CN111134662A - Electrocardio abnormal signal identification method and device based on transfer learning and confidence degree selection - Google Patents
Electrocardio abnormal signal identification method and device based on transfer learning and confidence degree selection Download PDFInfo
- Publication number
- CN111134662A CN111134662A CN202010096524.0A CN202010096524A CN111134662A CN 111134662 A CN111134662 A CN 111134662A CN 202010096524 A CN202010096524 A CN 202010096524A CN 111134662 A CN111134662 A CN 111134662A
- Authority
- CN
- China
- Prior art keywords
- data
- training
- short
- abnormal
- time
- 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.)
- Granted
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
- A61B5/7203—Signal processing specially adapted for physiological signals or for diagnostic purposes for noise prevention, reduction or removal
-
- 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
-
- 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
Abstract
The invention discloses an electrocardio abnormal signal identification method based on transfer learning and confidence degree selection, which comprises the following steps: s1, denoising a large amount of short-time electrocardio data; s2, building a CNN model, and initializing parameters in the CNN model at random; s3, training a CNN by using a large number of short-time electrocardiogram data sets; s4, cutting a small amount of long-time electrocardiogram data to enable the length to be matched with the network input; and S5, performing migration training by using the cut short-time data, and selecting k short-time data with the highest confidence level in the packets as input in each training round, and S6, realizing the function of recognizing the abnormal electrocardio signals. The method for recognizing the electrocardio abnormal signals based on the transfer learning and the confidence degree selection is used, information is obtained through pre-training of the convolutional neural network, the accuracy of the model for recognizing and classifying the electrocardio abnormal signals is improved in the transfer learning and confidence degree selection mode, reference can be provided for a doctor in an auxiliary mode, misdiagnosis and missed diagnosis rates are reduced, and the workload of the doctor is reduced.
Description
Technical Field
The invention relates to the technical field of electrocardiosignal identification and classification, in particular to an electrocardio abnormal signal identification method and device based on transfer learning and confidence degree selection.
Background
Cardiovascular disease (CVD) refers to a series of diseases associated with the heart or blood vessels, also known as circulatory diseases. The following are several important facts of several world health organization statistics: cardiovascular disease remains the leading cause of death worldwide, with the number of deaths annually from cardiovascular disease exceeding any other cause of death in all death states. Within 2016, an estimated 1790 million people die of cardiovascular disease, accounting for about 31% of the total number of worldwide deaths, and about 85% of them die of heart disease and stroke. In the diagnosis of heart diseases, Electrocardiogram (ECG, EKG) is a diagnostic technique for recording the electrophysiological activity of the heart in time units through the chest cavity, capturing its electrical signals by electrodes placed on the skin and plotting them as lines. As a non-invasive recording mode, the application of the electrocardiogram is the most extensive and authoritative.
In recent years, the level of technologies such as fuzzy recognition, artificial intelligence, and neural networks has been increasing. With the development of big data and artificial intelligence, the research on the automatic electrocardiogram diagnosis algorithm and system designed based on electrocardiogram signal data has been more in recent years, but most of the results still stay in the experimental stage, and a great distance is still needed to be left for the real commercial investment. Even if the part is put into commercial use, the precision is insufficient, the disease discrimination is not specific enough, and the like, so that the help of doctors is very limited.
The inventor of the present application finds that the method of the prior art has at least the following technical problems in the process of implementing the present invention:
the PDF of a 12-lead electrocardiogram is currently the most readily available data to hospitals or physicians. The data volume is large, the length is relatively short (PDF of the 12-lead electrocardiogram is short-time lead data), and a diagnosis result can be quickly obtained after the data is input into a neural network model. However, the whole sample size of partial long-time lead data is small due to the difficulty in acquisition, the partial long-time lead data is difficult to be directly used for training a model, most of the data only have global labels, and the condition of inaccurate labels may occur when the samples are segmented and input into the model, so that the overall performance of the model is poor.
Disclosure of Invention
In view of the above, the present invention provides a method and an apparatus for identifying abnormal cardiac electrical signals based on transfer learning and confidence level selection, so as to solve or at least partially solve the technical problem of inaccurate identification result in the prior art.
In order to solve the above technical problem, a first aspect of the present invention provides an abnormal cardiac electrical signal identification method based on transfer learning and confidence level selection, including:
s1: acquiring a first quantity of short-time electrocardiogram data, and denoising the acquired first quantity of short-time electrocardiogram data;
s2: building a CNN model, and randomly initializing parameters in the CNN model;
s3: defining a loss function by taking the denoised short-time electrocardiogram data of the first quantity as a training data set, and training a CNN model by adopting a preset algorithm to obtain an original classification model;
s4: obtaining a second quantity of long-time electrocardiogram data, taking the obtained second quantity of long-time electrocardiogram data as a training sample, and cutting the training sample, wherein the cut training sample comprises a plurality of short-time electrocardiogram data, and the second quantity is far smaller than the first quantity;
s5: determining the abnormal confidence coefficient of each short-time electrocardiogram data in the cut sample according to the original classification model, selecting target short-time data according to the abnormal confidence coefficient, and performing migration training on the initial classification model to obtain a target classification model;
s6: and identifying the electrocardiogram data to be identified by using the target classification model.
In one embodiment, the denoising processing is performed on the first number of short-time electrocardiographic data acquired in S1, and includes: and denoising by adopting a method of combining an integrated empirical mode decomposition algorithm and a wavelet soft threshold.
In one embodiment, the random initialization of the parameters in S2 follows a gaussian distribution:
where μ is desired, σ is standard deviation, σ is2Is the variance.
In one embodiment, the CNN model in step S2 includes a plurality of convolutional layers, a plurality of pooling layers, and a fully connected layer.
In one embodiment, the training algorithm in S3 is a back propagation algorithm, the classification model is trained based on a stochastic gradient descent algorithm, and the calculation formula of the convolutional layer output value is as follows:
wherein x isi+m,j+nThe (i + m) th row and (j + n) th column elements representing the image; wm,nRepresents the weight of the mth row and the nth column; wbA bias term representing a filter; a isi,jThe ith row and jth column elements representing the features obtained after the convolution operation; f represents an activation function; the back propagation algorithm specifically includes:
s3.1: calculating the output value a of each neuron in a forward directionj；
S3.2: inverse computation of error term δ for each neuronjIs an error term δjLoss function E of networkdPartial derivatives of weighted input netj to neurons, i.e.
