CN113052229A - Heart disease classification method and device based on electrocardiogram data - Google Patents
Heart disease classification method and device based on electrocardiogram data Download PDFInfo
- Publication number
- CN113052229A CN113052229A CN202110303683.8A CN202110303683A CN113052229A CN 113052229 A CN113052229 A CN 113052229A CN 202110303683 A CN202110303683 A CN 202110303683A CN 113052229 A CN113052229 A CN 113052229A
- Authority
- CN
- China
- Prior art keywords
- data
- classification model
- classification
- electrocardiogram
- electrocardiographic
- 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
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
- G06F18/2148—Generating training patterns; Bootstrap methods, e.g. bagging or boosting characterised by the process organisation or structure, e.g. boosting cascade
-
- 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
- A61B5/7246—Details of waveform analysis using correlation, e.g. template matching or determination of similarity
-
- 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
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/243—Classification techniques relating to the number of classes
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/243—Classification techniques relating to the number of classes
- G06F18/2431—Multiple classes
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/25—Fusion techniques
- G06F18/253—Fusion techniques of extracted features
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/25—Fusion techniques
- G06F18/254—Fusion techniques of classification results, e.g. of results related to same input data
- G06F18/256—Fusion techniques of classification results, e.g. of results related to same input data of results relating to different input data, e.g. multimodal recognition
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02A—TECHNOLOGIES FOR ADAPTATION TO CLIMATE CHANGE
- Y02A90/00—Technologies having an indirect contribution to adaptation to climate change
- Y02A90/10—Information and communication technologies [ICT] supporting adaptation to climate change, e.g. for weather forecasting or climate simulation
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Life Sciences & Earth Sciences (AREA)
- Data Mining & Analysis (AREA)
- Physics & Mathematics (AREA)
- Artificial Intelligence (AREA)
- Health & Medical Sciences (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Evolutionary Computation (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Evolutionary Biology (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Molecular Biology (AREA)
- General Health & Medical Sciences (AREA)
- Biomedical Technology (AREA)
- Biophysics (AREA)
- Animal Behavior & Ethology (AREA)
- Mathematical Physics (AREA)
- Pathology (AREA)
- Physiology (AREA)
- Heart & Thoracic Surgery (AREA)
- Medical Informatics (AREA)
- Psychiatry (AREA)
- Veterinary Medicine (AREA)
- Signal Processing (AREA)
- Surgery (AREA)
- Public Health (AREA)
- Fuzzy Systems (AREA)
- Computational Linguistics (AREA)
- Computing Systems (AREA)
- Software Systems (AREA)
- Measurement And Recording Of Electrical Phenomena And Electrical Characteristics Of The Living Body (AREA)
Abstract
The invention relates to a cardiac disease classification method based on electrocardiogram data, which comprises the following steps: establishing a corresponding electrocardiogram data set aiming at a set disease category; performing neural network training based on the electrocardiogram data set to obtain a multi-classification model; extracting specific lead data aiming at the electrocardio data of part of the set symptoms; training an image classification model based on specific lead data of the electrocardio data of the partial diseases to obtain a two-classification model; and fusing the multi-classification model and the two-classification model to realize classification of multiple diseases. The invention can accurately distinguish the heart diseases with more similar characteristics.
Description
Technical Field
The invention relates to the technical field of intelligent identification of electrocardiogram data, in particular to a method and a device for classifying heart diseases based on electrocardiogram data and a computer storage medium.
Background
Today, heart diseases are classified into a plurality of categories, and some heart diseases have a plurality of similar characteristics, which causes great difficulty for doctors in clinical diagnosis. For example, left bundle branch block, right ventricular pacing rhythm, pre-excitation syndrome A and pre-excitation syndrome B all occur in a heart chamber, have a plurality of similar characteristics such as large QRS wave width, and at present, no good intelligent judgment method for the diseases does not appear, and the methods and other diseases with different origins are combined into more than twenty types of electrocardiograms or even more than thirty types of electrocardiograms for multi-classification judgment. Generally, a certain data processing mode is carried out on data of all different diseases, then a pre-designed algorithm model is trained, and finally a fixed model is obtained to process other test set data to obtain a multi-classification prediction result containing twenty-three diseases.
