WO2022134472A1

WO2022134472A1 - Method, device and terminal for electrocardiosignal detection and storage medium

Info

Publication number: WO2022134472A1
Application number: PCT/CN2021/097285
Authority: WO
Inventors: 赵婷婷; 孙行智; 朱昭苇; 徐卓扬; 刘卓
Original assignee: 平安科技（深圳）有限公司
Priority date: 2020-12-25
Filing date: 2021-05-31
Publication date: 2022-06-30
Also published as: CN112690802A; CN112690802B

Abstract

A device and terminal for electrocardiosignal detection and a storage medium. The device comprises: a first obtaining unit, configured to obtain electrocardiosignals to be detected; and a processing unit, configured to input said electrocardiosignals into a trained electrocardiosignal detection model for processing to obtain the locations of abnormal electrocardiosignals in said electrocardiosignals and abnormal types corresponding to the abnormal electrocardiosignals. On the basis of an abnormal electrocardiosignal screening module in the electrocardiosignal detection model, the abnormal electrocardiosignals can be screened out from the electrocardiosignals, the locations of the abnormal electrocardiosignals are determined, and an abnormal electrocardiosignal classification module in the electrocardiosignal detection model can analyze the abnormal electrocardiosignals to obtain the abnormal types corresponding to the abnormal electrocardiosignals. According to the device, the abnormal electrocardiosignals in said electrocardiosignals can be accurately positioned, meanwhile, the abnormal types of the abnormal electrocardiosignals can be accurately recognized, and the electrocardiosignal reading speed and accuracy are improved. The terminal and the storage medium correspond to the device.

Description

A method, device, terminal and storage medium for detecting ECG signals

This application requires the Chinese Patent Office of the State Intellectual Property Office of the People's Republic of China with the application number 202011560066.8 and the invention titled "A method, device, terminal and storage medium for detecting electrocardiographic signals" on December 25, 2020. Priority to the patent application, the entire contents of which are incorporated herein by reference.

technical field

The present application belongs to the field of artificial intelligence, and in particular, relates to a method, device, terminal and storage medium for detecting ECG signals.

Background technique

As an important information reflecting the vital signs of patients, ECG signals have been widely used in the diagnosis of various cardiac abnormalities, and can also be used to predict the morbidity and mortality of cardiovascular diseases. Early, correct diagnosis of cardiac abnormalities can increase the chances of successful treatment. However, manual interpretation of ECG signals is time-consuming and labor-intensive, and requires higher requirements for medical personnel. Therefore, it is very necessary to automatically interpret ECG signals based on neural network models.

The inventor realized that the existing neural network model cannot identify the abnormal type corresponding to the abnormal ECG signal while locating the position of the abnormal ECG signal, which is not conducive to the interpretation of the ECG signal.

technical problem

One of the purposes of the embodiments of the present application is to provide a method, device, terminal and storage medium for detecting ECG signals, so as to solve the problem that the existing neural network model cannot identify the location of abnormal ECG signals while identifying the location of abnormal ECG signals. It is a technical problem that the abnormal type corresponding to the abnormal ECG signal is found, which is not conducive to the interpretation of the ECG signal.

technical solutions

In a first aspect, an embodiment of the present application provides a method for detecting an ECG signal, the method comprising:

Obtain the ECG signal to be detected;

Inputting the ECG signal into the trained ECG signal detection model for processing to obtain the position of the abnormal ECG signal in the ECG signal and the abnormal type corresponding to the abnormal ECG signal;

Wherein, the ECG signal detection model includes an abnormal ECG signal screening module and an abnormal ECG signal classification module, and the abnormal ECG signal screening module is used to screen out the abnormal ECG signals from the ECG signals, And determine the position of the abnormal ECG signal, and the abnormal ECG signal classification module is used for analyzing the abnormal ECG signal to obtain the abnormal type corresponding to the abnormal ECG signal.

In a second aspect, an embodiment of the present application provides a device for detecting an ECG signal, the device comprising:

a first acquisition unit, used to acquire the ECG signal to be detected;

a processing unit, configured to input the ECG signal into a trained ECG signal detection model for processing, and obtain the position of the abnormal ECG signal in the ECG signal and the abnormal type corresponding to the abnormal ECG signal; Wherein, the ECG signal detection model includes an abnormal ECG signal screening module and an abnormal ECG signal classification module, and the abnormal ECG signal screening module is used to screen out the abnormal ECG signals from the ECG signals, And determine the position of the abnormal ECG signal, and the abnormal ECG signal classification module is used for analyzing the abnormal ECG signal to obtain the abnormal type corresponding to the abnormal ECG signal.

In a third aspect, an embodiment of the present application provides a terminal for detecting electrocardiographic signals, including a memory, a processor, and a computer program stored in the memory and running on the processor, where the processor implements when executing the computer program:

Obtain the ECG signal to be detected;

In a fourth aspect, an embodiment of the present application provides a computer-readable storage medium, the computer-readable storage medium may be non-volatile or volatile, and the computer-readable storage medium stores a computer program, the computer program Implemented when executed by the processor:

Obtain the ECG signal to be detected;

beneficial effect

Compared with the prior art, the embodiment of the present application has the beneficial effect that the ECG signal to be detected is processed based on the pre-trained ECG signal detection model. The abnormal ECG signal screening module in the ECG signal detection model can screen out the abnormal ECG signal from the ECG signal, and determine the position of the abnormal ECG signal, and the classification module in the ECG signal detection model can detect the abnormal ECG signal. The abnormal ECG signal screened by the ECG signal screening module is analyzed to obtain the abnormal type corresponding to the abnormal ECG signal. Based on this method, not only the abnormal ECG signal in the ECG signal to be detected can be accurately located, but also the abnormal type of the abnormal ECG signal can be accurately identified, and the speed and accuracy of the interpretation of the ECG signal can be improved. .

Description of drawings

In order to illustrate the technical solutions in the embodiments of the present application more clearly, the following briefly introduces the accompanying drawings that are used in the description of the embodiments or exemplary technologies. Obviously, the drawings in the following description are only for the present application. In some embodiments, for those of ordinary skill in the art, other drawings can also be obtained according to these drawings without any creative effort.

1 is a schematic flowchart of a method for detecting an electrocardiogram signal provided by an embodiment of the present application;

FIG. 2 is a schematic flowchart of a method for detecting an ECG signal provided by another embodiment of the present application;

3 is a schematic diagram of ECG signal division provided by an embodiment of the present application;

FIG. 4 is a schematic flowchart of a method for detecting an ECG signal provided by another embodiment of the present application;

5 is a schematic flowchart of a method for detecting an ECG signal provided by another embodiment of the present application;

FIG. 6 is a schematic flowchart of a method for detecting an ECG signal provided by another embodiment of the present application;

FIG. 7 is a schematic diagram of a device for detecting an ECG signal provided by an embodiment of the present application;

FIG. 8 is a schematic diagram of a terminal for detecting an ECG signal provided by another embodiment of the present application.

Embodiments of the present invention

In order to make the purpose, technical solutions and advantages of the present application more clearly understood, the present application will be described in further detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application, but not to limit the present application.

However, during the training process of the existing neural network model, the selected training samples are incomplete and inaccurate, so that the neural network model obtained by training cannot locate the position of abnormal ECG signals, and cannot accurately identify the corresponding abnormal ECG signals. exception type.

In view of this, the present application provides a method for detecting an ECG signal, which processes the ECG signal to be detected based on a pre-trained ECG signal detection model. The abnormal ECG signal screening module in the ECG signal detection model can screen out the abnormal ECG signal from the ECG signal, and determine the position of the abnormal ECG signal, and the classification module in the ECG signal detection model can detect the abnormal ECG signal. The abnormal ECG signal screened by the ECG signal screening module is analyzed to obtain the abnormal type corresponding to the abnormal ECG signal. Based on this method, not only the abnormal ECG signal in the ECG signal to be detected can be accurately located, but also the abnormal type of the abnormal ECG signal can be accurately identified, and the speed and accuracy of the interpretation of the ECG signal can be improved. .