S3.3: calculating per neuron connection weight WjiIs of the formula
Where j represents the jth of the network.
In one embodiment, S5 specifically includes:
s5.1: determining the abnormal confidence coefficient of each short-time electrocardio data in the cut sample according to the original classification model;
s5.2: selecting preset short-time electrocardiogram data with the highest abnormal confidence coefficient as input of an initial classification model according to the magnitude of the abnormal confidence coefficient, and performing one round of training on the initial classification model;
s5.3: and repeatedly executing S5.1 and S5.2 until the classification model converges or a preset training round number is reached.
In one embodiment, the training samples include normal samples labeled as normal and abnormal samples labeled as abnormal, and S5.2 specifically includes:
according to the magnitude of the abnormal confidence coefficient, selecting preset short-time electrocardiogram data with the highest abnormal confidence coefficient from the abnormal samples as abnormal training data, and selecting preset short-time electrocardiogram data with the highest abnormal confidence coefficient from the normal samples as normal training data;
and taking the abnormal training data and the normal training data as the input of the initial classification model, and carrying out one round of training on the initial classification model.
Based on the same inventive concept, the second aspect of the present invention provides an abnormal cardiac electrical signal identification apparatus based on transfer learning and confidence level selection, comprising:
the short-time electrocardio-data acquisition module is used for acquiring a first amount of short-time electrocardio-data and denoising the acquired first amount of short-time electrocardio-data;
the CNN model building module is used for building a CNN model and randomly initializing parameters in the CNN model;
the CNN model training module is used for defining a loss function by taking the denoised short-time electrocardiogram data of the first quantity as a training data set and training the CNN model by adopting a preset algorithm to obtain an original classification model;
the long-time electrocardiogram data cutting module is used for acquiring a second amount of long-time electrocardiogram data, taking the acquired second amount of long-time electrocardiogram data as a training sample, cutting the training sample, wherein the cut training sample comprises a plurality of short-time electrocardiogram data, and the second amount is far smaller than the first amount;
the migration training module is used for determining the abnormal confidence coefficient of each short-time electrocardiogram data in the cut sample according to the original classification model, selecting target short-time data according to the abnormal confidence coefficient, and performing migration training on the initial classification model to obtain a target classification model;
and the identification module is used for identifying the electrocardiogram data to be identified by utilizing the target classification model.
Based on the same inventive concept, a third aspect of the present invention provides a computer-readable storage medium having stored thereon a computer program which, when executed, performs the method of the first aspect.
Based on the same inventive concept, a fourth aspect of the present invention provides a computer device comprising a memory, a processor and a computer program stored on the memory and executable on the processor, the processor implementing the method according to the first aspect when executing the program.
One or more technical solutions in the embodiments of the present application have at least one or more of the following technical effects:
the invention provides an electrocardio abnormal signal identification method based on transfer learning and confidence degree selection, which comprises the steps of firstly carrying out denoising processing on a first quantity of acquired short-time electrocardio data; training the set CNN model by using the short-time electrocardio data after denoising treatment to obtain an original classification model; by pre-training on a large amount of electrocardiogram data, the problem of insufficient prior knowledge caused by a small amount of part of electrocardiogram data is avoided, and the characteristics of various electrocardiogram data can be fully learned; then obtaining a second quantity of long-time electrocardiogram data, taking the obtained second quantity of long-time electrocardiogram data as training samples, cutting the training samples, wherein the second quantity is far smaller than the first quantity, namely the second quantity of long-time electrocardiogram data is a small quantity of data, the first quantity of short-time electrocardiogram data is a large quantity of data, then determining the abnormal confidence coefficient of each short-time electrocardiogram data in the cut samples according to the original classification model, and selecting target short-time data according to the abnormal confidence coefficient to perform migration training on the initial classification model to obtain a target classification model; the transfer learning method is used, and the pre-trained model is used for transfer learning, so that the parameters of the model are reduced, the model is more robust, the anti-overfitting effect is better, and the accuracy and generalization capability of the model are improved. And then the target classification model can be used for identifying the electrocardiogram data to be identified, thereby achieving the technical effect of improving the identification accuracy. Reliable assistance and reference is provided for medical personnel in specific applications. By repeatedly training a large amount of data and continuously optimizing the algorithm, the accuracy of the abnormal electrocardio identification and classification is improved to a certain extent, reliable assistance and reference are provided for medical personnel, and misdiagnosis and missed diagnosis rate are reduced.
Further, according to the magnitude of the abnormal confidence coefficient, selecting preset short-time electrocardiogram data with the highest abnormal confidence coefficient from the abnormal samples as abnormal training data, and selecting preset short-time electrocardiogram data with the highest abnormal confidence coefficient from the normal samples as normal training data; the abnormal training data and the normal training data are used as the input of an initial classification model, and the initial classification model is trained once, namely, the deep learning method is improved, the abnormal part in the abnormal electrocardiogram data is screened, and the interference of the normal part in the abnormal electrocardiogram data on the classification model is avoided.
Further, in order to make model prediction faster, the convolution operation is performed using one-dimensional CNN, which is easier to train, with only a few tens of Back Propagation (BP) periods.
Drawings
In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below, and it is obvious that the drawings in the following description are some embodiments of the present invention, and for those skilled in the art, other drawings can be obtained according to these drawings without creative efforts.
FIG. 1 is a schematic flow chart of an implementation of an abnormal cardiac electrical signal identification method based on transfer learning and confidence level selection according to an embodiment;
FIG. 2 is a comparison graph of the original data of the electrocardiograph signal and the de-noised data;
FIG. 3 is a schematic structural diagram of a CNN model (convolutional neural network);
FIG. 4 is a schematic diagram illustrating confidence level selection training in model transfer learning;
fig. 5 is a block diagram of an abnormal cardiac electrical signal identification apparatus based on transfer learning and confidence level selection according to an embodiment of the present invention;
FIG. 6 is a block diagram of a computer-readable storage medium according to an embodiment of the present invention;
fig. 7 is a block diagram of a computer device in an embodiment of the present invention.
