrocardiographic training data weighted feature matrices into the decoding network to obtain classification results corresponding to the plurality of electrocardiographic training data, and finally determine the classification model according to the classification results corresponding to the plurality of electrocardiographic training data. By adopting the method, the computer equipment can determine the classification model through a plurality of electrocardiogram training data and an initial classification network, and the attention network in the classification model is a trained attention network, and the attention coefficient of the attention network is the trained attention coefficient, so that the information characteristics of the electrocardiogram peak wave band can be enhanced when the electrocardiogram data is subjected to heartbeat classification, and the accuracy of the classification result is further improved.
For example, in the model training process, the above-mentioned electrocardiographic training data is input to the classification model in batches, each batch of electrocardiographic training data contains an artificially defined number of MIT-BIH data set heart beat sequences, for example, 64, and in each iteration training of the model: the electrocardio training data is input into the LSTM subunit of the coding network, multiplied by the attention coefficient in the attention network through the attention network to obtain input data (matrix) of the decoding network, and then input into the LSTM subunit of the decoding network to finally obtain a classification result. The computer equipment can calculate the error of the classification result and the real result through a cross entropy loss function, and the error is acted on the LSTM network and the attention network through an error back propagation mechanism to update the classification parameters. Optionally, the embodiment may adopt an Adaptive Moment Estimation (Adam) optimizer to perform iterative optimization of classification parameters, where the learning rate is 10 -4 The weight attenuation factor is 0.0005. And the method is used for updating the network weight, calculating the gradient in the error back propagation process of the network and updating the network parameters.
In this embodiment, the computer device obtains a plurality of initial electrocardiographic data, preprocesses the plurality of initial electrocardiographic data to obtain a plurality of electrocardiographic training data, and inputs the plurality of electrocardiographic training data into a preset initial classification model for training to obtain a classification model. By adopting the method, the computer equipment can be preprocessed based on a plurality of initial electrocardio data to obtain a plurality of electrocardio training data, so that the data processing is more convenient and the operation efficiency is high, then the initial classification model is trained based on a plurality of electrocardio training data to obtain the classification model, so that when the classification model is adopted to carry out heart beat classification processing on the electrocardio data, the information characteristics of the electrocardio peak wave band are enhanced through an attention mechanism, and the accuracy of the classification result is further improved.
Optionally, on the basis of the embodiment shown in fig. 5, after step S404, the method may further include: and inputting the electrocardio test data into the classification model, and determining the accuracy of the classification model according to the result output by the classification model. Specifically, the computer device may obtain electrocardiographic test data, optionally, the electrocardiographic test data may include 90 pieces of double-lead electrocardiographic data with peak class labels in an European ST-T data set, input the electrocardiographic test data to the trained classification model, and determine the accuracy of the classification model according to a classification result corresponding to the electrocardiographic test data output by the classification model. If the classification result corresponding to the output electrocardio test data is matched with the cardiac beat class marked by the electrocardio test data, the classification model is determined to be an accurate classification model, if the classification result corresponding to the output electrocardio test data is not matched with the cardiac beat class marked by the electrocardio test data, the classification model is determined to be an inaccurate classification model, and optionally, the classification model can be retrained again at the moment so as to ensure the accuracy of the classification model. And if the accuracy of the output result corresponding to the electrocardio test data cannot reach the expectation, modifying the super-parameter or the structure of the network, and repeating the whole training process. The hyper-parameters of the network can be parameters set by the artificial definition network, such as the number and size of convolution kernels.
In one embodiment, the method may further comprise: and if the heartbeat category is the abnormal heartbeat category, a statement or an alarm is sent out. Specifically, when the heart beat type determined by the computer device is a normal heart beat, the processing may not be performed, or the user may be prompted to be normal; when the heartbeat category determined by the computer device is abnormal heartbeat, for example, the heartbeat category is N or L, a statement or an alarm can be issued. Optionally, the specific form of the statement or the alarm sent by the computer device is not limited in this embodiment, and may be sending an alarm prompt sound, or by flashing an indicator light, or popping up an alarm dialog, etc.