Please refer to FIG. 1. FIG. 1 is a schematic flowchart of a method for detecting an ECG signal provided by an embodiment of the present application. The execution subject of the method for detecting an ECG signal in this embodiment is a terminal, and the terminal includes but is not limited to a mobile terminal such as a smart phone, a tablet computer, a computer, a Personal Digital Assistant (PDA), etc., and may also include a desktop computer, a server, etc. Wait for the terminal. The method for detecting an ECG signal as shown in FIG. 1 may include S101 to S102, and the details are as follows:

S101: Acquire an ECG signal to be detected.

A waveform diagram that can completely represent one cardiac cycle of the heart is called a cardiac beat signal, and the ECG signal to be detected may include several cardiac beat signals.

When it is necessary to detect whether the ECG signal of a certain user is abnormal, the user's ECG signal can be input to the terminal that detects the ECG signal, that is, the ECG signal to be detected is input, and the terminal that detects the ECG signal receives the ECG signal to be detected. ECG signal.

The ECG signal to be detected may also be stored in a preset folder of the terminal in advance, and when an instruction to detect the ECG signal is received, the ECG signal to be detected is extracted from the preset folder.

The medical detection device may also generate the ECG signal to be detected after detecting the user's heart, and transmit the ECG signal to be detected to a terminal that detects the ECG signal, and the terminal receives the ECG signal to be detected. This is only an exemplary description, and it is not limited.

S102: Input the ECG signal into a trained ECG signal detection model for processing, and obtain the position of the abnormal ECG signal in the ECG signal and the abnormal type corresponding to the abnormal ECG signal;

In this embodiment, a pre-trained ECG signal detection model is pre-stored in the terminal. The ECG signal detection model is obtained by using a machine learning algorithm to train the first sample training set based on the initial screening network and the second sample training set based on the initial classification network. It can be understood that the ECG signal detection model can be pre-trained by the terminal, or the file corresponding to the ECG signal detection model can be transplanted to the terminal after being pre-trained by other devices. That is to say, the executive body that trains the ECG signal detection model and the executive body that uses the ECG signal detection model may be the same or different.

Exemplarily, the trained ECG signal detection model may include an abnormal ECG signal screening module and an abnormal ECG signal classification module. The abnormal ECG signal screening module is obtained by training the first sample training set based on the initial screening network. The abnormal electrocardiographic signal screening module performs screening processing on the electrocardiographic signal to be detected, screens out the abnormal electrocardiographic signal in the electrocardiographic signal to be detected, and can determine the position of the abnormal electrocardiographic signal. It should be noted that, when there is no abnormal ECG signal in the ECG signal to be detected, the abnormal ECG signal screening module will output a null value if the abnormal ECG signal cannot be detected. It can be understood that when there is no abnormal ECG signal in the ECG signal to be detected, the output of the trained ECG signal detection model is empty.

The abnormal ECG signal classification module is obtained by training the second sample training set based on the initial classification network. The abnormal ECG signal classification module analyzes the abnormal ECG signals screened by the abnormal ECG signal screening module, and obtains the abnormal type corresponding to the abnormal ECG signals.

Wherein, the abnormal type refers to the type to which the abnormal ECG signal belongs. For example, abnormal types may include atrial fibrillation, first-degree AV block, left bundle branch block, right bundle branch block, premature atrial contractions, premature ventricular contractions, ST segment and T wave abnormalities (ST-T changes), P Wave abnormalities, abnormal QRS complexes, significant changes in QRS complexes and ST-T, and arrhythmias. The abnormality type can be used to assist the doctor in judging the user's disease. For example, the abnormality type corresponding to the abnormal ECG signal of the user is P-wave abnormality. According to previous experience, it can be judged that the user has a high probability of atrial hypertrophy; the abnormal ECG signal of the user is abnormal. The corresponding abnormal type is ST-T change, and the user's myocardial ischemia can be judged according to the usual experience. This is only an exemplary description, and it is not limited.

It is worth noting that when there is no abnormal ECG signal in the ECG signal to be detected, the abnormal ECG signal screening module will output a null value if it cannot detect the abnormal ECG signal. Accordingly, the abnormal ECG signal classification module will output a null value at this time. If there is no abnormal ECG signal that can be processed, a null value will also be output. It is finally reflected that the output of the trained ECG signal detection model is empty.

Please refer to FIG. 2. FIG. 2 is a schematic flowchart of a method for detecting an ECG signal provided by another embodiment of the present application. Optionally, in a possible implementation manner, as shown in FIG. 2 , the foregoing S102 may include S1021 to S1028, and the details are as follows:

S1021: Divide the ECG signal into a plurality of ECG signal segments.

Divide the ECG signal into a plurality of ECG signal segments with the same length in sequence. For example, according to the original order of each heartbeat signal in the ECG signal to be detected, every second heartbeat signal or every third heartbeat signal is divided to obtain a plurality of ECG signal segments, and each ECG signal segment contains 2 heartbeat signals or 3 heartbeat signals. This is only an exemplary description, and it is not limited.

Please refer to FIG. 3 , which is a schematic diagram of ECG signal division provided by an embodiment of the present application. Figure 3 only shows a certain segment of the ECG signal. As shown in Figure 3, the ECG signal segment selected by each rectangular frame represents the divided ECG signal segment. There are four ECG signal segments. The ECG signal segment contains 3 heartbeat signals, and the second ECG signal, the third ECG signal, and the third ECG signal each contain 2 heartbeat signals. This figure is only an exemplary illustration, which is not limited.

S1022: Select a first ECG signal segment and a second ECG signal segment from the plurality of ECG signal segments, the first ECG signal segment is adjacent to the second ECG signal segment, and the first ECG signal segment is adjacent to the second ECG signal segment. An ECG signal segment is any one of the plurality of ECG signal segments.

The first ECG signal segment refers to any one ECG signal segment among the divided multiple ECG signal segments, and the second ECG signal segment refers to an ECG signal segment adjacent to the first ECG signal segment. Usually, a first ECG signal segment is selected from a plurality of ECG signal segments according to the order of division, and the next ECG signal segment adjacent to the first ECG signal segment is selected as the second ECG signal segment.

Exemplarily, for the four ECG signal segments shown in FIG. 3 , the first ECG signal segment on the left is selected as the first ECG signal segment, and the next ECG signal segment adjacent to it is the second ECG signal segment. If the third ECG signal segment is selected as the first ECG signal segment, the last ECG signal segment adjacent to it is the second ECG signal segment. This figure is only an exemplary illustration, which is not limited.

S1023: Input the first ECG signal segment and the second ECG signal segment into the abnormal ECG signal screening module for screening processing to obtain the first ECG signal segment and the second ECG signal segment The degree of matching between signal fragments.

The degree of matching can be understood as the degree of similarity, that is, the degree of similarity between two ECG signal segments. In this embodiment, it is to determine the degree of similarity between the first ECG signal segment and the second ECG signal segment.

Exemplarily, the matching degree may be represented by 1 and 0, wherein 1 indicates that the two electrocardiographic signal segments have a high degree of matching, and 0 represents that the two electrocardiographic signal segments have a low matching degree. Input the first ECG signal segment and the second ECG signal segment into the abnormal ECG signal screening module for screening processing. When the abnormal ECG signal screening module outputs 1, it indicates the first ECG signal segment and the second ECG signal. The matching degree between the segments is high; when the abnormal ECG signal screening module outputs 0, it indicates that the matching degree between the first ECG signal segment and the second ECG signal segment is low.

Optionally, in a possible implementation manner, the matching degree may also be represented by a specific numerical value, percentage, etc., for example, the matching degree may be 95, 90, 80, 30, 95%, 60%, and so on. This is only an exemplary description, and it is not limited.

Comparing the matching degree with the preset threshold, when the matching degree is less than the preset threshold, execute S1024-S1027, and when the matching degree is greater than or equal to the preset threshold, execute S1028.

S1024: When it is detected that the matching degree is less than a preset threshold, mark the first ECG signal segment and the second ECG signal segment as abnormal ECG signals respectively, and determine the first ECG signal the position of the segment and the second ECG signal segment in the ECG signal, respectively.