Detailed Description
The invention aims to provide an electrocardio abnormal signal identification method based on transfer learning and confidence degree selection, which mainly solves the problems that the electrocardio abnormal signal identification method utilizes the electrocardio diagnosis knowledge which is learned by a neural network to transfer to an electrocardio data set for carrying out the identification and classification of the electrocardio abnormality, assists in providing reference for a doctor, reduces misdiagnosis and missed diagnosis rate and lightens the workload of the doctor; the characteristics of the electrocardiogram data are automatically learned by using a convolutional neural network, and the accuracy and generalization capability of the model are improved by using transfer learning and confidence degree selection.
In order to make the objects, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are some, but not all, embodiments of the present invention. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.
Example one
The embodiment provides an electrocardio abnormal signal identification method based on transfer learning and confidence degree selection, which comprises the following steps:
s1: and acquiring a first quantity of short-time electrocardio data, and denoising the acquired first quantity of short-time electrocardio data.
Specifically, the first amount of short-time electrocardiographic data is a large amount of short-time electrocardiographic data, such as PDF of a 12-lead electrocardiogram, and the first amount may be selected as needed, and is generally in the order of ten thousand, such as 2 ten thousand, 3 ten thousand, forty thousand, and so on. The denoising processing of the acquired first amount of short-time electrocardiographic data can adopt the existing method.
S2: and (4) building a CNN model, and randomly initializing parameters in the CNN model.
Specifically, the CNN model is a convolutional neural network model, and can be constructed by using existing neural network knowledge, and perform random initialization on parameters, where the parameters include a learning rate, the number of times of completing one forward calculation and one backward propagation calculation, and the like.
S3: and taking the first amount of short-time electrocardiogram data subjected to denoising processing as a training data set, defining a loss function, and training the CNN model by adopting a preset algorithm to obtain an original classification model.
In particular, the preset algorithm may be a back propagation algorithm.
S4: and obtaining a second quantity of long-time electrocardiogram data, taking the obtained second quantity of long-time electrocardiogram data as a training sample, and cutting the training sample, wherein the cut training sample comprises a plurality of short-time electrocardiogram data, and the second quantity is far smaller than the first quantity.
Specifically, the second number of long-time electrocardiographic data is a small number of long-time electrocardiographic data, which is much smaller than the first number, i.e., not at a data level, and the second number is several hundreds or several tens. The purpose of cutting the second amount of long-time electrocardiogram data is to make the length of the second amount of long-time electrocardiogram data adaptive to the network input, so as to obtain a data set for transfer learning.
S5: and determining the abnormal confidence coefficient of each short-time electrocardiogram data in the cut sample according to the original classification model, selecting target short-time data according to the abnormal confidence coefficient, and performing migration training on the initial classification model to obtain a target classification model.
Specifically, the original classification model gives the probabilities that the data are normal and abnormal when predicting, the probabilities are the abnormal confidence coefficients of the data, each short-time electrocardiogram data can obtain the corresponding abnormal confidence coefficient through the original classification model, and then target data are selected according to the abnormal confidence coefficients for migration training.
In each round of training process, k pieces of short-time data with the highest confidence coefficient are selected as input, a model with the highest accuracy on a verification set is stored as an optimal model, the function of recognizing the abnormal electrocardio signals is achieved, the last layer of a target model is a softmax layer for processing multi-classification tasks, Batchnormalization is carried out on each layer of convolution layer to accelerate neural network training, the sensitivity of network initialization is reduced, and the problem of gradient disappearance is relieved by using a LeakyRelu activation function.
S6: and identifying the electrocardiogram data to be identified by using the target classification model.
Specifically, this step is a specific application of the model obtained by the migration training in S5, and the electrocardiogram data to be recognized can be recognized by using this model.
Fig. 1 is a flow chart of implementing the method for identifying an abnormal cardiac electrical condition based on transfer learning and confidence level selection in a specific implementation process.
Compared with the prior art, the invention has the following advantages and beneficial technical effects:
1. the invention pre-trains a large amount of electrocardiogram data, avoids the problem of insufficient prior knowledge caused by a small amount of part of electrocardiogram data, and can fully learn the characteristics of various electrocardiogram data.
2. The invention uses the transfer learning method and the pre-trained model to perform transfer learning, reduces the parameters of the model, makes the model more robust, has better anti-overfitting effect, and improves the accuracy and generalization capability of the model.
3. The deep learning method is improved, the abnormal part in the abnormal electrocardiogram data is screened, and the interference of the normal part in the abnormal electrocardiogram data to the classification model is avoided.
4. For faster model prediction, the convolution operation is performed using one-dimensional CNN, which is easier to train, with only a few tens of Back Propagation (BP) periods.
5. The invention improves the accuracy and provides reliable assistance and reference for medical personnel. By repeatedly training a large amount of data and continuously optimizing the algorithm, the accuracy of the abnormal electrocardio identification and classification is improved to a certain extent, reliable assistance and reference are provided for medical personnel, and misdiagnosis and missed diagnosis rate are reduced.
In one embodiment, the denoising processing is performed on the first number of short-time electrocardiographic data acquired in S1, and includes: and denoising by adopting a method of combining an integrated empirical mode decomposition algorithm and a wavelet soft threshold.
In a specific implementation process, the comparison example before and after denoising is shown in fig. 2, where (a) shows a schematic diagram of original electrocardiographic data, and (a) shows a schematic diagram of denoised electrocardiographic data, and the electrocardiographic signal is a bioelectric signal collected from a body surface of a human body, and has the commonality of the bioelectric signals: weak amplitude, low frequency, large impedance, randomness and the like, most energy of electrocardiosignals is concentrated at 0.05-100 Hz, QRS complex energy is concentrated at 5-45 Hz, and P, T wave frequency is generally below 10 Hz. Three kinds of interference mainly exist in the electrocardio data, namely 50Hz power frequency interference; the base line drifts, the frequency range is usually between 0.15 Hz and 0.3Hz, and sometimes reaches 1 Hz; myoelectric interference and wide frequency range. In one embodiment, a denoising method combining an integrated empirical mode decomposition algorithm and a wavelet soft threshold is adopted, on one hand, the integrated empirical mode decomposition algorithm avoids the occurrence of an aliasing phenomenon of the empirical mode decomposition mode, and on the other hand, the wavelet soft threshold reduces the loss of useful information caused in the coefficient threshold processing process.