It should be understood that although the various steps in the flow charts of fig. 2-5 are shown in order as indicated by the arrows, the steps are not necessarily performed in order as indicated by the arrows. The steps are not limited to being performed in the exact order illustrated and, unless explicitly stated herein, may be performed in other orders. Moreover, at least some of the steps in fig. 2-5 may include multiple sub-steps or multiple stages that are not necessarily performed at the same time, but may be performed at different times, and the order of performance of the sub-steps or stages is not necessarily sequential, but may be performed in turn or alternating with other steps or at least some of the sub-steps or stages of other steps.
In one embodiment, as shown in fig. 6, there is provided an electrocardiographic data classifying apparatus comprising: an acquisition module 11 and a first processing module 12.
Specifically, the obtaining module 11 is configured to obtain electrocardiographic data to be classified.
The first processing module 12 is configured to classify heartbeats in the electrocardiographic data according to the electrocardiographic data and a preset classification model to obtain the category of the heartbeats, where the classification model includes a long-term and short-term memory network and an attention network.
In one embodiment, as shown in fig. 7, the apparatus may further include a second processing module 13.
And the second processing module 13 is configured to perform preprocessing on the electrocardiograph data to obtain standard electrocardiograph data, where the preprocessing includes cardioversion voltage value normalization and time length clipping.
In one embodiment, the classification model comprises: the device comprises an encoding network, an attention network and a decoding network, wherein the encoding network and the decoding network are convolutional neural networks or cyclic neural networks. The first processing module 12 is specifically configured to input the standard electrocardiographic data into the encoding network to obtain an electrocardiographic data feature matrix; inputting the electrocardiogram data characteristic matrix into the attention network, and performing characteristic weighting on the electrocardiogram data characteristic matrix through the attention network to obtain an electrocardiogram data attention weighted characteristic matrix; and inputting the electrocardio data weighted feature matrix into the decoding network to obtain a classification result corresponding to the electrocardio data.
In one embodiment, as shown in fig. 8, the apparatus may further include a third processing module 14. The third processing module 14 is specifically configured to acquire a plurality of pieces of initial electrocardiographic training data, and preprocess the plurality of pieces of initial electrocardiographic training data to obtain a plurality of pieces of standard electrocardiographic training data, where the preprocessing includes cardioversion voltage value normalization and time length clipping; and inputting the plurality of standard electrocardio training data into a preset initial classification model for training to obtain the classification model.
In an embodiment, the third processing module 14 may be specifically configured to input a plurality of standard electrocardiographic training data into the encoding network, so as to obtain a plurality of electrocardiographic training data feature matrices output through a hidden layer of the encoding network; inputting a plurality of the ECG training data feature matrixes into the attention network, and performing feature weighting on the ECG training data feature matrixes through the attention network to obtain a plurality of ECG training data weighted feature matrixes; inputting a plurality of the ECG training data weighted feature matrixes into the decoding network to obtain a plurality of classification results corresponding to the ECG training data; and determining the classification model according to the classification result corresponding to the standard electrocardio training data.
In one embodiment, the first processing module 12 may be further configured to input the electrocardiographic test data into the classification model; and determining the accuracy of the classification model according to the result output by the classification model.
In one embodiment, the first processing module 12 may be further configured to issue a statement or an alarm when the heartbeat category is an abnormal heartbeat category.
For specific limitations of the electrocardiographic data classification device, reference may be made to the above limitations on the electrocardiographic data classification method, which is not described herein again. All modules in the electrocardiogram data classification device can be completely or partially realized by software, hardware and a combination thereof. The modules can be embedded in a hardware form or independent from a processor in the computer device, and can also be stored in a memory in the computer device in a software form, so that the processor can call and execute operations corresponding to the modules.