Disclosure of Invention
In view of the above, there is a need to provide a method, an apparatus and a computer storage medium for classifying cardiac disorders based on electrocardiographic data, so as to solve the problem that many disorders with similar characteristics are difficult to be accurately distinguished.
The invention provides a cardiac disease classification method based on electrocardiogram data, which comprises the following steps:
establishing a corresponding electrocardiogram data set aiming at a set disease category;
performing neural network training based on the electrocardiogram data set to obtain a multi-classification model;
extracting specific lead data aiming at the electrocardio data of part of the set symptoms;
training an image classification model based on specific lead data of the electrocardio data of the partial diseases to obtain a two-classification model;
and fusing the multi-classification model and the two-classification model to realize classification of multiple diseases.
Further, a corresponding electrocardiogram data set is established for the set disease category, specifically:
acquiring electrocardiogram data in the same quantity range aiming at different set disease categories, wherein the electrocardiogram data comprises electrocardiogram data and Holter data;
and preprocessing the electrocardiogram data to obtain the electrocardiogram data set.
Further, the set condition categories include left bundle branch block, right ventricular pacing rhythm, priming syndrome type a, priming syndrome type B, and others.
Further, neural network training is performed based on the electrocardiogram data set to obtain a multi-classification model, which specifically comprises the following steps:
cutting the electrocardio data in the electrocardio data set into heart beat data, and establishing a heart beat data set;
performing neural network training based on the electrocardiogram data set to obtain a first multi-classification model;
performing neural network training based on the heartbeat data set to obtain a second multi-classification model;
and fusing the first multi-classification model and the second multi-classification model to obtain the multi-classification model.
Further, the first multi-classification model and the second multi-classification model are fused to obtain the multi-classification model, which specifically comprises:
inputting the electrocardiogram data into the first classification model to obtain a first feature vector;
inputting the heartbeat data into the second classification model to obtain a second feature vector;
splicing the first feature vector and the second feature vector to obtain a fusion feature vector;
and carrying out classification training on the fusion feature vector through XGboost to obtain the fused multi-classification model.
Further, training an image classification model based on specific lead data of the electrocardiographic data of the partial diseases to obtain a two-classification model, specifically:
and extracting first lead data of the electrocardio data of the partial symptoms for training to obtain the two classification models.
Further, the multi-classification model and the two-classification model are fused to realize classification of multiple diseases, specifically:
inputting the data to be tested into the multi-classification model to obtain a prediction result;
and judging whether the prediction result is one of the partial diseases, if so, inputting the data to be detected into the two-classification model again to obtain a final prediction result, otherwise, directly outputting the prediction result of the multi-classification model.
The invention also provides a heart disease classification device based on the electrocardio data, which comprises a processor and a memory, wherein the memory is stored with a computer program, and the computer program is executed by the processor to realize the heart disease classification method based on the electrocardio data.
The present invention also provides a computer storage medium having a computer program stored thereon, which, when executed by a processor, implements the method for classifying a cardiac disorder based on electrocardiographic data.
Has the advantages that: according to the method, firstly, an electrocardiogram data set is established aiming at the set diseases with more similar characteristics, a more targeted model is established based on the electrocardiogram data set, and more accurate capture can be realized on the difference characteristics of the set diseases. Meanwhile, aiming at partial diseases with higher judgment difficulty and larger difference in data quantity, a binary model is trained by extracting specific lead data, and the set difference of the diseases on the specific leads is extracted and distinguished by the binary model, so that the identification rate of the partial diseases with higher judgment difficulty can be further improved. The model is fused, so that the difference of the set diseases can be grasped from the most comprehensive angle, and the disease recognition rate is improved.
Drawings
FIG. 1 is a flowchart of a first embodiment of a method for classifying cardiac disorders based on electrocardiographic data according to the present invention;
fig. 2 is a flowchart illustrating a first embodiment of a method for classifying cardiac disorders based on electrocardiographic data according to the present invention.