Preferably, the denoising processing can be performed on the short-time electrocardiogram data obtained after cutting by the same method.
In one embodiment, the random initialization of the parameters in S2 follows a gaussian distribution:
where μ is desired, σ is standard deviation, σ is2Is the variance.
In one embodiment, the CNN model in step S2 includes a plurality of convolutional layers, a plurality of pooling layers, and a fully connected layer.
Specifically, for the short-time electrocardiographic data used in the present invention, the sizes of the convolution kernel and the pooling layer are adjusted to adapt to the length of the electrocardiographic data, and the construction of the CNN model is completed by the construction of the number of layers and the configuration of the parameters of each layer, and the structure of the CNN model is shown in fig. 3.
For faster model prediction, the convolution operation is performed using one-dimensional CNN, which is easier to train, with only a few tens of Back Propagation (BP) periods.
In one embodiment, the training algorithm in S3 is a back propagation algorithm, the classification model is trained based on a stochastic gradient descent algorithm, and the calculation formula of the convolutional layer output value is as follows:
wherein x isi+m,j+nThe (i + m) th row and (j + n) th column elements representing the image; wm,nRepresents the weight of the mth row and the nth column; wbA bias term representing a filter; a isi,jAn ith row and a jth column element representing a feature map; f denotes the activation function, it should be noted that at each convolution layer of cnn, the data exists in two dimensions, which can be regarded as the superposition of a plurality of one-dimensional data, wherein each one-dimensional data is called a feature map, i.e. the feature obtained after the convolution operation, i.e. the filter, i.e. the convolution kernel.
The back propagation algorithm specifically includes:
s3.1: calculating the output value a of each neuron in a forward directionj；
S3.2: inverse computation of error term δ for each neuronjIs an error term δjLoss function E of networkdPartial derivatives of weighted input netj to neurons, i.e.
S3.3: calculating per neuron connection weight WjiIs of the formula
Where j represents the jth of the network.
Specifically, the labels of the electrocardiogram data all adopt a one-hot form, the total loss function of the model is defined as the sum of cross entropy loss functions of all the electrocardiogram data participating in training, and the labels and the prediction results of the single electrocardiogram data are respectively labeliAnd predictioniThen the model total loss function is defined as follows:
the implementation of training the classification model based on the stochastic gradient descent algorithm may be: and in a specific implementation process, the initial learning rate value is 0.001, the EPOCH value is 100, the batch _ size value is 64, the learning rate attenuation step size is 8000 and the single learning rate attenuation rate is 0.96.
Specifically, the LeakyRelu activation function is
The LeakyRelu activation function is to assign a non-zero slope to all negative values, where xiRepresents input, yiRepresents the output, aiRepresents a fixed parameter in the interval (1, + ∞).
The target classification model realizes classification of all electrocardiogram data, the number of classified classes is 2, and the class names corresponding to the classes are as follows: normal electrocardiogram data Normal, Abnormal data Abnormal.
In one embodiment, S5 specifically includes:
s5.1: determining the abnormal confidence coefficient of each short-time electrocardio data in the cut sample according to the original classification model;
s5.2: selecting preset short-time electrocardiogram data with the highest abnormal confidence coefficient as input of an initial classification model according to the magnitude of the abnormal confidence coefficient, and performing one round of training on the initial classification model;
s5.3: and repeatedly executing S5.1 and S5.2 until the classification model converges or a preset training round number is reached.
Specifically, for each short-time electrocardiogram data in the cut sample, the probability that the data is abnormal can be obtained through the original classification model, and the probability is the abnormal confidence. The preset number can be set according to actual conditions, such as 5, 6 and the like.
In one embodiment, the training samples include normal samples labeled as normal and abnormal samples labeled as abnormal, and S5.2 specifically includes:
according to the magnitude of the abnormal confidence coefficient, selecting preset short-time electrocardiogram data with the highest abnormal confidence coefficient from the abnormal samples as abnormal training data, and selecting preset short-time electrocardiogram data with the highest abnormal confidence coefficient from the normal samples as normal training data;
and taking the abnormal training data and the normal training data as the input of the initial classification model, and carrying out one round of training on the initial classification model.
Specifically, referring to fig. 4, a training diagram is selected for the confidence level used in model migration learning, and each cut block is short-time electrocardiographic data.
Generally speaking, the invention utilizes the information of the first quantity (large quantity) of short-time electrocardiogram data sets to help a small quantity of electrocardiogram data to perform migration training, and further performs identification and classification on the electrocardiogram abnormity, thereby avoiding the problem of insufficient priori knowledge caused by less quantity of partial electrocardiogram data, improving the stability of the model, screening the abnormal part in the abnormal electrocardiogram data, avoiding the interference of the normal part in the abnormal electrocardiogram data on the classification model, and ensuring the robustness of the model. The method can improve the detection rate, the identification precision and the identification efficiency of abnormal data in the electrocardiogram data, assist the diagnosis of doctors, reduce the workload of the doctors, improve the diagnosis efficiency and provide objective and accurate diagnosis results for the doctors, thereby having greater social and practical values.
Example two
Based on the same inventive concept, the present embodiment provides an abnormal cardiac electrical signal identification apparatus based on transfer learning and confidence level selection, please refer to fig. 5, the apparatus includes:
the short-time electrocardiographic data acquisition module 201 is configured to acquire a first number of short-time electrocardiographic data, and perform denoising processing on the acquired first number of short-time electrocardiographic data;
the CNN model building module 202 is used for building a CNN model and randomly initializing parameters in the CNN model;
the CNN model training module 203 is used for defining a loss function by taking the denoised short-time electrocardiogram data of the first quantity as a training data set, and training the CNN model by adopting a preset algorithm to obtain an original classification model;
the long-time electrocardiogram data cutting module 204 is configured to obtain a second amount of long-time electrocardiogram data, use the obtained second amount of long-time electrocardiogram data as a training sample, and cut the training sample, where the cut training sample includes a plurality of short-time electrocardiogram data, and the second amount is far smaller than the first amount;
the migration training module 205 is configured to determine an abnormal confidence of each short-time electrocardiographic data in the cut sample according to the original classification model, select target short-time data according to the abnormal confidence, perform migration training on the initial classification model, and obtain a target classification model;
the identification module 206 is configured to identify the electrocardiographic data to be identified by using the target classification model.