The preset threshold is used for comparison with the matching degree, and the comparison result is used to assist in determining whether to mark the first ECG signal segment and the second ECG signal segment as abnormal ECG signals. When the expressions of the matching degree are different, the preset threshold can also be adjusted accordingly.

Exemplarily, when the matching degree is represented by 1 and 0, the preset threshold may be set to 1. For example, the matching degree between the first ECG signal segment and the second ECG signal segment is 0. At this time, it is detected that the matching degree 0 is less than the preset threshold value 1, which proves that the first ECG signal segment and the second ECG signal are The segment matching degree is low, that is, the similarity degree is low. It can be understood that the first ECG signal segment and the second ECG signal segment are two ECG signal segments with large differences, which are likely to be abnormal ECG signal segments. The first ECG signal segment and the second ECG signal segment are respectively marked as abnormal ECG signals.

While the first ECG signal segment and the second ECG signal segment are respectively marked as abnormal ECG signals, the respective differences between the first ECG signal segment and the second ECG signal segment in the entire ECG signal to be detected are acquired. The position is to obtain the coordinates of the first ECG signal segment and the second ECG signal segment in the entire ECG signal to be detected, respectively. Optionally, the position or coordinates can also be represented by the description of the number of segments. For example, the position of the first ECG signal segment in the entire ECG signal to be detected is the first ECG signal segment, the second ECG signal segment The position of the electrical signal segment in the entire ECG signal to be detected is the second ECG signal segment. This is only an exemplary description, and it is not limited.

Optionally, the ECG signal segment marked as abnormal ECG signal can be marked with a rectangular frame, for example, the first ECG signal segment and the first ECG signal segment marked as abnormal ECG signal can be marked with a rectangular frame different from the color of the ECG signal. The second ECG signal segment is framed, so that the doctor can directly check which of the ECG signals to be detected are abnormal ECG signals.

S1025: Input the third ECG signal segment and the fourth ECG signal segment from the multiple ECG signal segments into the abnormal ECG signal screening module for screening processing, to obtain the third ECG signal segment and the ECG signal segment. Matching degree between fourth ECG signal segments, the first ECG signal segment, the second ECG signal segment, the third ECG signal segment, and the fourth ECG signal segment are adjacent in sequence .

The next ECG signal segment adjacent to the second ECG signal segment is the third ECG signal segment, and the next ECG signal segment adjacent to the third ECG signal segment is the fourth ECG signal segment. That is, the first ECG signal segment, the second ECG signal segment, the third ECG signal segment, and the fourth ECG signal segment are adjacent in sequence.

Exemplarily, as shown in FIG. 3, the four ECG signal segments are, from left to right, the first ECG signal segment, the second ECG signal segment, the third ECG signal segment, and the fourth ECG signal segment . In S1023 and S1024, the first ECG signal segment and the second ECG signal segment are processed, and then the third ECG signal segment and the fourth ECG signal segment are processed. Specifically, the third ECG signal segment and the fourth ECG signal segment are input into the abnormal ECG signal screening module for screening processing to obtain the matching degree between the third ECG signal segment and the fourth ECG signal segment. When it is detected that the matching degree is less than the preset threshold, the third ECG signal segment and the fourth ECG signal segment are respectively marked as abnormal ECG signals, and it is determined that the third ECG signal segment and the fourth ECG signal segment are respectively position in the ECG signal. When it is detected that the matching degree is greater than or equal to the preset threshold, the third ECG signal segment or the fourth ECG signal segment is retained, and the retained ECG signal segment and the fifth ECG signal segment are input to the abnormal ECG signal The screening process is performed in the screening module, and the fourth ECG signal segment is adjacent to the fifth ECG signal segment. This cycle is used until the processing of multiple ECG signal segments corresponding to the ECG signal to be detected is completed. It can be understood that the first ECG signal segment, the second ECG signal segment, the third ECG signal segment, and the fourth ECG signal segment are all used to refer to multiple ECG signal segments that need to be processed. A certain ECG signal segment is for the purpose of explaining the solution more clearly, rather than restricting that it must be the first ECG signal segment, the second ECG signal segment, and so on.

S1026: When it is detected that the matching degree between the third ECG signal segment and the fourth ECG signal segment is less than a preset threshold, compare the third ECG signal segment and the fourth ECG signal segment The segments are respectively marked as abnormal ECG signals, and the respective positions of the third ECG signal segment and the fourth ECG signal segment in the ECG signal are determined.

Exemplarily, when the matching degree is represented by 1 and 0, the preset threshold may be set to 1. For example, the matching degree between the third ECG signal segment and the fourth ECG signal segment is 0. At this time, it is detected that the matching degree 0 is less than the preset threshold value 1, which proves that the third ECG signal segment and the fourth ECG signal are The segment matching degree is low, that is, the similarity degree is low. It can be understood that the third ECG signal segment and the fourth ECG signal segment are two ECG signal segments with large differences, which are probably abnormal ECG signal segments. The third ECG signal segment and the fourth ECG signal segment are respectively marked as abnormal ECG signals.

While marking the third ECG signal segment and the fourth ECG signal segment as abnormal ECG signals, respectively, obtain the difference between the third ECG signal segment and the fourth ECG signal segment in the entire ECG signal to be detected. The position is to obtain the coordinates of the third ECG signal segment and the fourth ECG signal segment in the entire ECG signal to be detected, respectively. Optionally, the position or coordinates can also be represented by the description of the number of segments. For example, the location of the third ECG signal segment in the entire ECG signal to be detected is the third ECG signal segment, and the fourth ECG signal segment. The position of the electrical signal segment in the entire ECG signal to be detected is the fourth ECG signal segment. This is only an exemplary description, and it is not limited.

Optionally, the ECG signal segment marked as abnormal ECG signal may be marked with a rectangular frame, for example, the third ECG signal segment and the first ECG signal segment marked as abnormal ECG signal may be marked with a rectangular frame different from the color of the ECG signal. The four ECG signal segments are framed, so that the doctor can directly check which of the ECG signals to be detected are abnormal ECG signals.

S1027: Analyze the abnormal electrocardiographic signal based on the abnormal electrocardiographic signal classification module to obtain an abnormality type corresponding to the abnormal electrocardiographic signal.

The abnormal ECG signal is analyzed based on the abnormal ECG signal classification module to obtain the abnormal type corresponding to the abnormal ECG signal.

The ECG signal segments marked as abnormal ECG signals are analyzed by the abnormal ECG signal classification module, and the abnormal type corresponding to each abnormal ECG signal is obtained. For example, a feature vector of an abnormal electrocardiogram signal can be extracted by an abnormal electrocardiogram signal classification module, and the feature vector can be classified to obtain an abnormality type corresponding to the abnormal electrocardiogram signal.

Optionally, when the matching degree is greater than or equal to the preset threshold, perform S1028.

S1028: When it is detected that the degree of matching between the first ECG signal segment and the second ECG signal segment is greater than or equal to the preset threshold, put the first ECG signal segment or the second ECG signal segment The second ECG signal segment is retained, and the retained ECG signal segment and the third ECG signal segment are input into the abnormal ECG signal screening module for screening processing, and the retained ECG signal segment and the third ECG signal segment are obtained. The degree of matching between ECG signal segments.