Detailed Description
The accompanying drawings, which are incorporated in and constitute a part of this application, illustrate preferred embodiments of the invention and together with the description, serve to explain the principles of the invention and not to limit the scope of the invention.
Example 1
As shown in fig. 1, embodiment 1 of the present invention provides a method for classifying cardiac disorders based on electrocardiographic data, including the steps of:
s1, establishing a corresponding electrocardiogram data set aiming at the set disease category;
s2, carrying out neural network training based on the electrocardiogram data set to obtain a multi-classification model;
s3, extracting specific lead data aiming at the electrocardio data of a part of symptoms in the set symptoms;
s4, training an image classification model based on special lead data of the electrocardio data of the partial diseases to obtain a two-classification model;
and S5, fusing the multi-classification model and the two-classification model to realize classification of multiple diseases.
The embodiment aims to design a method for intelligently classifying according to common characteristics of set symptoms and differences among the set symptoms, so that the requirements of doctors on the high accuracy of the symptoms can be met, the characteristics of high-risk diseases on electrocardiogram data can be captured in time, meanwhile, the burden of daily large-scale image viewing of the doctors is relieved, and the doctors are more concentrated on more complex diagnosis of the electrocardiogram symptoms.
Specifically, in this embodiment, an electrocardiographic data set is first established for a set disease condition with a large number of selected similar features, and a more targeted model is established based on the electrocardiographic data set, so that more accurate capture of the difference features of the set disease condition can be achieved. In the model training process, the training data dimensionality is changed every 10 times in an iteration mode, the data set distribution is enriched, the detection capability of the model on multi-scale data is improved, and the model can obtain a good prediction result on different input sizes. Meanwhile, aiming at partial diseases with higher judgment difficulty and larger difference in data quantity, a binary model is trained by extracting specific lead data, and the set difference of the diseases on the specific leads is extracted and distinguished by the binary model, so that the identification rate of the partial diseases with higher judgment difficulty can be further improved. The model is fused, so that the difference of the set diseases can be grasped from the most comprehensive angle, and the disease recognition rate is improved.
Preferably, the set disorder categories include left bundle branch block, right ventricular pacing rhythm, type a pre-excitation syndrome, type B pre-excitation syndrome, and others.
In particular, the predetermined condition may be characterized by an abnormal electrocardiogram.
The quantity of electrocardiogram data is huge, the categories are numerous, the ordinary doctor with professional skills has a large amount of treatment workload, the five types of electrocardiograms are often caused by serious diseases of other parts of a body and are mixed with other numerous electrocardio diseases, the difficulty in accurately distinguishing the types of diseases is increased, in the existing research, all the diseases are processed in the same mode and in the same flow, the scheme design which is prominent on the single item of the five types of diseases is omitted, the accuracy of the obtained intelligent judgment result is not high, and the accuracy required in the medical field can be ensured by re-determining the result by the doctor.
In this embodiment, the above five types of set symptoms are taken as an example, and the classification method is specifically described for the above five types of set symptoms.
Preferably, a corresponding electrocardiogram data set is established for a set disease category, specifically:
acquiring electrocardiogram data in the same quantity range aiming at different set disease categories, wherein the electrocardiogram data comprises electrocardiogram data and Holter data;
and preprocessing the electrocardiogram data to obtain the electrocardiogram data set.
As shown in fig. 2, in the present embodiment, the collected electrocardiographic data is first classified into six categories, i.e., a left bundle branch block, a right ventricular pacing rhythm, a type a pre-excitation syndrome, and a type B pre-excitation syndrome. In the actual treatment process, the latter three types are not common, and the pre-excitation syndrome is more in type A and less in type B. In order to ensure the balance of various data, Holter data is also introduced to serve as the supplementary data of the last three types in addition to the data of a common electrocardiograph, so that the aim of balancing the distribution of different types of data is fulfilled. After the electrocardio data of different types of set diseases are collected, the electrocardio data are preprocessed, and the electrocardio data after noise removal are obtained through processing such as power frequency filtering removal. Meanwhile, in order to enrich data distribution, electrocardiograph data and Holter data are brought into an integral data set, and after different lengths of data (the length of the electrocardiograph data is different from that of the Holter data) are subjected to a training process and are separated by a certain batch, the length of input data is randomly changed, and the prediction capability of a model on multi-scale data is improved.