Since the device described in the second embodiment of the present invention is a device used for implementing the abnormal cardiac electrical signal identification method based on transfer learning and confidence level selection in the first embodiment of the present invention, a person skilled in the art can understand the specific structure and deformation of the device based on the method described in the first embodiment of the present invention, and thus details are not described herein. All the devices adopted in the method of the first embodiment of the present invention belong to the protection scope of the present invention.
EXAMPLE III
Referring to fig. 6, based on the same inventive concept, the present application further provides a computer-readable storage medium 300, on which a computer program 311 is stored, which when executed implements the method according to the first embodiment.
Because the computer-readable storage medium introduced in the third embodiment of the present invention is a computer-readable storage medium used for implementing the abnormal cardiac electrical signal identification method based on transfer learning and confidence level selection in the first embodiment of the present invention, based on the method introduced in the first embodiment of the present invention, persons skilled in the art can understand the specific structure and deformation of the computer-readable storage medium, and thus details are not described here. Any computer readable storage medium used in the method of the first embodiment of the present invention is within the scope of the present invention.
Example four
Based on the same inventive concept, the present application further provides a computer device, please refer to fig. 7, which includes a storage 401, a processor 402, and a computer program 403 stored in the storage and running on the processor, and when the processor 402 executes the above program, the method in the first embodiment is implemented.
Because the computer device introduced in the fourth embodiment of the present invention is a computer device used for implementing the abnormal cardiac electrical signal identification method based on transfer learning and confidence level selection in the first embodiment of the present invention, based on the method introduced in the first embodiment of the present invention, persons skilled in the art can understand the specific structure and deformation of the computer device, and thus details are not described here. All the computer devices used in the method in the first embodiment of the present invention are within the scope of the present invention.
As will be appreciated by one skilled in the art, embodiments of the present invention may be provided as a method, system, or computer program product. Accordingly, the present invention may take the form of an entirely hardware embodiment, an entirely software embodiment or an embodiment combining software and hardware aspects. Furthermore, the present invention may take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, and the like) having computer-usable program code embodied therein.
The present invention is described with reference to flowchart illustrations and/or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention. It will be understood that each flow and/or block of the flow diagrams and/or block diagrams, and combinations of flows and/or blocks in the flow diagrams and/or block diagrams, can be implemented by computer program instructions. These computer program instructions may be provided to a processor of a general purpose computer, special purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, create means for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.
While preferred embodiments of the present invention have been described, additional variations and modifications in those embodiments may occur to those skilled in the art once they learn of the basic inventive concepts. Therefore, it is intended that the appended claims be interpreted as including preferred embodiments and all such alterations and modifications as fall within the scope of the invention.
It will be apparent to those skilled in the art that various modifications and variations can be made in the embodiments of the present invention without departing from the spirit or scope of the embodiments of the invention. Thus, if such modifications and variations of the embodiments of the present invention fall within the scope of the claims of the present invention and their equivalents, the present invention is also intended to encompass such modifications and variations.
Claims (10)
1. An electrocardio abnormal signal identification method based on transfer learning and confidence degree selection is characterized by comprising the following steps:
s1: acquiring a first quantity of short-time electrocardiogram data, and denoising the acquired first quantity of short-time electrocardiogram data;
s2: building a CNN model, and randomly initializing parameters in the CNN model;
s3: defining a loss function by taking the denoised short-time electrocardiogram data of the first quantity as a training data set, and training a CNN model by adopting a preset algorithm to obtain an original classification model;
s4: obtaining a second quantity of long-time electrocardiogram data, taking the obtained second quantity of long-time electrocardiogram data as a training sample, and cutting the training sample, wherein the cut training sample comprises a plurality of short-time electrocardiogram data, and the second quantity is far smaller than the first quantity;
s5: determining the abnormal confidence coefficient of each short-time electrocardiogram data in the cut sample according to the original classification model, selecting target short-time data according to the abnormal confidence coefficient, and performing migration training on the initial classification model to obtain a target classification model;
s6: and identifying the electrocardiogram data to be identified by using the target classification model.
2. The method as claimed in claim 1, wherein the denoising processing is performed on the first amount of short-time electrocardiographic data acquired in S1, and comprises: and denoising by adopting a method of combining an integrated empirical mode decomposition algorithm and a wavelet soft threshold.
3. The method of claim 2, wherein the random initialization of parameters in S2 follows a gaussian distribution:
where μ is desired, σ is standard deviation, σ is2Is the variance.
4. The method of claim 1, wherein the CNN model in step S2 includes a plurality of convolutional layers, a plurality of pooling layers, and a fully-connected layer.
5. The method of claim 1, wherein the training algorithm in S3 is a back propagation algorithm, the classification model is trained based on a stochastic gradient descent algorithm, and the formula for calculating the convolutional layer output value is as follows:
wherein x isi+m,j+nThe (i + m) th row and (j + n) th column elements representing the image; wm,nRepresents the weight of the mth row and the nth column; wbA bias term representing a filter; a isi,jThe ith row and jth column elements representing the features obtained after the convolution operation; f represents an activation function; the back propagation algorithm specifically includes:
s3.1: calculating the output value a of each neuron in a forward directionj；
S3.2: inverse computation of error term δ for each neuronjIs an error term δjLoss function E of networkdPartial derivatives of weighted input netj to neurons, i.e.
S3.3: calculating per neuron connection weight WjiIs of the formula
Where j represents the jth of the network.
6. The method of claim 5, wherein S5 specifically comprises:
s5.1: determining the abnormal confidence coefficient of each short-time electrocardio data in the cut sample according to the original classification model;
s5.2: selecting preset short-time electrocardiogram data with the highest abnormal confidence coefficient as input of an initial classification model according to the magnitude of the abnormal confidence coefficient, and performing one round of training on the initial classification model;
s5.3: and repeatedly executing S5.1 and S5.2 until the classification model converges or a preset training round number is reached.