In one embodiment, a computer device is provided, comprising a memory, a processor, and a computer program stored on the memory and executable on the processor, the processor being configured to perform the steps of: acquiring electrocardiogram data to be classified; classifying the heart beats in the electrocardiogram data according to the electrocardiogram data and a preset classification model to obtain the heart beat category, wherein the classification model comprises a long-term and short-term memory network and an attention network.
In one embodiment, the processor, when executing the computer program, further performs the steps of: and preprocessing the electrocardio data to obtain standard electrocardio data, wherein the preprocessing comprises heart beating voltage value normalization and time length cutting.
In one embodiment, the classification model comprises: the encoding network, the attention network and the decoding network are convolutional neural networks or cyclic neural networks, and the following steps are further realized when the processor executes the computer program: inputting the standard electrocardiogram data into the coding network to obtain an electrocardiogram data characteristic matrix; inputting the electrocardiogram data characteristic matrix into the attention network, and performing characteristic weighting on the electrocardiogram data characteristic matrix through the attention network to obtain an electrocardiogram data attention weighted characteristic matrix; and inputting the electrocardio data weighted feature matrix into the decoding network to obtain a classification result corresponding to the electrocardio data.
In one embodiment, the processor, when executing the computer program, further performs the steps of: acquiring a plurality of initial electrocardiogram training data, and preprocessing the plurality of initial electrocardiogram training data to obtain a plurality of standard electrocardiogram training data, wherein the preprocessing comprises cardiocapture voltage value normalization and time length clipping; and inputting the plurality of standard electrocardio training data into a preset initial classification model for training to obtain the classification model.
In one embodiment, the processor, when executing the computer program, further performs the steps of: inputting a plurality of standard ECG training data into the coding network to obtain a plurality of ECG training data characteristic matrixes output through a hidden layer of the coding network; inputting a plurality of the ECG training data feature matrixes into the attention network, and performing feature weighting on the ECG training data feature matrixes through the attention network to obtain a plurality of ECG training data weighted feature matrixes; inputting a plurality of the ECG training data weighted feature matrixes into the decoding network to obtain a plurality of classification results corresponding to the ECG training data; and determining the classification model according to the classification result corresponding to the standard electrocardio training data.
In one embodiment, the processor when executing the computer program further performs the steps of: inputting electrocardiographic test data into the classification model; and determining the accuracy of the classification model according to the result output by the classification model.
In one embodiment, the processor, when executing the computer program, further performs the steps of: and if the heartbeat category is the abnormal heartbeat category, a statement or an alarm is sent out.
It should be clear that, in the embodiments of the present application, the process of executing the computer program by the processor is consistent with the process of executing the steps in the above method, and specific reference may be made to the description above.
In one embodiment, a computer-readable storage medium is provided, having a computer program stored thereon, which when executed by a processor, performs the steps of: acquiring electrocardiogram data to be classified; classifying the heart beats in the electrocardiogram data according to the electrocardiogram data and a preset classification model to obtain the heart beat category, wherein the classification model comprises a long-term and short-term memory network and an attention network.
In one embodiment, the computer program when executed by the processor further performs the steps of: and preprocessing the electrocardio data to obtain standard electrocardio data, wherein the preprocessing comprises heart beat voltage value normalization and time length clipping.
In one embodiment, the classification model comprises: an encoding network, an attention network and a decoding network, the encoding network and the decoding network being a convolutional neural network or a cyclic neural network, the computer program when executed by a processor further implementing the steps of: inputting the standard electrocardiogram data into the coding network to obtain an electrocardiogram data characteristic matrix; inputting the electrocardiogram data characteristic matrix into the attention network, and performing characteristic weighting on the electrocardiogram data characteristic matrix through the attention network to obtain an electrocardiogram data attention weighted characteristic matrix; and inputting the electrocardio data weighted feature matrix into the decoding network to obtain a classification result corresponding to the electrocardio data.