Exemplarily, when the matching degree is represented by 1 and 0, the preset threshold may be set to 1. For example, the matching degree between the first ECG signal segment and the second ECG signal segment is 1, and it is detected that the matching degree 1 is equal to the preset threshold 1, which proves that the first ECG signal segment and the second ECG signal segment The segment matching degree is high, that is, the similarity degree is high. It can be understood that the first ECG signal segment and the second ECG signal segment are two ECG signal segments with small differences, which are probably normal ECG signal segments. One ECG signal segment or the second ECG signal segment is retained, and the retained ECG signal segment is used to continue the screening process with the third ECG signal segment to obtain the retained ECG signal segment and the third ECG signal segment The matching degree between them, the ECG signal segment that is not retained can be eliminated. After obtaining the matching degree between the retained ECG signal segment and the third ECG signal segment, compare the matching degree with the preset threshold value, when the matching degree between the retained ECG signal segment and the third ECG signal segment is When the value is less than the preset threshold, the reserved ECG signal segment and the third ECG signal segment are respectively marked as abnormal ECG signals, and the respective positions of the retained ECG signal segment and the third ECG signal segment in the ECG signal are determined. . When the matching degree between the retained ECG signal segment and the third ECG signal segment is greater than or equal to the preset threshold, the retained ECG signal segment or the third ECG signal segment is retained, and the ECG signal segment retained at this time is The electrical signal fragment and the fourth electrocardiographic signal fragment are input into the abnormal electrocardiographic signal screening module for screening processing to obtain the matching degree between the remaining electrocardiographic signal fragment and the fourth electrocardiographic signal fragment. This process is repeated until all the ECG signal segments in the plurality of ECG signal segments are screened. This is only an exemplary description, and it is not limited.

For example, the first ECG signal segment is retained, and the first ECG signal segment and the third ECG signal segment are input into the abnormal ECG signal screening module for screening processing to obtain the first ECG signal segment and the third ECG signal segment. The degree of matching between the signal segments is determined, and then the above steps are performed cyclically until the processing of the multiple ECG signal segments corresponding to the ECG signal to be detected is completed.

Alternatively, the second ECG signal segment is retained, and the second ECG signal segment and the third ECG signal segment are input into the abnormal ECG signal screening module for screening processing to obtain the second ECG signal segment and the third ECG signal segment. The degree of matching between the signal segments is determined, and then the above steps are performed cyclically until the processing of the multiple ECG signal segments corresponding to the ECG signal to be detected is completed.

Using this method can filter out a large number of identical ECG signal segments, on the one hand, the workload of the subsequent ECG signal detection model for classifying abnormal ECG signals is reduced, and on the other hand, the classification results are made accurate.

Please refer to FIG. 4. FIG. 4 is a schematic flowchart of a method for detecting an ECG signal provided by another embodiment of the present application. Optionally, in a possible implementation manner, as shown in FIG. 4 , the foregoing S1023 may include S10231 to S10233, and the details are as follows:

S10231: Extract a first feature vector corresponding to the first ECG signal segment based on the abnormal ECG signal screening module.

The abnormal ECG signal screening module can be obtained by training based on the twin network. The twin network is actually implemented by two parallel networks with all parameters shared and the same structure, and the network structure adopted by the twin network is not limited. It can be VGG-16 structure, Resnet-50 structure and Resnet-101 structure, etc. The two parallel networks in the twin network process the first ECG signal segment and the second ECG signal segment respectively, extract the first feature vector corresponding to the first ECG signal segment, and extract the first ECG signal segment corresponding to the second ECG signal segment. Two eigenvectors.

In this embodiment, the twin network includes a plurality of convolution layers, the first ECG signal segment can be normalized first, and the normalized first ECG signal segment is passed to the first convolution layer, the first convolution layer performs convolution processing on the first ECG signal segment, extracts features corresponding to the first ECG signal segment, and outputs a feature map based on the extracted features. The first convolutional layer inputs the feature map to the first sampling layer, the first sampling layer performs feature selection on the feature map, removes redundant features, reconstructs a new feature map, and passes the new feature map to the second convolutional layer. The second convolutional layer performs secondary feature extraction on the new feature map, and outputs the feature map again based on the extracted features. The second convolutional layer passes the re-output feature map to the second sampling layer, and the second The sampling layer performs secondary feature selection and reconstructs the feature map. By analogy, the first feature vector corresponding to the first ECG signal segment is obtained after the last sampling layer in the Siamese network completes its processing. This is only an exemplary description, and is not limited thereto.

S10232: Extract the second feature vector corresponding to the second ECG signal segment based on the abnormal ECG signal screening module;

For the specific process of extracting the second eigenvector corresponding to the second ECG signal segment by the abnormal ECG signal screening module, please refer to the specific process of extracting the first eigenvector corresponding to the first ECG signal segment by the abnormal ECG signal screening module in S10231, It will not be repeated here.

S10233: Calculate the similarity between the first feature vector and the second feature vector to obtain the matching degree between the first ECG signal segment and the second ECG signal segment.

Input the first eigenvector and the second eigenvector into the cosine distance formula for calculation, and the obtained value is the matching degree between the first ECG signal segment and the second ECG signal segment. The cosine distance formula is as follows:

In the above formula (1), cosθ represents the degree of matching, and the closer the value of cosθ is to 1, the more similar the first eigenvector and the second eigenvector are, that is, the more similar the first ECG signal segment and the second ECG signal segment are; A represents the first eigenvector, B represents the second eigenvector; i represents the dimension corresponding to the first eigenvector and the second eigenvector, that is, _i in A i represents the dimension corresponding to the first eigenvector, and i in B _i represents The dimension corresponding to the second feature vector.

Optionally, the Pearson correlation coefficient can also be used to determine the matching degree between the first ECG signal segment and the second ECG signal segment. The first feature vector and the second feature vector are input into the preset formula (2) for calculation to obtain the matching degree between the first ECG signal segment and the second ECG signal segment. The preset formula (2) is as follows:

In the above formula (2), X represents the first eigenvector, Y represents the second eigenvector, ρ _{x, y} represents the Pearson correlation coefficient between the first eigenvector and the second eigenvector, which can also be understood as the first eigenvector. The degree of matching between the electrical signal segment and the second ECG signal segment; cov(X, Y) represents the covariance of X and Y, σ _X represents the standard deviation of X, and σ _Y represents the standard deviation of Y.

Optionally, in a possible implementation manner, when the matching degree is represented by 1 and 0, the values calculated by the above two formulas may be mapped to the manner represented by 1 and 0. For example, if the calculated value is within the first preset value range, the matching degree is 1; if the calculated value is within the second preset value range, the matching degree is 0. The first preset value range and the second preset value range are set by the user according to the actual situation, and can be used to assist in judging the degree of similarity between the two ECG signal segments. This is only an exemplary description, and it is not limited.

Please refer to FIG. 5. FIG. 5 is a schematic flowchart of a method for detecting an ECG signal provided by another embodiment of the present application. Optionally, in a possible implementation manner, as shown in FIG. 5 , the foregoing S1027 may include S10271 to S10272, and the details are as follows:

S10271: Extract an abnormal feature vector corresponding to the abnormal ECG signal based on the abnormal ECG signal classification module.

The abnormal ECG signal classification module can be obtained by training based on a Long-Short Term Memory (LSTM) artificial neural network. The LSTM model contains multiple network layers, and the abnormal ECG signal contains 3 heartbeat signals as an example to illustrate. The abnormal ECG signal is input into the LSTM model, the first network layer extracts the feature vector corresponding to the first heartbeat signal, and the feature vector is input into the second network layer; the second network layer extracts the second heartbeat signal The corresponding feature vector, and the feature vector corresponding to the second heart beat signal is fused with the feature vector corresponding to the first heart beat signal and transmitted to the third network layer; the third network layer extracts the feature corresponding to the third heart beat signal. vector, and fuse the feature vector corresponding to the third heartbeat signal with the feature vector output by the second network layer to obtain the abnormal feature vector corresponding to the abnormal ECG signal. It can be understood that, when the abnormal ECG signal contains multiple heartbeat signals, the above steps may be continued until all the heartbeat signals are processed. This is only an exemplary description, and it is not limited.

S10272: Classify the abnormal feature vector to obtain the abnormal type corresponding to the abnormal ECG signal.

The abnormal feature vector extracted in S10271 is passed to the output layer in the LSTM model, that is, passed to the fully connected layer. The normalized exponential function (Softmax function) in the fully connected layer classifies the feature vector passed from the previous network layer to obtain the abnormal type corresponding to the abnormal ECG signal. That is, it is determined which one of atrial fibrillation, first-degree atrioventricular block, left bundle branch block and the like the abnormal electrocardiographic signal belongs to.