7. The method of claim 6, wherein the training samples include normal samples labeled normal and abnormal samples labeled abnormal, and S5.2 specifically includes:
according to the magnitude of the abnormal confidence coefficient, selecting preset short-time electrocardiogram data with the highest abnormal confidence coefficient from the abnormal samples as abnormal training data, and selecting preset short-time electrocardiogram data with the highest abnormal confidence coefficient from the normal samples as normal training data;
and taking the abnormal training data and the normal training data as the input of the initial classification model, and carrying out one round of training on the initial classification model.
8. An abnormal cardiac electrical signal recognition device based on transfer learning and confidence degree selection, comprising:
the short-time electrocardio-data acquisition module is used for acquiring a first amount of short-time electrocardio-data and denoising the acquired first amount of short-time electrocardio-data;
the CNN model building module is used for building a CNN model and randomly initializing parameters in the CNN model;
the CNN model training module is used for defining a loss function by taking the denoised short-time electrocardiogram data of the first quantity as a training data set and training the CNN model by adopting a preset algorithm to obtain an original classification model;
the long-time electrocardiogram data cutting module is used for acquiring a second amount of long-time electrocardiogram data, taking the acquired second amount of long-time electrocardiogram data as a training sample, cutting the training sample, wherein the cut training sample comprises a plurality of short-time electrocardiogram data, and the second amount is far smaller than the first amount;
the migration training module is used for determining the abnormal confidence coefficient of each short-time electrocardiogram data in the cut sample according to the original classification model, selecting target short-time data according to the abnormal confidence coefficient, and performing migration training on the initial classification model to obtain a target classification model;
and the identification module is used for identifying the electrocardiogram data to be identified by utilizing the target classification model.
9. A computer-readable storage medium, on which a computer program is stored, characterized in that the program, when executed, implements the method of any one of claims 1 to 7.
10. A computer device comprising a memory, a processor and a computer program stored on the memory and executable on the processor, characterized in that the processor implements the method according to any of claims 1 to 7 when executing the program.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010096524.0A CN111134662B (en) | 2020-02-17 | 2020-02-17 | Electrocardio abnormal signal identification method and device based on transfer learning and confidence degree selection |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010096524.0A CN111134662B (en) | 2020-02-17 | 2020-02-17 | Electrocardio abnormal signal identification method and device based on transfer learning and confidence degree selection |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN111134662A true CN111134662A (en) | 2020-05-12 |
| CN111134662B CN111134662B (en) | 2021-04-16 |
Family
ID=70527530
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202010096524.0A Active CN111134662B (en) | 2020-02-17 | 2020-02-17 | Electrocardio abnormal signal identification method and device based on transfer learning and confidence degree selection |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN111134662B (en) |
Cited By (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111990989A (en) * | 2020-08-31 | 2020-11-27 | 武汉大学 | Electrocardiosignal identification method based on generation countermeasure and convolution cyclic network |
| CN112286761A (en) * | 2020-10-29 | 2021-01-29 | 山东中创软件商用中间件股份有限公司 | Database state detection method and device, electronic equipment and storage medium |
| CN112383440A (en) * | 2020-12-01 | 2021-02-19 | 苏州中德双智科创发展有限公司 | Cloud model accuracy improving method and device, electronic equipment and storage medium |
| CN112401903A (en) * | 2020-11-03 | 2021-02-26 | 沈阳东软智能医疗科技研究院有限公司 | Electrocardio data identification method and device, storage medium and electronic equipment |
| CN112998652A (en) * | 2021-02-23 | 2021-06-22 | 华南理工大学 | Photoelectric volume pulse wave pressure identification method and system |
| CN113325773A (en) * | 2021-05-30 | 2021-08-31 | 滨州学院 | Intelligent pilot physical training monitoring system based on big data |
| CN113469218A (en) * | 2021-06-02 | 2021-10-01 | 安徽心之声医疗科技有限公司 | Electrocardiosignal critical value processing method, system and medium based on deep learning |
| CN115640507A (en) * | 2022-12-09 | 2023-01-24 | 南京备力医疗科技有限公司 | Abnormal data screening method based on electrocardio-heart sound joint analysis |
| WO2023004572A1 (en) * | 2021-07-27 | 2023-02-02 | 京东方科技集团股份有限公司 | Model training method, signal recognition method and apparatus, computing processing device, computer program, and computer-readable medium |
Citations (26)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104616031A (en) * | 2015-01-22 | 2015-05-13 | 哈尔滨工业大学深圳研究生院 | Transfer learning method and device |
| CN106355248A (en) * | 2016-08-26 | 2017-01-25 | 深圳先进技术研究院 | Deep convolution neural network training method and device |
| CN106901723A (en) * | 2017-04-20 | 2017-06-30 | 济南浪潮高新科技投资发展有限公司 | A kind of electrocardiographic abnormality automatic diagnosis method |
| CN107657318A (en) * | 2017-11-13 | 2018-02-02 | 成都蓝景信息技术有限公司 | A kind of electrocardiogram sorting technique based on deep learning model |
| CN107822622A (en) * | 2017-09-22 | 2018-03-23 | 成都比特律动科技有限责任公司 | Electrocardiographic diagnosis method and system based on depth convolutional neural networks |