Preferably, neural network training is performed based on the electrocardiogram data set to obtain a multi-classification model, which specifically comprises:
cutting the electrocardio data in the electrocardio data set into heart beat data, and establishing a heart beat data set;
performing neural network training based on the electrocardiogram data set to obtain a first multi-classification model;
performing neural network training based on the heartbeat data set to obtain a second multi-classification model;
and fusing the first multi-classification model and the second multi-classification model to obtain the multi-classification model.
As shown in fig. 2, since all the five types of setting disorders have the characteristic of a wide QRS wave, we firstly adopt a QRS wave detection algorithm such as a PT algorithm to determine an R point, and segment the original electrocardiographic data to obtain cardioversion data, which is used as an individual cardioversion data set. In the embodiment, the neural network selects network structures such as ResNeXt and the like which are better in classification effect, and after each set number of iterations in the network training process, data with one length is randomly selected as input, and the network is automatically adjusted to continue the training process. And respectively carrying out the model training process on the original electrocardio data and the 12-derivative data of the cardiotomy data to obtain a first six classification model and a second six classification model, and then carrying out fusion to obtain a final fusion six classification model.
In the aspect of neural network selection, the method comprises the steps of but not limited to ResNeXt and other network structures which perform well on image classification recognition, and identifying a Draknet19 network which is faster to train multi-scale data and has a better effect.
Preferably, the first multi-classification model and the second multi-classification model are fused to obtain the multi-classification model, specifically:
inputting the electrocardiogram data into the first classification model to obtain a first feature vector;
inputting the heartbeat data into the second classification model to obtain a second feature vector;
splicing the first feature vector and the second feature vector to obtain a fusion feature vector;
and carrying out classification training on the fusion feature vector through XGboost to obtain the fused multi-classification model.
Specifically, each piece of electrocardiographic data is subjected to a last full-connection layer of a first six-classification model to obtain a group of feature vectors, meanwhile, corresponding cardioid data is also subjected to a group of feature vectors obtained through a second six-classification network, the two groups of feature vectors are spliced to obtain feature representation of the electrocardiographic data, then the feature representations obtained by all training data are subjected to classification training through XGboost, and a fused multi-classification model is obtained. The test data is subjected to feature representation through two six classification models, and then is subjected to XGboost model to obtain a final six classification prediction result.
On the fusion of the two multi-classification models, a voting method or a weighted average method which consumes less computing resources and has higher speed can be taken as a preference, a deep learning method which has higher precision and occupies more computing resources, such as a decision tree, can be considered, and the requirements on the detection efficiency and the detection precision in practical application are taken as important indexes of a final selection method.
Preferably, training of an image classification model is performed based on specific lead data of the electrocardiographic data of the partial disease to obtain a two-classification model, which specifically comprises:
and extracting first lead data of the electrocardio data of the partial symptoms for training to obtain the two classification models. As shown in fig. 2, since the five setting conditions targeted by the present embodiment are closely related to WPW-a (abbreviation for pre-excitation syndrome type a) and WPW-B (abbreviation for pre-excitation syndrome type B) and the volume of WPW-B data is small, the first lead data of the two conditions, i.e. the V1 lead data, is selected and the image classification model is trained to obtain a binary model.
In particular, the first lead may be a limb-one lead and the V1 lead may be the first lead of a chest lead.
Preferably, the multi-classification model and the two-classification model are fused to realize classification of multiple diseases, specifically:
inputting the data to be tested into the multi-classification model to obtain a prediction result;
and judging whether the prediction result is one of the partial diseases, if so, inputting the data to be detected into the two-classification model again to obtain a final prediction result, otherwise, directly outputting the prediction result of the multi-classification model.