In this embodiment of the present application, the ECG signal to be detected is processed based on a pre-trained ECG signal detection model. The abnormal ECG signal screening module in the ECG signal detection model can screen out the abnormal ECG signal from the ECG signal, and determine the position of the abnormal ECG signal, and the classification module in the ECG signal detection model can detect the abnormal ECG signal. The abnormal ECG signal screened by the ECG signal screening module is analyzed to obtain the abnormal type corresponding to the abnormal ECG signal. Based on this method, the abnormal ECG signal in the ECG signal to be detected can be accurately located, which is convenient for the user to quickly and accurately find the position of the abnormal ECG signal in the whole ECG signal, and can also accurately identify the abnormal ECG signal. The abnormal type of the electrical signal is convenient to assist the doctor in diagnosing the patient's disease according to the abnormal type. Based on this method, the speed and accuracy of interpreting the ECG signal are improved.

Please refer to FIG. 6. FIG. 6 is a schematic flowchart of a method for detecting an ECG signal provided by another embodiment of the present application. The method may include S201-S207. For steps S206 to S207 shown in FIG. 6 , reference may be made to the relevant descriptions of S101 to S102 in the embodiment corresponding to FIG. 1 , and for brevity, details are not repeated here. Steps S201 to S205 will be specifically described below.

S201: Obtain a first sample training set; the first sample training set includes a plurality of sample ECG signal segments.

The first sample training set may include multiple sample ECG signal segments. Exemplarily, a large number of sample ECG signals of different users may be collected in advance on the network or in hospitals, etc. These sample ECG signals may include the ECG signals of normal people, and may also include the ECG signals of people with various diseases. . These sample ECG signals are divided into multiple sample ECG signal segments, and correspondingly, multiple abnormal sample ECG signal segments and multiple normal sample ECG signal segments are obtained. For the division manner, reference may be made to the description in S1021, which will not be repeated here.

S202: Perform a screening process on the plurality of sample ECG signal fragments based on the initial screening network to obtain a second sample training set; the sample ECG signal fragments in the second sample training set include abnormal sample ECG signal fragments and normal samples Fragment of ECG signal.

The initial screening network can use a twin network. The sample ECG signal fragments in the second sample training set are obtained after screening, so the obtained second sample training set is filtered out of a large number of the same ECG signal fragments, and also filtered out. A large number of normal sample ECG signal fragments. That is to say, although both the first sample training set and the second sample training set contain abnormal sample ECG signal segments and normal sample ECG signal segments, the abnormal sample ECG signal segments and normal sample ECG signal segments in the second sample training set The number of electrical signal segments is less than the number in the first sample training set, especially the number of normal sample ECG signal segments is far less than the number of normal sample ECG signal segments in the first sample training set.

The specific process of using the initial screening network to screen multiple sample ECG signal fragments may be to use the initial screening network to screen multiple sample ECG signal fragments corresponding to each sample ECG signal. For details, please refer to The process of screening and processing the ECG signal by the abnormal ECG signal screening module in S102 will not be repeated here.

Optionally, in a possible implementation manner, in order to ensure that the number of abnormal sample ECG signal segments and normal sample ECG signal segments is balanced, an appropriate ECG signal segment can be selected according to the number of abnormal sample ECG signal segments obtained after screening. The number of normal sample ECG signal fragments. In this way, when the ECG signal detection model is trained, the model can learn various types of features, thereby making the classification result of the ECG signal detection model obtained by training more accurate.

S203: Label each sample ECG signal segment in the second sample training set to obtain a classification type corresponding to each sample ECG signal segment.

The manual labeling method is used to label each sample ECG signal segment in the second sample training set, that is, label the classification type corresponding to each sample ECG signal segment.

S204: Train an initial classification network based on each sample ECG signal segment in the second sample training set and the classification type corresponding to each sample ECG signal segment, and update parameters of the initial classification network based on the training result.

The initial classification network can use the LSTM network. Exemplarily, the LSTM network is used to classify each sample ECG signal segment, and output the actual classification type corresponding to each sample ECG signal segment. For the specific classification process, refer to the description in S1027 above, which is not repeated here.

The loss value is calculated based on the actual classification type corresponding to each sample ECG signal segment output by the LSTM network and the classification type manually marked for the sample ECG signal segment. Compare the size between the loss value and the preset loss threshold, when the loss value is greater than the preset loss threshold, adjust the parameters of the LSTM network, and continue to train the LSTM network; when the loss value is less than or equal to the preset loss value When the loss threshold is reached, the training is stopped, and the LSTM network has been trained at this time. It can be understood that the LSTM network at this time is the trained abnormal ECG signal classification module.

S205: When the loss function corresponding to the initial classification network converges, generate the trained ECG signal detection model based on the initial classification network at this time and the initial screening network.

Optionally, in a possible implementation manner, in the process of training the initial classification network, it may also be determined whether the loss function corresponding to the initial classification network has converged. When the loss function corresponding to the initial classification network does not converge, adjust the parameters of the initial classification network and continue to train the initial classification network. When the loss function corresponding to the initial classification network converges, it is equivalent to obtaining a trained abnormal ECG signal classification module. After the initial screening network has been screened for multiple sample ECG signal fragments, the network has become very stable, which is equivalent to obtaining a trained abnormal ECG signal screening module. Based on the abnormal ECG signal screening module and the abnormal ECG signal classification module, a trained ECG signal detection model is constructed.

In the embodiment of the present application, the ECG signal detection model trained in the above manner includes an abnormal ECG signal screening module and an abnormal ECG signal classification module. During use, the ECG signal detection model passes the abnormal ECG signal screening module. A large number of identical ECG signal segments in the ECG signal to be detected can be filtered out, and it can also be understood as filtering out a large number of normal ECG signal segments in the ECG signal to be detected (usually, abnormal ECG signals are in the whole ECG signal. Only part of it, most of them are normal ECG signals). When the abnormal ECG signal classification module classifies and processes the abnormal ECG signals, the workload is reduced, and the processed abnormal type is more accurate.

Optionally, in a possible implementation manner, the ECG signal detection model and the abnormality type corresponding to the abnormal ECG signal may also be uploaded to the blockchain.

In this embodiment, the ECG signal detection model and the abnormal type corresponding to the abnormal ECG signal are uploaded to the blockchain, which can ensure its security and fairness and transparency to users. In addition, the ECG signal detection model and the abnormal type corresponding to the abnormal ECG signal are uploaded to the blockchain. With the feature that the files on the blockchain cannot be tampered with at will, the abnormal ECG signal detection model and the abnormal ECG signal can be avoided. The type is maliciously tampered with, so that the subsequent user can directly and accurately obtain the abnormal type corresponding to the abnormal ECG signal, and it is also convenient for the subsequent user to use the ECG signal detection model.

The blockchain referred to in this example is a new application mode of computer technology such as distributed data storage, point-to-point transmission, consensus mechanism, and encryption algorithm. Blockchain, essentially a decentralized database, is a series of data blocks associated with cryptographic methods. Each data block contains a batch of network transaction information to verify its Validity of information (anti-counterfeiting) and generation of the next block. The blockchain can include the underlying platform of the blockchain, the platform product service layer, and the application service layer.

Please refer to FIG. 7 . FIG. 7 is a schematic diagram of an apparatus for detecting an ECG signal provided by an embodiment of the present application. Each unit included in the apparatus is used to perform each step in the embodiment corresponding to FIG. 1 , FIG. 2 , and FIG. 4 to FIG. 6 . For details, please refer to the relevant descriptions in the corresponding embodiments of FIG. 1 , FIG. 2 , and FIG. 4 to FIG. 6 . For convenience of description, only the parts related to this embodiment are shown. See Figure 7, including:

a first acquiring unit 310, configured to acquire the ECG signal to be detected;

The processing unit 320 is configured to input the ECG signal into the trained ECG signal detection model for processing, and obtain the position of the abnormal ECG signal in the ECG signal and the abnormal type corresponding to the abnormal ECG signal ; wherein, the ECG signal detection model includes an abnormal ECG signal screening module and an abnormal ECG signal classification module, and the abnormal ECG signal screening module is used to screen out the abnormal ECG signal from the ECG signal , and determine the position of the abnormal ECG signal, and the abnormal ECG signal classification module is used for analyzing the abnormal ECG signal to obtain the abnormal type corresponding to the abnormal ECG signal.