| CN109241397A (en) * | 2018-07-06 | 2019-01-18 | 四川斐讯信息技术有限公司 | A kind of method and apparatus for cleaning data |
| CN109276242A (en) * | 2018-08-02 | 2019-01-29 | 深圳市理邦精密仪器股份有限公司 | The method and apparatus of electrocardiosignal type identification |
| CN109303560A (en) * | 2018-11-01 | 2019-02-05 | 杭州质子科技有限公司 | A kind of atrial fibrillation recognition methods of electrocardiosignal in short-term based on convolution residual error network and transfer learning |
| CN109544603A (en) * | 2018-11-28 | 2019-03-29 | 上饶师范学院 | Method for tracking target based on depth migration study |
| CN109620205A (en) * | 2018-12-26 | 2019-04-16 | 上海联影智能医疗科技有限公司 | Electrocardiogram (ECG) data classification method, device, computer equipment and storage medium |
| US20190159737A1 (en) * | 2015-08-14 | 2019-05-30 | Elucid Bioimaging Inc. | Methods and systems for utilizing quantitative imaging |
| WO2019104003A1 (en) * | 2017-11-21 | 2019-05-31 | Beth Israel Deaconess Medical Center, Inc | Systems and methods for automatically interpreting images of microbiological samples |
| CN109887595A (en) * | 2019-01-16 | 2019-06-14 | 成都蓝景信息技术有限公司 | Heartbeat anomalous identification algorithm based on depth learning technology |
| CN110090012A (en) * | 2019-03-15 | 2019-08-06 | 上海图灵医疗科技有限公司 | A kind of human body diseases detection method and testing product based on machine learning |
| CN110109543A (en) * | 2019-04-30 | 2019-08-09 | 福州大学 | C-VEP recognition methods based on subject migration |
| CN110148466A (en) * | 2019-05-15 | 2019-08-20 | 东北大学 | A kind of heart impact signal atrial fibrillation computer aided diagnosing method based on transfer learning |
| CN110147878A (en) * | 2018-11-28 | 2019-08-20 | 腾讯科技（深圳）有限公司 | Data processing method, device and equipment |
| CN110200619A (en) * | 2019-06-06 | 2019-09-06 | 中山大学孙逸仙纪念医院 | It is a kind of to establish major cardiovascular disease risks Early-warning Model method and device |
| CN110211684A (en) * | 2019-06-05 | 2019-09-06 | 四川长虹电器股份有限公司 | The electrocardiogram classification method of BP neural network based on genetic algorithm optimization |
| CN110299205A (en) * | 2019-07-23 | 2019-10-01 | 上海图灵医疗科技有限公司 | Biomedicine signals characteristic processing and evaluating method, device and application based on artificial intelligence |
| CN110427924A (en) * | 2019-09-09 | 2019-11-08 | 桂林电子科技大学 | A kind of heart impact signal based on LSTM more classifying identification methods automatically |
| CN110584649A (en) * | 2019-09-09 | 2019-12-20 | 武汉大学 | Method and system for identifying abnormal electrocardiosignals in dynamic single-lead electrocardiogram |
| CN110619322A (en) * | 2019-10-18 | 2019-12-27 | 武汉大学 | Multi-lead electrocardio abnormal signal identification method and system based on multi-flow convolution cyclic neural network |
| CN110693486A (en) * | 2019-09-27 | 2020-01-17 | 武汉中旗生物医疗电子有限公司 | Electrocardiogram abnormity labeling method and device |
| CN110751131A (en) * | 2019-11-16 | 2020-02-04 | 李汭傧 | Arrhythmia detection device |
| CN110786847A (en) * | 2018-08-02 | 2020-02-14 | 深圳市理邦精密仪器股份有限公司 | Electrocardiogram signal library building method and analysis method |
-
2020
- 2020-02-17 CN CN202010096524.0A patent/CN111134662B/en active Active
Patent Citations (26)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104616031A (en) * | 2015-01-22 | 2015-05-13 | 哈尔滨工业大学深圳研究生院 | Transfer learning method and device |
| US20190159737A1 (en) * | 2015-08-14 | 2019-05-30 | Elucid Bioimaging Inc. | Methods and systems for utilizing quantitative imaging |
| CN106355248A (en) * | 2016-08-26 | 2017-01-25 | 深圳先进技术研究院 | Deep convolution neural network training method and device |
| CN106901723A (en) * | 2017-04-20 | 2017-06-30 | 济南浪潮高新科技投资发展有限公司 | A kind of electrocardiographic abnormality automatic diagnosis method |
| CN107822622A (en) * | 2017-09-22 | 2018-03-23 | 成都比特律动科技有限责任公司 | Electrocardiographic diagnosis method and system based on depth convolutional neural networks |
| CN107657318A (en) * | 2017-11-13 | 2018-02-02 | 成都蓝景信息技术有限公司 | A kind of electrocardiogram sorting technique based on deep learning model |
| WO2019104003A1 (en) * | 2017-11-21 | 2019-05-31 | Beth Israel Deaconess Medical Center, Inc | Systems and methods for automatically interpreting images of microbiological samples |
| CN109241397A (en) * | 2018-07-06 | 2019-01-18 | 四川斐讯信息技术有限公司 | A kind of method and apparatus for cleaning data |
| CN109276242A (en) * | 2018-08-02 | 2019-01-29 | 深圳市理邦精密仪器股份有限公司 | The method and apparatus of electrocardiosignal type identification |
| CN110786847A (en) * | 2018-08-02 | 2020-02-14 | 深圳市理邦精密仪器股份有限公司 | Electrocardiogram signal library building method and analysis method |
| CN109303560A (en) * | 2018-11-01 | 2019-02-05 | 杭州质子科技有限公司 | A kind of atrial fibrillation recognition methods of electrocardiosignal in short-term based on convolution residual error network and transfer learning |
| CN109544603A (en) * | 2018-11-28 | 2019-03-29 | 上饶师范学院 | Method for tracking target based on depth migration study |
| CN110147878A (en) * | 2018-11-28 | 2019-08-20 | 腾讯科技（深圳）有限公司 | Data processing method, device and equipment |
| CN109620205A (en) * | 2018-12-26 | 2019-04-16 | 上海联影智能医疗科技有限公司 | Electrocardiogram (ECG) data classification method, device, computer equipment and storage medium |
| CN109887595A (en) * | 2019-01-16 | 2019-06-14 | 成都蓝景信息技术有限公司 | Heartbeat anomalous identification algorithm based on depth learning technology |
| CN110090012A (en) * | 2019-03-15 | 2019-08-06 | 上海图灵医疗科技有限公司 | A kind of human body diseases detection method and testing product based on machine learning |
| CN110109543A (en) * | 2019-04-30 | 2019-08-09 | 福州大学 | C-VEP recognition methods based on subject migration |
| CN110148466A (en) * | 2019-05-15 | 2019-08-20 | 东北大学 | A kind of heart impact signal atrial fibrillation computer aided diagnosing method based on transfer learning |
| CN110211684A (en) * | 2019-06-05 | 2019-09-06 | 四川长虹电器股份有限公司 | The electrocardiogram classification method of BP neural network based on genetic algorithm optimization |