As shown in fig. 2, two six classification models are fused, and the fusion modes include non-deep learning modes such as voting and weighting, and deep learning modes such as XGBoost, and the like, and the fusion mode with better effect is selected according to a specific experimental result. This step results in a fused six-class model. And the next step is to fuse the model and a two-classification model, test data firstly pass through a six-classification model to obtain a prediction result, data predicted to be WPW-A or WPW-B is further passed through a two-classification model to obtain a new prediction result, and the new prediction result is merged with other types of results in the prediction result of the multi-classification model to obtain a final prediction result.
According to the embodiment, the common characteristics of the five types of set diseases and the difference between the five types of set diseases are intelligently judged, so that the requirement of a doctor on the high accuracy of the judgment of the five types of diseases can be met, the characteristics of the high-risk diseases on an electrocardiogram can be captured in time, meanwhile, the burden of the doctor on seeing a large amount of pictures in daily life is relieved, and the doctor can concentrate on more complex diagnosis of the electrocardiogram diseases.
In the scheme, the five types of data are specially selected to establish a more targeted model, and the difference characteristics of the five types of data can be captured more accurately. Aiming at common points on QRS waveforms, heart beat data is intercepted, and the five types of differences of characteristics on the heart beat are trained by a network more pertinently, so that the recognition rate is improved. In the model training process, the training data dimensionality is changed every 10 times in an iteration mode, the data set distribution is enriched, the detection capability of the model on multi-scale data is improved, and the model can obtain a good prediction result on different input sizes. Aiming at two types of WPW data which are more difficult to distinguish and have larger difference in data quantity, and the main difference of the two types of data in form appears on the V1 lead, a V1 lead training binary model is taken, and the recognition rate of the two types of data can be further improved. The three models are fused, so that the differences of the five types of symptoms can be grasped from the perspective as comprehensive as possible, and the symptom identification rate is improved. On the training data, long data, heartbeat data and V1 lead data aiming at WPW class are respectively selected as training data to be trained respectively, and the accuracy of the five classes of diseases is improved more specifically. And finally, three model results are fused, so that the integral model can more accurately capture the important characteristics of the five diseases, the model prediction capability is improved, the detection rate of the five diseases is enhanced, and the risk is reduced.
Example 2
Embodiment 2 of the present invention provides an apparatus for classifying cardiac disorders based on electrocardiographic data, including a processor and a memory, where the memory stores a computer program, and when the computer program is executed by the processor, the method for classifying cardiac disorders based on electrocardiographic data according to embodiment 1 is implemented.
The cardiac disorder classification device based on the electrocardiographic data provided by the embodiment of the invention is used for realizing the cardiac disorder classification method based on the electrocardiographic data, so that the cardiac disorder classification method based on the electrocardiographic data has the technical effects which are also possessed by the cardiac disorder classification device based on the electrocardiographic data, and the details are not repeated here.
Example 3
The computer storage medium provided by the embodiment of the invention is used for realizing the cardiac disorder classification method based on the electrocardiographic data, so that the technical effect of the cardiac disorder classification method based on the electrocardiographic data is also achieved by the computer storage medium, and the description is omitted here.
The above description is only for the preferred embodiment of the present invention, but the scope of the present invention is not limited thereto, and any changes or substitutions that can be easily conceived by those skilled in the art within the technical scope of the present invention are included in the scope of the present invention.
Claims (9)
1. A heart disease classification method based on electrocardiogram data is characterized by comprising the following steps:
establishing a corresponding electrocardiogram data set aiming at a set disease category;
performing neural network training based on the electrocardiogram data set to obtain a multi-classification model;
extracting specific lead data aiming at the electrocardio data of part of the set symptoms;
training an image classification model based on specific lead data of the electrocardio data of the partial diseases to obtain a two-classification model;
and fusing the multi-classification model and the two-classification model to realize classification of multiple diseases.
2. The method for classifying cardiac disorders based on electrocardiographic data according to claim 1, wherein a corresponding electrocardiographic data set is established for a set disorder category, specifically:
acquiring electrocardiogram data in the same quantity range aiming at different set disease categories, wherein the electrocardiogram data comprises electrocardiogram data and Holter data;
and preprocessing the electrocardiogram data to obtain the electrocardiogram data set.