Optionally, the processing unit 320 includes:

a dividing unit, configured to divide the ECG signal into a plurality of ECG signal segments;

a selection unit, configured to select a first ECG signal segment and a second ECG signal segment from the plurality of ECG signal segments, where the first ECG signal segment is adjacent to the second ECG signal segment, The first ECG signal segment is any one of the multiple ECG signal segments;

The first screening unit is configured to input the first ECG signal segment and the second ECG signal segment into the abnormal ECG signal screening module for screening processing, and obtain the first ECG signal segment and all the ECG signal segments. matching degree between the second ECG signal segments;

a marking unit, configured to respectively mark the first ECG signal segment and the second ECG signal segment as abnormal ECG signals when the matching degree is detected to be less than a preset threshold, and determine the first ECG signal segment the respective positions of the ECG signal segment and the second ECG signal segment in the ECG signal;

The second screening unit is configured to input the third ECG signal segment and the fourth ECG signal segment among the plurality of ECG signal segments into the abnormal ECG signal screening module for screening processing to obtain the third ECG signal segment degree of matching between the signal segment and the fourth ECG signal segment, the first ECG signal segment, the second ECG signal segment, the third ECG signal segment, and the fourth ECG signal segment The signal segments are adjacent in sequence;

A detection unit, configured to, when it is detected that the matching degree between the third ECG signal segment and the fourth ECG signal segment is less than a preset threshold, compare the third ECG signal segment with the fourth ECG signal segment The ECG signal segments are respectively marked as abnormal ECG signals, and the respective positions of the third ECG signal segment and the fourth ECG signal segment in the ECG signal are determined;

an analysis unit, configured to analyze the abnormal ECG signal based on the abnormal ECG signal classification module to obtain an abnormal type corresponding to the abnormal ECG signal;

a third screening unit, configured to select the first ECG signal when the matching degree between the first ECG signal segment and the second ECG signal segment is greater than or equal to the preset threshold The segment or the second ECG signal segment is retained, and the retained ECG signal segment and the third ECG signal segment are input into the abnormal ECG signal screening module for screening processing to obtain the retained ECG signal segment and the degree of matching between the third ECG signal segment.

Optionally, the first screening unit is specifically used for:

Extract the first feature vector corresponding to the first ECG signal segment based on the abnormal ECG signal screening module;

Extract the second feature vector corresponding to the second ECG signal segment based on the abnormal ECG signal screening module;

The similarity between the first feature vector and the second feature vector is calculated to obtain the matching degree between the first ECG signal segment and the second ECG signal segment.

Optionally, the analysis unit is specifically used for:

Extract the abnormal feature vector corresponding to the abnormal ECG signal based on the abnormal ECG signal classification module;

The abnormal feature vector is classified to obtain the abnormal type corresponding to the abnormal ECG signal.

Optionally, the device further includes:

a second obtaining unit, configured to obtain a first sample training set; the first sample training set includes a plurality of sample ECG signal segments;

A third acquiring unit, configured to perform screening processing on the plurality of sample ECG signal fragments based on the initial screening network to obtain a second sample training set; the sample ECG signal fragments in the second sample training set include abnormal sample ECGs Signal fragments and normal sample ECG signal fragments;

a labeling unit, configured to label each sample ECG signal segment in the second sample training set to obtain a classification type corresponding to each sample ECG signal segment;

A first training unit, configured to train an initial classification network based on each sample ECG signal segment in the second sample training set and the classification type corresponding to each sample ECG signal segment, and update the initial classification network based on the training result The parameters of the classification network;

The second training unit is configured to generate the trained ECG signal detection model based on the initial classification network at this time and the initial screening network when the loss function corresponding to the initial classification network converges.

Optionally, the device further includes:

The uploading unit is configured to upload the ECG signal detection model and the abnormality type corresponding to the abnormal ECG signal to the blockchain.

Please refer to FIG. 8. FIG. 8 is a schematic diagram of a terminal for detecting an ECG signal provided by another embodiment of the present application. As shown in FIG. 8 , the terminal 4 for detecting electrocardiographic signals in this embodiment includes: a processor 40, a memory 41, and computer instructions 42 stored in the memory 41 and executable on the processor 40. When the processor 40 executes the computer instructions 42 , the steps in each of the foregoing embodiments of the method for detecting an ECG signal are implemented, for example, S101 to S102 shown in FIG. 1 . Alternatively, when the processor 40 executes the computer instructions 42, the functions of the units in the foregoing embodiments are implemented, for example, the functions of the units 310 to 320 shown in FIG. 7 .

Illustratively, the computer instructions 42 may be divided into one or more units, and the one or more units are stored in the memory 41 and executed by the processor 40 to complete the present application. The one or more units may be a series of computer instruction segments capable of completing specific functions, and the instruction segments are used to describe the execution process of the computer instruction 42 in the terminal 4 for detecting ECG signals. For example, the computer instruction 42 may be divided into a first obtaining unit and a processing unit, and the specific functions of each unit are as described above.

The terminal that detects the ECG signal may include, but is not limited to, the processor 40 and the memory 41 . Those skilled in the art can understand that FIG. 8 is only an example of the terminal 4 for detecting ECG signals, and does not constitute a limitation on the terminal for detecting ECG signals, and may include more or less components than those shown in the figure, or a combination of certain Some components, or different components, for example, the terminal that detects the ECG signal may also include an input and output terminal, a network access terminal, a bus, and the like.

The so-called processor 40 may be a central processing unit (Central Processing Unit, CPU), or other general-purpose processors, digital signal processors (Digital Signal Processors, DSP), application specific integrated circuits (Application Specific Integrated Circuit, ASIC), Off-the-shelf programmable gate array (Field-Programmable Gate Array, FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. A general purpose processor may be a microprocessor or the processor may be any conventional processor or the like.

The memory 41 may be an internal storage unit of the terminal for detecting ECG signals, such as a hard disk or memory of the terminal for detecting ECG signals. The memory 41 may also be an external storage terminal of the terminal for detecting ECG signals, such as a plug-in hard disk, a smart memory card (Smart Media Card, SMC), a secure digital (Secure Digital, SD) card, flash memory card (Flash Card), etc. Further, the memory 41 may also include both an internal storage unit of the terminal for detecting ECG signals and an external storage terminal. The memory 41 is used to store the computer instructions and other programs and data required by the terminal. The memory 41 can also be used to temporarily store data that has been output or will be output.

The present application also provides computer storage media, which can be non-volatile or volatile. The computer-readable storage medium stores a computer program, and when the computer program is executed by the processor, realizes: acquiring the ECG signal to be detected; inputting the ECG signal into the trained ECG signal detection model for processing, and obtaining The position of the abnormal ECG signal in the ECG signal and the abnormal type corresponding to the abnormal ECG signal; wherein, the ECG signal detection model includes an abnormal ECG signal screening module and an abnormal ECG signal classification module, the The abnormal ECG signal screening module is used to screen out the abnormal ECG signal from the ECG signal, and determine the position of the abnormal ECG signal, and the abnormal ECG signal classification module is used to classify the abnormal ECG signal. The electrical signal is analyzed to obtain the abnormal type corresponding to the abnormal ECG signal.

The present application also provides a computer program product, which, when the computer program product runs on a terminal, enables the terminal to implement the steps of each of the above-mentioned methods for detecting electrocardiographic signals.

The embodiments of the present application also provide a chip or integrated circuit, the chip or integrated circuit includes: a processor for invoking and running a computer program from a memory, so that a terminal installed with the chip or integrated circuit executes the above-mentioned various detection methods The steps of the method of an electrical signal.