| CN110200619A (en) * | 2019-06-06 | 2019-09-06 | 中山大学孙逸仙纪念医院 | It is a kind of to establish major cardiovascular disease risks Early-warning Model method and device |
| CN110299205A (en) * | 2019-07-23 | 2019-10-01 | 上海图灵医疗科技有限公司 | Biomedicine signals characteristic processing and evaluating method, device and application based on artificial intelligence |
| CN110427924A (en) * | 2019-09-09 | 2019-11-08 | 桂林电子科技大学 | A kind of heart impact signal based on LSTM more classifying identification methods automatically |
| CN110584649A (en) * | 2019-09-09 | 2019-12-20 | 武汉大学 | Method and system for identifying abnormal electrocardiosignals in dynamic single-lead electrocardiogram |
| CN110693486A (en) * | 2019-09-27 | 2020-01-17 | 武汉中旗生物医疗电子有限公司 | Electrocardiogram abnormity labeling method and device |
| CN110619322A (en) * | 2019-10-18 | 2019-12-27 | 武汉大学 | Multi-lead electrocardio abnormal signal identification method and system based on multi-flow convolution cyclic neural network |
| CN110751131A (en) * | 2019-11-16 | 2020-02-04 | 李汭傧 | Arrhythmia detection device |
Cited By (12)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111990989A (en) * | 2020-08-31 | 2020-11-27 | 武汉大学 | Electrocardiosignal identification method based on generation countermeasure and convolution cyclic network |
| CN112286761A (en) * | 2020-10-29 | 2021-01-29 | 山东中创软件商用中间件股份有限公司 | Database state detection method and device, electronic equipment and storage medium |
| CN112286761B (en) * | 2020-10-29 | 2023-07-07 | 山东中创软件商用中间件股份有限公司 | Database state detection method and device, electronic equipment and storage medium |
| CN112401903A (en) * | 2020-11-03 | 2021-02-26 | 沈阳东软智能医疗科技研究院有限公司 | Electrocardio data identification method and device, storage medium and electronic equipment |
| CN112401903B (en) * | 2020-11-03 | 2023-12-22 | 沈阳东软智能医疗科技研究院有限公司 | Electrocardiogram data identification method and device, storage medium and electronic equipment |
| CN112383440A (en) * | 2020-12-01 | 2021-02-19 | 苏州中德双智科创发展有限公司 | Cloud model accuracy improving method and device, electronic equipment and storage medium |
| CN112998652A (en) * | 2021-02-23 | 2021-06-22 | 华南理工大学 | Photoelectric volume pulse wave pressure identification method and system |
| CN113325773A (en) * | 2021-05-30 | 2021-08-31 | 滨州学院 | Intelligent pilot physical training monitoring system based on big data |
| CN113469218A (en) * | 2021-06-02 | 2021-10-01 | 安徽心之声医疗科技有限公司 | Electrocardiosignal critical value processing method, system and medium based on deep learning |
| WO2023004572A1 (en) * | 2021-07-27 | 2023-02-02 | 京东方科技集团股份有限公司 | Model training method, signal recognition method and apparatus, computing processing device, computer program, and computer-readable medium |
| CN115640507A (en) * | 2022-12-09 | 2023-01-24 | 南京备力医疗科技有限公司 | Abnormal data screening method based on electrocardio-heart sound joint analysis |
| CN115640507B (en) * | 2022-12-09 | 2024-03-15 | 镜电(南京)科技发展有限公司 | Abnormal data screening method based on electrocardiographic and heart sound combined analysis |
Also Published As
| Publication number | Publication date |
|---|---|
| CN111134662B (en) | 2021-04-16 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN111134662B (en) | Electrocardio abnormal signal identification method and device based on transfer learning and confidence degree selection | |
| Yildirim | A novel wavelet sequence based on deep bidirectional LSTM network model for ECG signal classification | |
| CN110619322A (en) | Multi-lead electrocardio abnormal signal identification method and system based on multi-flow convolution cyclic neural network | |
| CN111990989A (en) | Electrocardiosignal identification method based on generation countermeasure and convolution cyclic network | |
| Meintjes et al. | Fundamental heart sound classification using the continuous wavelet transform and convolutional neural networks | |
| Prasad et al. | Classification of ECG arrhythmias using multi-resolution analysis and neural networks | |
| CN111772619B (en) | Heart beat identification method based on deep learning, terminal equipment and storage medium | |
| Wu et al. | A novel method for classification of ECG arrhythmias using deep belief networks | |
| CN111291727B (en) | Method and device for detecting signal quality by using photoplethysmography | |
| Sai | A review on arrhythmia classification using ECG signals | |
| Rahman et al. | ECG classification for detecting ECG arrhythmia empowered with deep learning approaches | |
| Dallali et al. | Fuzzy c-means clustering, Neural Network, WT, and HRV for classification of cardiac arrhythmia | |
| CN111759345A (en) | Heart valve abnormality analysis method, system and device based on convolutional neural network | |
| Rath et al. | Improved heart disease detection from ECG signal using deep learning based ensemble model | |
| CN113180679A (en) | Single-lead electrocardio abnormal signal identification method | |
| Prakash et al. | A system for automatic cardiac arrhythmia recognition using electrocardiogram signal | |
| Singh et al. | Short and noisy electrocardiogram classification based on deep learning | |
| Yuen et al. | Detecting noisy ECG QRS complexes using WaveletCNN autoencoder and ConvLSTM | |
| Awati et al. | Smart heart disease detection using particle swarm optimization and support vector machine | |
| Marzog et al. | ECG-Signal classification using efficient machine learning approach | |
| Ajitkumar Singh et al. | Heart abnormality classification using PCG and ECG recordings | |
| Allam et al. | Patient-specific ECG beat classification using EMD and deep learning-based technique | |
| Rahman et al. | Automated detection of cardiac arrhythmia based on a hybrid CNN-LSTM network | |
| JP2023104885A (en) | Electrocardiographic heart rate multi-type prediction method based on graph convolution | |
| Mitrokhin et al. | Deep learning approach for QRS wave detection in ECG monitoring |
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 |