3. The method of classifying cardiac disorders based on electrocardiographic data according to claim 1, wherein said set disorder categories include left bundle branch block, right ventricular pacing rhythm, type a pre-excitation syndrome, type B pre-excitation syndrome, and others.
4. The cardiac disorder classification method based on electrocardiographic data according to claim 1, wherein neural network training is performed based on the electrocardiographic data set to obtain a multi-classification model, specifically:
cutting the electrocardio data in the electrocardio data set into heart beat data, and establishing a heart beat data set;
performing neural network training based on the electrocardiogram data set to obtain a first multi-classification model;
performing neural network training based on the heartbeat data set to obtain a second multi-classification model;
and fusing the first multi-classification model and the second multi-classification model to obtain the multi-classification model.
5. The method for classifying cardiac disorders based on electrocardiographic data according to claim 4, wherein the first multi-classification model and the second multi-classification model are fused to obtain the multi-classification model, and specifically:
inputting the electrocardiogram data into the first classification model to obtain a first feature vector;
inputting the heartbeat data into the second classification model to obtain a second feature vector;
splicing the first feature vector and the second feature vector to obtain a fusion feature vector;
and carrying out classification training on the fusion feature vector through XGboost to obtain the fused multi-classification model.
6. The method for classifying cardiac disorders based on electrocardiographic data according to claim 1, wherein training of an image classification model is performed based on specific lead data of the electrocardiographic data of the partial disorders to obtain a binary classification model, specifically:
and extracting first lead data of the electrocardio data of the partial symptoms for training to obtain the two classification models.
7. The cardiac disorder classification method based on electrocardiographic data according to claim 1, wherein the multi-classification model and the bi-classification model are fused to realize classification of multiple disorders, specifically:
inputting the data to be tested into the multi-classification model to obtain a prediction result;
and judging whether the prediction result is one of the partial diseases, if so, inputting the data to be detected into the two-classification model again to obtain a final prediction result, otherwise, directly outputting the prediction result of the multi-classification model.
8. A cardiac disorder classification apparatus based on electrocardiographic data, comprising a processor and a memory, the memory storing thereon a computer program, which when executed by the processor, implements the cardiac disorder classification method based on electrocardiographic data according to any one of claims 1 to 7.
9. A computer storage medium having a computer program stored thereon, wherein the computer program, when executed by a processor, implements a method for cardiac disorder classification based on electrocardiographic data according to any one of claims 1 to 7.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110303683.8A CN113052229B (en) | 2021-03-22 | 2021-03-22 | Heart condition classification method and device based on electrocardiographic data |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110303683.8A CN113052229B (en) | 2021-03-22 | 2021-03-22 | Heart condition classification method and device based on electrocardiographic data |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN113052229A true CN113052229A (en) | 2021-06-29 |
| CN113052229B CN113052229B (en) | 2023-08-29 |
Family
ID=76514582
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202110303683.8A Active CN113052229B (en) | 2021-03-22 | 2021-03-22 | Heart condition classification method and device based on electrocardiographic data |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN113052229B (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114587378A (en) * | 2022-04-11 | 2022-06-07 | 平安科技（深圳）有限公司 | Deep learning-based electrocardiogram classification method, device, equipment and storage medium |
Citations (15)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107837082A (en) * | 2017-11-27 | 2018-03-27 | 乐普（北京）医疗器械股份有限公司 | Electrocardiogram automatic analysis method and device based on artificial intelligence self study |
| CN108836302A (en) * | 2018-03-19 | 2018-11-20 | 武汉海星通技术股份有限公司 | Electrocardiogram intelligent analysis method and system based on deep neural network |
| US20190082988A1 (en) * | 2017-09-19 | 2019-03-21 | Tata Consultancy Services Limited | Cascaded binary classifier for identifying rhythms in a single-lead electrocardiogram (ecg) signal |
| CN109893119A (en) * | 2019-03-24 | 2019-06-18 | 浙江好络维医疗技术有限公司 | A kind of P wave recognition positioning method based on multi-lead clustering |
| CN110236518A (en) * | 2019-04-02 | 2019-09-17 | 武汉大学 | Method and device for joint classification of ECG and cardiac shock signals based on neural network |
| CN110288028A (en) * | 2019-06-27 | 2019-09-27 | 北京邮电大学 | ECG detection method, system, device and computer-readable storage medium |
| US20190328243A1 (en) * | 2016-12-21 | 2019-10-31 | Emory University | Methods and Systems for Determining Abnormal Cardiac Activity |
| CN110522442A (en) * | 2019-08-16 | 2019-12-03 | 广州视源电子科技股份有限公司 | Multi-lead electrocardiographic abnormality detection device, electronic apparatus, and storage medium |
| CN110801218A (en) * | 2019-11-13 | 2020-02-18 | 清华大学 | Electrocardiogram data processing method, device, electronic device and computer readable medium |
| CN110852288A (en) * | 2019-11-15 | 2020-02-28 | 苏州大学 | Cell image classification method based on two-stage convolutional neural network |
| CN111012332A (en) * | 2020-01-20 | 2020-04-17 | 太原理工大学 | Multi-label classification method of 12-lead ECG signal based on neural network |
| CN111160139A (en) * | 2019-12-13 | 2020-05-15 | 中国科学院深圳先进技术研究院 | ECG signal processing method, device and terminal equipment |
| US20200237246A1 (en) * | 2017-11-27 | 2020-07-30 | Lepu Medical Technology (Bejing) Co., Ltd. | Automatic recognition and classification method for electrocardiogram heartbeat based on artificial intelligence |
| CN111700608A (en) * | 2020-07-24 | 2020-09-25 | 武汉中旗生物医疗电子有限公司 | Multi-classification method and device for electrocardiosignals |
| US20210035689A1 (en) * | 2018-04-17 | 2021-02-04 | Bgi Shenzhen | Modeling method and apparatus for diagnosing ophthalmic disease based on artificial intelligence, and storage medium |
-
2021
- 2021-03-22 CN CN202110303683.8A patent/CN113052229B/en active Active
Patent Citations (15)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20190328243A1 (en) * | 2016-12-21 | 2019-10-31 | Emory University | Methods and Systems for Determining Abnormal Cardiac Activity |
| US20190082988A1 (en) * | 2017-09-19 | 2019-03-21 | Tata Consultancy Services Limited | Cascaded binary classifier for identifying rhythms in a single-lead electrocardiogram (ecg) signal |
| CN107837082A (en) * | 2017-11-27 | 2018-03-27 | 乐普（北京）医疗器械股份有限公司 | Electrocardiogram automatic analysis method and device based on artificial intelligence self study |
| US20200237246A1 (en) * | 2017-11-27 | 2020-07-30 | Lepu Medical Technology (Bejing) Co., Ltd. | Automatic recognition and classification method for electrocardiogram heartbeat based on artificial intelligence |
| CN108836302A (en) * | 2018-03-19 | 2018-11-20 | 武汉海星通技术股份有限公司 | Electrocardiogram intelligent analysis method and system based on deep neural network |
| US20210035689A1 (en) * | 2018-04-17 | 2021-02-04 | Bgi Shenzhen | Modeling method and apparatus for diagnosing ophthalmic disease based on artificial intelligence, and storage medium |
| CN109893119A (en) * | 2019-03-24 | 2019-06-18 | 浙江好络维医疗技术有限公司 | A kind of P wave recognition positioning method based on multi-lead clustering |
| CN110236518A (en) * | 2019-04-02 | 2019-09-17 | 武汉大学 | Method and device for joint classification of ECG and cardiac shock signals based on neural network |
| CN110288028A (en) * | 2019-06-27 | 2019-09-27 | 北京邮电大学 | ECG detection method, system, device and computer-readable storage medium |