In one embodiment, the computer program when executed by the processor further performs the steps of: acquiring a plurality of initial electrocardiogram training data, and preprocessing the plurality of initial electrocardiogram training data to obtain a plurality of standard electrocardiogram training data, wherein the preprocessing comprises cardiocapture voltage value normalization and time length clipping; and inputting the plurality of standard electrocardio training data into a preset initial classification model for training to obtain the classification model.
In one embodiment, the computer program when executed by the processor further performs the steps of: inputting a plurality of standard ECG training data into the coding network to obtain a plurality of ECG training data characteristic matrixes output through a hidden layer of the coding network; inputting a plurality of the electrocardiogram training data feature matrixes into the attention network, and performing feature weighting on the electrocardiogram training data feature matrixes through the attention network to obtain a plurality of electrocardiogram training data weighted feature matrixes; inputting a plurality of the ECG training data weighted feature matrixes into the decoding network to obtain a plurality of classification results corresponding to the ECG training data; and determining the classification model according to the classification result corresponding to the standard electrocardio training data.
In one embodiment, the computer program when executed by the processor further performs the steps of: inputting electrocardiographic test data into the classification model; and determining the accuracy of the classification model according to the result output by the classification model.
In one embodiment, the computer program when executed by the processor further performs the steps of: and if the heartbeat category is the abnormal heartbeat category, a statement or an alarm is sent out.
It should be clear that, in the embodiments of the present application, the process of executing the computer program by the processor is consistent with the process of executing the steps in the above method, and specific reference may be made to the description above.
It will be understood by those skilled in the art that all or part of the processes of the methods of the embodiments described above can be implemented by hardware instructions of a computer program, which can be stored in a non-volatile computer-readable storage medium, and when executed, can include the processes of the embodiments of the methods described above. Any reference to memory, storage, database, or other medium used in the embodiments provided herein may include non-volatile and/or volatile memory, among others. Non-volatile memory can include read-only memory (ROM), programmable ROM (PROM), electrically Programmable ROM (EPROM), electrically Erasable Programmable ROM (EEPROM), or flash memory. Volatile memory can include Random Access Memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in a variety of forms such as Static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double Data Rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous Link DRAM (SLDRAM), rambus (Rambus) direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM).
All possible combinations of the technical features in the above embodiments may not be described for the sake of brevity, but should be considered as being within the scope of the present disclosure as long as there is no contradiction between the combinations of the technical features.
The above-mentioned embodiments only express several embodiments of the present application, and the description thereof is more specific and detailed, but not construed as limiting the scope of the invention. It should be noted that, for a person skilled in the art, several variations and modifications can be made without departing from the concept of the present application, which falls within the scope of protection of the present application. Therefore, the protection scope of the present patent shall be subject to the appended claims.
Claims (10)
1. A method for classifying electrocardiographic data, the method comprising:
acquiring electrocardiogram data to be classified;
preprocessing the electrocardio data to obtain standard electrocardio data corresponding to a plurality of heartbeats, wherein the preprocessing comprises heart beat voltage value normalization and time length clipping;
inputting the standard electrocardiogram data into an encoding network to obtain an electrocardiogram data characteristic matrix through a hidden layer of the encoding network; the coding network comprises a plurality of long-short term memory artificial neural network subunits;
inputting the electrocardio data characteristic matrix into an attention network, and multiplying each layer of attention coefficients in the attention network as weight factors by electrocardio data characteristic vectors output by each hidden layer of the coding network to obtain an electrocardio data attention weighted characteristic matrix, wherein the attention network introduces an attention mechanism for enhancing the information characteristics of an electrocardio peak wave band of the standard electrocardio data;
inputting the electrocardio data attention weighting characteristic matrix into a decoding network to obtain a classification result corresponding to the electrocardio data center; the decoding network comprises a plurality of long-short term memory artificial neural network subunits.
2. The method according to claim 1, wherein the classification result corresponding to the heartbeat includes: normal heart beat, ventricular premature beat, isolated QRS-like artifact, and left bundle branch block.