The above-mentioned embodiments are only used to illustrate the technical solutions of the present application, but not to limit them; although the present application has been described in detail with reference to the above-mentioned embodiments, those of ordinary skill in the art should understand that: it can still be used for the above-mentioned implementations. The technical solutions recorded in the examples are modified, or some technical features thereof are equivalently replaced; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the spirit scope of the technical solutions in the embodiments of the application, and should be included in the present application. within the scope of protection of the application.

Claims

A method for detecting ECG signals, comprising:

Obtain the ECG signal to be detected;

Inputting the ECG signal into the trained ECG signal detection model for processing to obtain the position of the abnormal ECG signal in the ECG signal and the abnormal type corresponding to the abnormal ECG signal;

Wherein, the ECG signal detection model includes an abnormal ECG signal screening module and an abnormal ECG signal classification module, and the abnormal ECG signal screening module is used to screen out the abnormal ECG signals from the ECG signals, And determine the position of the abnormal ECG signal, and the abnormal ECG signal classification module is used for analyzing the abnormal ECG signal to obtain the abnormal type corresponding to the abnormal ECG signal.
The method according to claim 1, wherein the ECG signal is input into a trained ECG signal detection model for processing to obtain the location of the abnormal ECG signal and the abnormality in the ECG signal. Abnormal types corresponding to ECG signals, including:

dividing the ECG signal into a plurality of ECG signal segments;

A first ECG signal segment and a second ECG signal segment are selected from the plurality of ECG signal segments, the first ECG signal segment is adjacent to the second ECG signal segment, and the first ECG signal segment is adjacent to the second ECG signal segment. The electrical signal segment is any one of the plurality of ECG signal segments;

Inputting the first ECG signal segment and the second ECG signal segment into the abnormal ECG signal screening module for screening processing to obtain the first ECG signal segment and the second ECG signal segment match between;

When it is detected that the matching degree is smaller than a preset threshold, the first ECG signal segment and the second ECG signal segment are respectively marked as abnormal ECG signals, and the first ECG signal segment and the second ECG signal segment are respectively marked as abnormal ECG signals. the respective positions of the second ECG signal segments in the ECG signal;

Inputting the third ECG signal segment and the fourth ECG signal segment in the multiple ECG signal segments into the abnormal ECG signal screening module for screening processing to obtain the third ECG signal segment and the fourth ECG signal segment The degree of matching between ECG signal segments, the first ECG signal segment, the second ECG signal segment, the third ECG signal segment, and the fourth ECG signal segment are adjacent in sequence;

When it is detected that the matching degree between the third ECG signal segment and the fourth ECG signal segment is less than a preset threshold, the third ECG signal segment and the fourth ECG signal segment are respectively Mark the abnormal ECG signal, and determine the respective positions of the third ECG signal segment and the fourth ECG signal segment in the ECG signal;

Analyzing the abnormal ECG signal based on the abnormal ECG signal classification module to obtain the abnormal type corresponding to the abnormal ECG signal;

or,

When it is detected that the matching degree between the first ECG signal segment and the second ECG signal segment is greater than or equal to the preset threshold, the first ECG signal segment or the second ECG signal segment is The electrical signal segment is retained, and the retained ECG signal segment and the third ECG signal segment are input into the abnormal ECG signal screening module for screening processing, and the retained ECG signal segment and the third ECG signal segment are obtained. The degree of matching between signal fragments.
The method according to claim 2, wherein the first ECG signal segment and the second ECG signal segment are input into the abnormal ECG signal screening module for screening processing to obtain the first ECG signal segment. The matching degree between the ECG signal segment and the second ECG signal segment, including:

Extract the first feature vector corresponding to the first ECG signal segment based on the abnormal ECG signal screening module;

Extract the second feature vector corresponding to the second ECG signal segment based on the abnormal ECG signal screening module;

The similarity between the first feature vector and the second feature vector is calculated to obtain the matching degree between the first ECG signal segment and the second ECG signal segment.
The method according to claim 2, wherein the analyzing the abnormal ECG signal based on the abnormal ECG signal classification module to obtain the abnormal type corresponding to the abnormal ECG signal, comprising:

Extract the abnormal feature vector corresponding to the abnormal ECG signal based on the abnormal ECG signal classification module;

The abnormal feature vector is classified to obtain the abnormal type corresponding to the abnormal ECG signal.
The method according to any one of claims 1 to 4, wherein, before acquiring the ECG signal to be detected, the method further comprises:

obtaining a first sample training set; the first sample training set includes a plurality of sample ECG signal segments;

Perform screening processing on the plurality of sample ECG signal segments based on the initial screening network to obtain a second sample training set; the sample ECG signal segments in the second sample training set include abnormal sample ECG signal segments and normal sample ECG signal segments signal fragment;

Marking each sample ECG signal segment in the second sample training set to obtain a classification type corresponding to each sample ECG signal segment;

The initial classification network is trained based on each sample ECG signal segment in the second sample training set and the classification type corresponding to each sample ECG signal segment, and the parameters of the initial classification network are updated based on the training result;

When the loss function corresponding to the initial classification network converges, the trained ECG signal detection model is generated based on the initial classification network at this time and the initial screening network.
The method according to any one of claims 1 to 4, wherein the ECG signal is input into a trained ECG signal detection model for processing to obtain an abnormal ECG signal in the ECG signal. After the location and the abnormal type corresponding to the abnormal ECG signal, the method further includes:

Upload the ECG signal detection model and the abnormality type corresponding to the abnormal ECG signal to the blockchain.
A device for detecting ECG signals, comprising:

a first acquisition unit, used to acquire the ECG signal to be detected;

a processing unit, configured to input the ECG signal into a trained ECG signal detection model for processing, and obtain the position of the abnormal ECG signal in the ECG signal and the abnormal type corresponding to the abnormal ECG signal; Wherein, the ECG signal detection model includes an abnormal ECG signal screening module and an abnormal ECG signal classification module, and the abnormal ECG signal screening module is used to screen out the abnormal ECG signals from the ECG signals, And determine the position of the abnormal ECG signal, and the abnormal ECG signal classification module is used for analyzing the abnormal ECG signal to obtain the abnormal type corresponding to the abnormal ECG signal.
The apparatus of claim 7, wherein the apparatus further comprises:

a second obtaining unit, configured to obtain a first sample training set; the first sample training set includes a plurality of sample ECG signal segments;

A third acquiring unit, configured to perform screening processing on the plurality of sample ECG signal fragments based on the initial screening network to obtain a second sample training set; the sample ECG signal fragments in the second sample training set include abnormal sample ECGs Signal fragments and normal sample ECG signal fragments;

a labeling unit, configured to label each sample ECG signal segment in the second sample training set to obtain a classification type corresponding to each sample ECG signal segment;

A first training unit, configured to train an initial classification network based on each sample ECG signal segment in the second sample training set and the classification type corresponding to each sample ECG signal segment, and update the initial classification network based on the training result The parameters of the classification network;

The second training unit is configured to generate the trained ECG signal detection model based on the initial classification network at this time and the initial screening network when the loss function corresponding to the initial classification network converges.
A terminal for detecting electrocardiographic signals, comprising a memory, a processor, and a computer program stored in the memory and running on the processor, wherein the processor implements when the processor executes the computer program:

Obtain the ECG signal to be detected;

Inputting the ECG signal into the trained ECG signal detection model for processing to obtain the position of the abnormal ECG signal in the ECG signal and the abnormal type corresponding to the abnormal ECG signal;

Wherein, the ECG signal detection model includes an abnormal ECG signal screening module and an abnormal ECG signal classification module, and the abnormal ECG signal screening module is used to screen out the abnormal ECG signals from the ECG signals, And determine the position of the abnormal ECG signal, and the abnormal ECG signal classification module is used for analyzing the abnormal ECG signal to obtain the abnormal type corresponding to the abnormal ECG signal.
The terminal according to claim 9, wherein, when the processor executes the computer program, it further implements:

dividing the ECG signal into a plurality of ECG signal segments;

A first ECG signal segment and a second ECG signal segment are selected from the plurality of ECG signal segments, the first ECG signal segment is adjacent to the second ECG signal segment, and the first ECG signal segment is adjacent to the second ECG signal segment. The electrical signal segment is any one of the plurality of ECG signal segments;

Inputting the first ECG signal segment and the second ECG signal segment into the abnormal ECG signal screening module for screening processing to obtain the first ECG signal segment and the second ECG signal segment match between;

When it is detected that the matching degree is smaller than a preset threshold, the first ECG signal segment and the second ECG signal segment are respectively marked as abnormal ECG signals, and the first ECG signal segment and the second ECG signal segment are respectively marked as abnormal ECG signals. the respective positions of the second ECG signal segments in the ECG signal;

Inputting the third ECG signal segment and the fourth ECG signal segment in the multiple ECG signal segments into the abnormal ECG signal screening module for screening processing to obtain the third ECG signal segment and the fourth ECG signal segment The degree of matching between ECG signal segments, the first ECG signal segment, the second ECG signal segment, the third ECG signal segment, and the fourth ECG signal segment are adjacent in sequence;

When it is detected that the matching degree between the third ECG signal segment and the fourth ECG signal segment is less than a preset threshold, the third ECG signal segment and the fourth ECG signal segment are respectively Mark the abnormal ECG signal, and determine the respective positions of the third ECG signal segment and the fourth ECG signal segment in the ECG signal;

Analyzing the abnormal ECG signal based on the abnormal ECG signal classification module to obtain the abnormal type corresponding to the abnormal ECG signal;

or,

When it is detected that the matching degree between the first ECG signal segment and the second ECG signal segment is greater than or equal to the preset threshold, the first ECG signal segment or the second ECG signal segment is The electrical signal segment is retained, and the retained ECG signal segment and the third ECG signal segment are input into the abnormal ECG signal screening module for screening processing, and the retained ECG signal segment and the third ECG signal segment are obtained. The degree of matching between signal fragments.
The terminal of claim 10, wherein, when the processor executes the computer program, it further implements:

Extract the first feature vector corresponding to the first ECG signal segment based on the abnormal ECG signal screening module;

Extract the second feature vector corresponding to the second ECG signal segment based on the abnormal ECG signal screening module;

The similarity between the first feature vector and the second feature vector is calculated to obtain the matching degree between the first ECG signal segment and the second ECG signal segment.
The terminal of claim 10, wherein, when the processor executes the computer program, it further implements:

Extract the abnormal feature vector corresponding to the abnormal ECG signal based on the abnormal ECG signal classification module;

The abnormal feature vector is classified to obtain the abnormal type corresponding to the abnormal ECG signal.
The terminal according to any one of claims 9 to 12, wherein, when the processor executes the computer program, it further implements:

obtaining a first sample training set; the first sample training set includes a plurality of sample ECG signal segments;

Perform screening processing on the plurality of sample ECG signal segments based on the initial screening network to obtain a second sample training set; the sample ECG signal segments in the second sample training set include abnormal sample ECG signal segments and normal sample ECG signal segments signal fragment;

Marking each sample ECG signal segment in the second sample training set to obtain a classification type corresponding to each sample ECG signal segment;

The initial classification network is trained based on each sample ECG signal segment in the second sample training set and the classification type corresponding to each sample ECG signal segment, and the parameters of the initial classification network are updated based on the training result;

When the loss function corresponding to the initial classification network converges, the trained ECG signal detection model is generated based on the initial classification network at this time and the initial screening network.
The terminal according to any one of claims 9 to 12, wherein, when the processor executes the computer program, it further implements:

Upload the ECG signal detection model and the abnormality type corresponding to the abnormal ECG signal to the blockchain.
A computer-readable storage medium storing a computer program, wherein the computer program is executed by a processor to realize:

Obtain the ECG signal to be detected;

Inputting the ECG signal into the trained ECG signal detection model for processing, to obtain the position of the abnormal ECG signal in the ECG signal and the abnormal type corresponding to the abnormal ECG signal;

Wherein, the ECG signal detection model includes an abnormal ECG signal screening module and an abnormal ECG signal classification module, and the abnormal ECG signal screening module is used to screen out the abnormal ECG signals from the ECG signals, And determine the position of the abnormal ECG signal, and the abnormal ECG signal classification module is used for analyzing the abnormal ECG signal to obtain the abnormal type corresponding to the abnormal ECG signal.
The computer-readable storage medium of claim 15, wherein the computer program, when executed by the processor, further implements:

dividing the ECG signal into a plurality of ECG signal segments;

A first ECG signal segment and a second ECG signal segment are selected from the plurality of ECG signal segments, the first ECG signal segment is adjacent to the second ECG signal segment, and the first ECG signal segment is adjacent to the second ECG signal segment. The electrical signal segment is any one of the plurality of ECG signal segments;

Inputting the first ECG signal segment and the second ECG signal segment into the abnormal ECG signal screening module for screening processing to obtain the first ECG signal segment and the second ECG signal segment match between;

When it is detected that the matching degree is less than a preset threshold, the first ECG signal segment and the second ECG signal segment are respectively marked as abnormal ECG signals, and the first ECG signal segment and the second ECG signal segment are respectively marked as abnormal ECG signals. the respective positions of the second ECG signal segments in the ECG signal;

Inputting the third ECG signal segment and the fourth ECG signal segment in the multiple ECG signal segments into the abnormal ECG signal screening module for screening processing to obtain the third ECG signal segment and the fourth ECG signal segment The degree of matching between ECG signal segments, the first ECG signal segment, the second ECG signal segment, the third ECG signal segment, and the fourth ECG signal segment are adjacent in sequence;

When it is detected that the matching degree between the third ECG signal segment and the fourth ECG signal segment is less than a preset threshold, the third ECG signal segment and the fourth ECG signal segment are respectively Mark the abnormal ECG signal, and determine the respective positions of the third ECG signal segment and the fourth ECG signal segment in the ECG signal;

Analyzing the abnormal ECG signal based on the abnormal ECG signal classification module to obtain the abnormal type corresponding to the abnormal ECG signal;

or,

When it is detected that the matching degree between the first ECG signal segment and the second ECG signal segment is greater than or equal to the preset threshold, the first ECG signal segment or the second ECG signal segment is The electrical signal segment is retained, and the retained ECG signal segment and the third ECG signal segment are input into the abnormal ECG signal screening module for screening processing, and the retained ECG signal segment and the third ECG signal segment are obtained. The degree of matching between signal fragments.
The computer-readable storage medium of claim 16, wherein the computer program, when executed by the processor, further implements:

Extract the first feature vector corresponding to the first ECG signal segment based on the abnormal ECG signal screening module;

Extract the second feature vector corresponding to the second ECG signal segment based on the abnormal ECG signal screening module;

The similarity between the first feature vector and the second feature vector is calculated to obtain the matching degree between the first ECG signal segment and the second ECG signal segment.
The computer-readable storage medium of claim 16, wherein the computer program, when executed by the processor, further implements:

Extract the abnormal feature vector corresponding to the abnormal ECG signal based on the abnormal ECG signal classification module;

The abnormal feature vector is classified to obtain the abnormal type corresponding to the abnormal ECG signal.
The computer-readable storage medium of any one of claims 15 to 18, wherein the computer program, when executed by the processor, further implements:

obtaining a first sample training set; the first sample training set includes a plurality of sample ECG signal segments;

Perform screening processing on the plurality of sample ECG signal segments based on the initial screening network to obtain a second sample training set; the sample ECG signal segments in the second sample training set include abnormal sample ECG signal segments and normal sample ECG signal segments signal fragment;

Marking each sample ECG signal segment in the second sample training set to obtain a classification type corresponding to each sample ECG signal segment;

The initial classification network is trained based on each sample ECG signal segment in the second sample training set and the classification type corresponding to each sample ECG signal segment, and the parameters of the initial classification network are updated based on the training result;

When the loss function corresponding to the initial classification network converges, the trained ECG signal detection model is generated based on the initial classification network at this time and the initial screening network.
The computer-readable storage medium of any one of claims 15 to 18, wherein the computer program, when executed by the processor, further implements:

Upload the ECG signal detection model and the abnormality type corresponding to the abnormal ECG signal to the blockchain.