| CN110522442A (en) * | 2019-08-16 | 2019-12-03 | 广州视源电子科技股份有限公司 | Multi-lead electrocardiographic abnormality detection device, electronic apparatus, and storage medium |
| CN110801218A (en) * | 2019-11-13 | 2020-02-18 | 清华大学 | Electrocardiogram data processing method, device, electronic device and computer readable medium |
| CN110852288A (en) * | 2019-11-15 | 2020-02-28 | 苏州大学 | Cell image classification method based on two-stage convolutional neural network |
| CN111160139A (en) * | 2019-12-13 | 2020-05-15 | 中国科学院深圳先进技术研究院 | ECG signal processing method, device and terminal equipment |
| CN111012332A (en) * | 2020-01-20 | 2020-04-17 | 太原理工大学 | Multi-label classification method of 12-lead ECG signal based on neural network |
| CN111700608A (en) * | 2020-07-24 | 2020-09-25 | 武汉中旗生物医疗电子有限公司 | Multi-classification method and device for electrocardiosignals |
Non-Patent Citations (4)
| Title |
|---|
| 何龙 著: "《深入理解XGBoost 高效机器学习算法与进阶》", 31 January 2020, 机械工业出版社, pages: 5 - 7 * |
| 周群一，等: "人工神经网络在心拍分类中的应用", 中国医疗器械杂志, no. 03, pages 41 - 46 * |
| 庞彦伟，等: "基于卷积神经网络的第一导联心电图心拍分类", 天津大学学报(自然科学与工程技术版), no. 10, pages 5 - 12 * |
| 熊鹏，等: "基于稀疏降噪自动编码机的心律失常自动分类", 激光杂志, no. 04, pages 156 - 160 * |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114587378A (en) * | 2022-04-11 | 2022-06-07 | 平安科技（深圳）有限公司 | Deep learning-based electrocardiogram classification method, device, equipment and storage medium |
Also Published As
| Publication number | Publication date |
|---|---|
| CN113052229B (en) | 2023-08-29 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Shaker et al. | Generalization of convolutional neural networks for ECG classification using generative adversarial networks | |
| Golany et al. | SimGANs: Simulator-based generative adversarial networks for ECG synthesis to improve deep ECG classification | |
| KR102451795B1 (en) | ECG signal detection method | |
| US9968303B2 (en) | System and method of adaptive interpretation of ECG waveforms | |
| CN112603330B (en) | Electrocardiosignal identification and classification method | |
| CN109117730A (en) | Electrocardiogram auricular fibrillation real-time judge method, apparatus, system and storage medium | |
| CN109091138A (en) | The judgment means and Mapping System of arrhythmia cordis originating point | |
| CN108256452A (en) | A kind of method of the ECG signal classification of feature based fusion | |
| CN112690802B (en) | Method, device, terminal and storage medium for detecting electrocardiosignals | |
| CN112022141B (en) | Electrocardiosignal class detection method, electrocardiosignal class detection device and storage medium | |
| CN111956208B (en) | A method for ECG signal classification based on ultra-lightweight convolutional neural network | |
| Vasconcellos et al. | Siamese convolutional neural network for heartbeat classification using limited 12-lead ECG datasets | |
| Lima et al. | Heartbeat anomaly detection using adversarial oversampling | |
| Wu et al. | ECG classification using ICA features and support vector machines | |
| Chu et al. | A novel two-lead arrhythmia classification system based on CNN and LSTM | |
| Li | Deep intermediate representation and in-set voting scheme for multiple-beat electrocardiogram classification | |
| CN108960050A (en) | Classification model training method, ECG data classifying method, device and equipment | |
| Pinto et al. | Explaining ECG biometrics: Is it all in the QRS? | |
| CN113486752A (en) | Emotion identification method and system based on electrocardiosignals | |
| Shah et al. | ECG‐TransCovNet: a hybrid transformer model for accurate arrhythmia detection using electrocardiogram signals | |
| CN113052229B (en) | Heart condition classification method and device based on electrocardiographic data | |
| CN112989971B (en) | Electrocardiogram data fusion method and device for different data sources | |
| Lin et al. | Explainable deep neural network for identifying cardiac abnormalities using class activation map | |
| CN109077719A (en) | signal identification method, device, equipment and storage medium | |
| Roland et al. | An automated system for arrhythmia detection using ECG records from MITDB |
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 |