3. The method of any one of claims 1 to 2, wherein prior to entering the standard electrocardiographic data into the encoding network, the method further comprises:
acquiring a plurality of initial electrocardiographic training data, and preprocessing the plurality of initial electrocardiographic training data to obtain a plurality of standard electrocardiographic training data, wherein the preprocessing comprises cardiocapture voltage value normalization and time length clipping;
inputting the standard electrocardio training data into a preset initial classification model for training to obtain the classification model; the classification model includes the encoding network, the attention network, and the decoding network.
4. The method of claim 3, wherein the pre-processing further comprises randomly adding noise processing.
5. The method according to claim 4, wherein the inputting the plurality of standard electrocardiographic training data into a preset initial classification model for training to obtain the classification model comprises:
inputting a plurality of standard ECG training data into the coding network to obtain a plurality of ECG training data characteristic matrixes output through a hidden layer of the coding network;
inputting a plurality of the ECG training data feature matrixes into the attention network, and performing feature weighting on the ECG training data feature matrixes through the attention network to obtain a plurality of ECG training data weighted feature matrixes;
inputting a plurality of the ECG training data weighted feature matrixes into the decoding network to obtain a plurality of classification results corresponding to the ECG training data;
and determining the classification model according to the classification result corresponding to the standard electrocardio training data.
6. The method according to claim 5, wherein after determining the classification model according to the classification result corresponding to the plurality of standard electrocardiographic training data, further comprising:
inputting electrocardiographic test data into the classification model;
and determining the accuracy of the classification model according to the result output by the classification model.
7. The method of claim 6, further comprising:
and if the classification result corresponding to the heartbeat is an abnormal heartbeat class, a statement or an alarm is sent out.
8. An electrocardiographic data sorting apparatus, characterized in that the apparatus comprises: the device comprises an acquisition module, a second processing module and a first processing module;
the acquisition module is used for acquiring the electrocardiogram data to be classified;
the second processing module is used for preprocessing the electrocardio data to obtain standard electrocardio data corresponding to a plurality of heartbeats, and the preprocessing comprises heart beat voltage value normalization and time length clipping;
the first processing module is used for inputting the standard electrocardiogram data into an encoding network so as to obtain an electrocardiogram data characteristic matrix through a hidden layer of the encoding network, and the encoding network comprises a plurality of long-term and short-term memory artificial neural network subunits; inputting the electrocardiogram data feature matrix into an attention network, and multiplying the electrocardiogram data feature vectors output by all hidden layers of the coding network by using each layer of attention coefficients in the attention network as weight factors to obtain an electrocardiogram data attention weighted feature matrix, wherein the attention network introduces an attention mechanism for enhancing the information features of the electrocardiogram peak wave band of the standard electrocardiogram data; inputting the electrocardio data attention weighting characteristic matrix into a decoding network to obtain a classification result corresponding to the electrocardio data center; the decoding network comprises a plurality of long-short term memory artificial neural network subunits.
9. A computer device comprising a memory and a processor, the memory storing a computer program, wherein the processor when executing the computer program performs the steps of the method according to any of claims 1 to 7.
10. A computer-readable storage medium, on which a computer program is stored, which, when being executed by a processor, carries out the steps of the method of any one of claims 1 to 7.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811597800.0A CN109620205B (en) | 2018-12-26 | 2018-12-26 | Electrocardiogram data classification method and device, computer equipment and storage medium |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811597800.0A CN109620205B (en) | 2018-12-26 | 2018-12-26 | Electrocardiogram data classification method and device, computer equipment and storage medium |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN109620205A CN109620205A (en) | 2019-04-16 |
| CN109620205B true CN109620205B (en) | 2022-10-28 |
Family
ID=66077743
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201811597800.0A Active CN109620205B (en) | 2018-12-26 | 2018-12-26 | Electrocardiogram data classification method and device, computer equipment and storage medium |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109620205B (en) |
Families Citing this family (18)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110090016B (en) * | 2019-04-28 | 2021-06-25 | 心医国际数字医疗系统(大连)有限公司 | Method and system for positioning R wave position and R wave automatic detection method using LSTM neural network |
| CN112116926A (en) * | 2019-06-19 | 2020-12-22 | 北京猎户星空科技有限公司 | Audio data processing method and device and model training method and device |
| CN110288028B (en) * | 2019-06-27 | 2021-11-02 | 北京邮电大学 | Electrocardio detection method, system, equipment and computer readable storage medium |
| CN110547792B (en) * | 2019-07-31 | 2022-02-15 | 平安科技（深圳）有限公司 | Atrial fibrillation detection method and device, computer equipment and storage medium |
| CN110393519B (en) * | 2019-08-19 | 2022-06-24 | 广州视源电子科技股份有限公司 | Electrocardiosignal analysis method and device, storage medium and processor |
| CN110890155B (en) * | 2019-11-25 | 2022-10-28 | 中国科学技术大学 | Multi-class arrhythmia detection method based on lead attention mechanism |
| CN111134662B (en) * | 2020-02-17 | 2021-04-16 | 武汉大学 | Electrocardio abnormal signal identification method and device based on transfer learning and confidence degree selection |
| CN111297350B (en) * | 2020-02-27 | 2021-08-31 | 福州大学 | Three-heart beat multi-model comprehensive decision-making electrocardiogram feature classification method integrating source end influence |
| CN111523502B (en) * | 2020-04-30 | 2023-06-20 | 上海交通大学医学院附属瑞金医院 | Electrocardiogram image processing method, equipment, medium and electrocardiograph |
| CN111759304B (en) * | 2020-07-01 | 2022-09-30 | 杭州脉流科技有限公司 | Electrocardiogram abnormity identification method and device, computer equipment and storage medium |
| CN111956207B (en) * | 2020-08-19 | 2024-02-20 | 广州视源电子科技股份有限公司 | Electrocardiogram recording labeling method, device, equipment and storage medium |
| CN111920405A (en) * | 2020-09-15 | 2020-11-13 | 齐鲁工业大学 | Atrial fibrillation signal identification system and method |
| CN112043260B (en) * | 2020-09-16 | 2022-11-15 | 杭州师范大学 | Electrocardiogram classification method based on local mode transformation |
| CN112883803B (en) * | 2021-01-20 | 2023-09-01 | 武汉中旗生物医疗电子有限公司 | Electrocardiogram signal classification method, device and storage medium based on deep learning |
| CN112932433B (en) * | 2021-02-05 | 2022-11-08 | 青岛大学 | Arrhythmia automatic classification method of electrocardiogram |
| CN113012807B (en) * | 2021-03-04 | 2023-06-09 | 武汉中旗生物医疗电子有限公司 | ECG signal marking method, system, device and computer readable storage medium |
| CN113876354B (en) * | 2021-09-30 | 2023-11-21 | 深圳信息职业技术学院 | Fetal heart rate signal processing method and device, electronic equipment and storage medium |
| CN117796817B (en) * | 2024-02-29 | 2024-08-02 | 山东大学齐鲁医院 | Method and system for rapid detection and early warning of acute myocardial infarction |
Family Cites Families (12)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US10524123B2 (en) * | 2016-03-30 | 2019-12-31 | Zoll Medical Corporation | Establishing secure communication at an emergency care scene |
| KR20180001367A (en) * | 2016-06-27 | 2018-01-04 | 현대자동차주식회사 | Apparatus and Method for detecting state of driver based on biometric signals of driver |
| CN107463609B (en) * | 2017-06-27 | 2020-06-19 | 浙江大学 | Method for solving video question-answering by using layered space-time attention codec network mechanism |
| CN107890348B (en) * | 2017-11-21 | 2018-12-25 | 郑州大学 | One kind is extracted based on the automation of deep approach of learning electrocardio tempo characteristic and classification method |
| CN108804495B (en) * | 2018-04-02 | 2021-10-22 | 华南理工大学 | Automatic text summarization method based on enhanced semantics |
| CN108852339A (en) * | 2018-05-09 | 2018-11-23 | 广东工业大学 | A kind of electrocardiogram (ECG) data analytical equipment, method and system |
| CN108766557B (en) * | 2018-05-12 | 2021-07-20 | 鲁东大学 | Automatic arrhythmia analysis method based on channel signal fusion neural network |
| CN108717439A (en) * | 2018-05-16 | 2018-10-30 | 哈尔滨理工大学 | A kind of Chinese Text Categorization merged based on attention mechanism and characteristic strengthening |
| CN108926338B (en) * | 2018-05-31 | 2019-06-18 | 中南民族大学 | Heart rate prediction technique and device based on deep learning |
| CN108875807B (en) * | 2018-05-31 | 2022-05-27 | 陕西师范大学 | Image description method based on multiple attention and multiple scales |
| CN108805087B (en) * | 2018-06-14 | 2021-06-15 | 南京云思创智信息科技有限公司 | Time sequence semantic fusion association judgment subsystem based on multi-modal emotion recognition system |
| CN108985443B (en) * | 2018-07-04 | 2022-03-29 | 北京旷视科技有限公司 | Action recognition method and neural network generation method and device thereof, and electronic equipment |
-
2018
- 2018-12-26 CN CN201811597800.0A patent/CN109620205B/en active Active
Also Published As
| Publication number | Publication date |
|---|---|
| CN109620205A (en) | 2019-04-16 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109620205B (en) | Electrocardiogram data classification method and device, computer equipment and storage medium | |
| US20220093216A1 (en) | Discovering novel features to use in machine learning techniques, such as machine learning techniques for diagnosing medical conditions | |
| US20220093215A1 (en) | Discovering genomes to use in machine learning techniques | |
| CN111990989A (en) | Electrocardiosignal identification method based on generation countermeasure and convolution cyclic network | |
| CN111626114B (en) | Electrocardiosignal arrhythmia classification system based on convolutional neural network | |
| Jahmunah et al. | Uncertainty quantification in DenseNet model using myocardial infarction ECG signals | |
| US20210204858A1 (en) | Automatic sensing of features within an electrocardiogram | |
| CN116782829A (en) | Method and apparatus for converting bioelectric signal data into numerical vector and method and apparatus for analyzing disease using the same | |
| CN111657925A (en) | Electrocardiosignal classification method, system, terminal and storage medium based on machine learning | |
| CN110801218B (en) | Electrocardiogram data processing method and device, electronic equipment and computer readable medium | |
| CN112270212A (en) | Method and device for generating heartbeat label data sequence based on multi-lead electrocardiosignals | |
| CN111759304B (en) | Electrocardiogram abnormity identification method and device, computer equipment and storage medium | |
| WO2023073092A1 (en) | Managing a model trained using a machine learning process | |
| CN115778403A (en) | Electrocardiogram analysis method, electrocardiogram analysis device, electronic equipment and storage medium | |
| US20240206821A1 (en) | Apparatus and a method for predicting a physiological indicator | |
| Lu et al. | An end-to-end model for ECG signals classification based on residual attention network | |
| Kurian et al. | Prediction of risk in cardiovascular disease using machine learning algorithms | |
| Sonia et al. | Cardiac abnormalities from 12‐Lead ECG signals prediction based on deep convolutional neural network optimized with nomadic people optimization algorithm | |
| KR102560516B1 (en) | An apparatus and method for predicting heart disease based on deep learning models using ECG data and body information | |
| CN110992312A (en) | Medical image processing method, device, storage medium and computer equipment | |
| Celebiler | heart disease diagnosis using neural networks on electrocardiogram datasets | |
| CN112397179A (en) | Medical electrocardiogram intelligent analysis system and analysis method | |
| US20240363247A1 (en) | Method and an apparatus for detecting a level of cardiovascular disease | |
| CN214505007U (en) | Medical electrocardiogram intelligent analysis system | |
| Kurniawan et al. | Arrhythmia Analysis in the Long-term Electrocardiogram Monitoring System. |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |