Abnormal electrocardiogram recognition method, device, equipment, medium and product

Abstract

The present application discloses an abnormal ECG recognition method, device, equipment, medium and product, which relates to the field of ECG detection. The method comprises: filtering the original ECG data to be recognized to obtain the ECG data after filtering; extracting features from the ECG data after filtering to obtain ECG feature data; the ECG feature data comprises basic ECG features and target ECG features, and the target ECG features are obtained by extracting features from the ECG data after filtering using a normal distribution probability density function and a histogram probability density curve; the ECG feature data are input into an abnormal ECG recognition model to obtain an abnormal ECG recognition result; the abnormal ECG recognition model is a trained machine learning model. The present application can improve the accuracy of abnormal ECG signal recognition.

Description

Technical Field

[0001] The present application relates to the field of electrocardiogram (ECG) detection technology, and in particular to an abnormal ECG identification method, device, equipment, medium and product. Background Art

[0002] In recent years, ECG monitoring technology has developed rapidly, and electrocardiogram (ECG) has been widely used as a non-invasive, simple and effective means of monitoring heart health. However, the interpretation of ECG usually requires professional medical knowledge and experience. For non-professionals or when resources are limited, it is particularly important to achieve automated and accurate ECG analysis.

[0003] Existing ECG analysis methods mainly rely on traditional feature extraction and classification algorithms. However, these methods still lack accuracy and robustness when processing complex ECG signals. Summary of the invention

[0004] The purpose of this application is to provide an abnormal ECG recognition method, device, equipment, medium and product, which can improve the accuracy of abnormal ECG signal recognition.

[0005] To achieve the above objectives, this application provides the following solutions:

[0006] In a first aspect, the present application provides an abnormal electrocardiogram recognition method, comprising:

[0007] Performing filtering on the original ECG data to be identified to obtain filtered ECG data;

[0008] Performing feature extraction on the filtered ECG data to obtain ECG feature data; the ECG feature data includes basic ECG features and target ECG features, and the target ECG features are obtained by performing feature extraction on the filtered ECG data using a normal distribution probability density function and a histogram probability density curve;

[0009] The ECG feature data is input into an abnormal ECG recognition model to obtain an abnormal ECG recognition result; the abnormal ECG recognition model is a trained machine learning model.

[0010] Optionally, feature extraction is performed on the filtered ECG data to obtain ECG feature data, specifically including:

[0011] The filtered ECG data is sampled at a set interval to determine the absolute value of the signal at each sampling point;

[0012] Count the number of absolute values ​​of the signal at each sampling point that are greater than the average value and less than the maximum value, and record it as the signal statistics value;

[0013] According to the sampling order, the sampling points whose signal absolute values ​​are less than the set threshold are assigned 0, and the sampling points whose signal absolute values ​​are greater than the set threshold are assigned 1, and a binary string is obtained;

[0014] Calculate the coefficient of variation of the binary string and the frequency of occurrence of 1;

[0015] Calculating the normalized complexity of the binary string;

[0016] The ECG data after filtering is multiplied by a Hamming window and then subjected to discrete Fourier transform, and the amplitude spectrum area of ​​the data after discrete Fourier transform is calculated;

[0017] Calculate the frequency corresponding to the maximum value of the power spectrum density of the discrete Fourier transformed data within the first set frequency range;

[0018] Determine a second set frequency range according to the frequency corresponding to the maximum value of the power spectrum density;

[0019] Calculate the amplitude sum of the discrete Fourier transformed data within a second set frequency range;

[0020] Calculate the standard deviation of the time complexity of the filtered ECG data;

[0021] The signal statistics, the coefficient of variation, the frequency of occurrence of 1, the normalized complexity, the amplitude spectrum area, the amplitude sum and the standard deviation of the time complexity constitute the basic electrocardiographic characteristics.

[0022] Optionally, the first set frequency range is 0.5 Hz to 9 Hz, the second set frequency range is 0.7F to 1.4F, and F is a frequency corresponding to a maximum value of a power spectrum density within the first set frequency range.

[0023] Optionally, feature extraction is performed on the filtered ECG data to obtain ECG feature data, which specifically includes:

[0024] Calculate the normal distribution probability density function of the filtered ECG data;

[0025] Generate a histogram of the filtered ECG data;

[0026] Calculating the count of each interval in the histogram and normalizing the probability density of each interval count;

[0027] Calculate the center position of each interval after the probability density is normalized;

[0028] According to the counts of each interval and each center position after probability density normalization, a histogram probability density curve is fitted using a linear interpolation method;

[0029] Calculate the area of ​​the histogram probability density curve under the normal distribution probability density curve, and record it as a first area;

[0030] Calculate the area of ​​the histogram probability density curve above the normal distribution probability density curve, and record it as the second area;

[0031] The ratio of the first area to the second area is used as the target ECG feature.

[0032] Optionally, filtering the original ECG data to be identified to obtain filtered ECG data specifically includes:

[0033] The original ECG data to be identified is subjected to bandpass filtering to obtain filtered ECG data; the starting frequency of the bandpass filtering is 0.7 Hz and the cutoff frequency is 30 Hz.

[0034] Optionally, the machine learning model is a random forest model, a decision tree model, an XGBoost model or an LGB model.

[0035] In a second aspect, the present application provides an abnormal electrocardiogram recognition device, comprising:

[0036] A filtering processing module is used to filter the original ECG data to be identified to obtain the ECG data after filtering;

[0037] A feature extraction module is used to extract features from the filtered ECG data to obtain ECG feature data; the ECG feature data includes basic ECG features and target ECG features, and the target ECG features are obtained by extracting features from the filtered ECG data using a normal distribution probability density function and a histogram probability density curve;

[0038] The abnormal ECG recognition module is used to input the ECG feature data into an abnormal ECG recognition model to obtain an abnormal ECG recognition result; the abnormal ECG recognition model is a trained machine learning model.

[0039] In a third aspect, the present application provides a computer device, comprising: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the steps of any of the above-described abnormal ECG identification methods.

[0040] In a fourth aspect, the present application provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of any of the above-mentioned abnormal ECG identification methods.

[0041] In a fifth aspect, the present application provides a computer program product, including a computer program, which, when executed by a processor, implements the steps of any of the abnormal ECG identification methods described above.

[0042] According to the specific embodiments provided in this application, this application discloses the following technical effects:

[0043] The present application provides an abnormal ECG recognition method, device, equipment, medium and product, which perform feature extraction on filtered ECG data to obtain ECG feature data, wherein the ECG feature data includes not only basic ECG features but also target ECG features. The target ECG features are obtained by extracting features from the filtered ECG data using a normal distribution probability density function and a histogram probability density curve, which can reflect subtle changes in ECG signals and thus improve the accuracy of abnormal ECG signal recognition. BRIEF DESCRIPTION OF THE DRAWINGS

[0044] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings required for use in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0045] Figure 1 A flowchart of an abnormal ECG recognition method provided in one embodiment of the present application;

[0046] Figure 2 A schematic diagram of a feature extraction process provided in an embodiment of the present application;

[0047] Figure 3 A schematic diagram of functional modules of an abnormal ECG recognition device provided in one embodiment of the present application;

[0048] Figure 4 A schematic diagram of the structure of a computer device provided in one embodiment of the present application. DETAILED DESCRIPTION

[0049] The following will be combined with the drawings in the embodiments of the present application to clearly and completely describe the technical solutions in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.

[0050] In order to make the above-mentioned objects, features and advantages of the present application more obvious and easy to understand, the present application is further described in detail below with reference to the accompanying drawings and specific implementation methods.

[0051] The present application provides an abnormal ECG recognition method, which is executed by a computer device, and can be executed by a computer device such as a terminal or a server alone, or by a terminal and a server together, such as Figure 1 As shown, an abnormal ECG identification method includes the following steps 101 to 103.

[0052] Step 101: filter the original ECG data to be identified to obtain filtered ECG data.

[0053] Step 102: extract features from the filtered ECG data to obtain ECG feature data; the ECG feature data includes basic ECG features and target ECG features, and the target ECG features are obtained by extracting features from the filtered ECG data using a normal distribution probability density function and a histogram probability density curve.

[0054] Step 103: input the ECG feature data into an abnormal ECG recognition model to obtain an abnormal ECG recognition result; the abnormal ECG recognition model is a trained machine learning model.

[0055] The original ECG data to be identified is the original ECG data of a preset time length. The original ECG data is a signal amplitude sequence, and the original ECG data is recorded by an ECG curve with the horizontal axis being time and the vertical axis being voltage.

[0056] The abnormal ECG recognition result is abnormal ECG or normal ECG.

[0057] Existing methods often fail to fully capture the key features in ECG signals, resulting in poor classification performance. Based on this, this application performs target ECG feature extraction, making full use of the subtle changes and implicit information of ECG signals, thereby improving the model's sensitivity and recognition accuracy to abnormal ECG signals. This new feature enhances the detection capability of ECG abnormalities, effectively reduces the misjudgment rate, and further improves the reliability of classification results.

[0058] Wherein, step 101 specifically includes: performing bandpass filtering on the original ECG data to be identified to obtain the ECG data after filtering; the starting frequency of the bandpass filtering is 0.7 Hz, and the cutoff frequency is 30 Hz.

[0059] The extraction process of basic ECG features in step 102 specifically includes:

[0060] The filtered ECG data is sampled at a set interval to determine the absolute value of the signal at each sampling point.

[0061] The number of absolute values ​​of the signal at each sampling point that are greater than the average value and less than the maximum value is counted and recorded as the signal statistical value (feature count2).

[0062] According to the sampling order, the sampling points whose absolute values ​​of the signals are less than the set threshold are assigned a value of 0, and the sampling points whose absolute values ​​of the signals are greater than the set threshold are assigned a value of 1, and a binary string is obtained.

[0063] The coefficient of variation of the binary string and the frequency of occurrence of 1 are calculated and recorded as feature Covarying and feature Grebin respectively.

[0064] The normalized complexity of the binary string is calculated and recorded as feature Comp.

[0065] The filtered ECG data is multiplied by a Hamming window and then discrete Fourier transformed. The amplitude spectrum area of ​​the data after discrete Fourier transform is calculated and recorded as feature AMSA.

[0066] Calculate the frequency corresponding to the maximum value of the power spectrum density of the discrete Fourier transformed data within the first set frequency range.

[0067] A second set frequency range is determined according to the frequency corresponding to the maximum value of the power spectrum density.

[0068] The amplitude sum of the discrete Fourier transformed data within the second set frequency range is calculated and recorded as feature A2.

[0069] Calculate the standard deviation of the time complexity of the ECG data after filtering, recorded as feature std_TCI.

[0070] The signal statistics, the coefficient of variation, the frequency of occurrence of 1, the normalized complexity, the amplitude spectrum area, the amplitude sum and the standard deviation of the time complexity constitute the basic electrocardiographic characteristics.

[0071] The first set frequency range is 0.5 Hz to 9 Hz, the second set frequency range is 0.7F to 1.4F, and F is the frequency corresponding to the maximum value of the power spectrum density in the first set frequency range.

[0072] The target ECG feature extraction process in step 102 specifically includes:

[0073] Calculating the normal distribution probability density function of the filtered ECG data specifically includes: calculating the mean and standard deviation based on the filtered ECG data, the formula is as follows:

[0074]

[0075] Among them, u is the mean, σ is the standard deviation, N is the total number of sampling points, and x j is the ECG signal value of the jth sampling point.

[0076] The normal distribution probability density function of the data is calculated by the mean and standard deviation. The formula is as follows:

[0077]

[0078] Where x represents the variable of the ECG signal.

[0079] Generate a histogram of the ECG data after filtering. Calculate the count of each interval in the histogram and normalize the probability density of each interval count. The calculation result is recorded as p i , record the boundary values ​​e of each interval of the histogram of the array variable x i . p i is the count of the ith interval.

[0080] Calculate the center position of each interval after the probability density is normalized. For the interval boundaries {e1,e2,e3,…,e n+1}, the center position of each interval is calculated by the following formula:

[0081]

[0082] Among them, c i represents the center position of the i-th interval; e i represents the left boundary of the i-th interval; e i+1 represents the right boundary of the i-th interval, and n is the number of intervals.

[0083] According to the counts of each interval after probability density normalization and the center positions, a linear interpolation method is used to fit the histogram probability density curve, specifically including: using a linear interpolation method to fit the probability density p of each interval count i The n points are expanded to 1000 points, denoted as p ii , the center position c of each interval is calculated by linear interpolation method i The n points are expanded to 1000 points, denoted as c ii , with the interpolated p ii and c ii Fit a smooth histogram probability density curve g(x).

[0084] Calculate the area of ​​the histogram probability density curve under the normal distribution probability density curve, and record it as the first area:

[0085]

[0086] Calculate the area of ​​the histogram probability density curve above the normal distribution probability density curve, and record it as the second area:

[0087]

[0088] Among them, area_below represents the first area, and area_above represents the second area.

[0089] The ratio of the first area to the second area is used as the target ECG feature:

[0090]

[0091] Among them, A_ratio represents the target ECG feature.

[0092] The machine learning model is a random forest model, a decision tree model, an XGBoost model, or an LGB model.

[0093] The beneficial effects of the present application are as follows: by establishing target features, abnormal ECG can be identified more accurately, and by entering the target features into the original machine learning model for training, a more accurate abnormal ECG recognition model can be obtained.

[0094] Based on the same inventive concept, the embodiment of the present application also provides an abnormal ECG recognition device for implementing the abnormal ECG recognition method involved above. The implementation scheme for solving the problem provided by the device is similar to the implementation scheme recorded in the above method, so the specific limitations in one or more abnormal ECG recognition device embodiments provided below can refer to the limitations of the abnormal ECG recognition method above, and will not be repeated here.

[0095] In an exemplary embodiment, Figure 3 As shown, an abnormal ECG identification device is provided, including: a filtering processing module 201, which is used to filter the original ECG data to be identified to obtain the filtered ECG data.

[0096] The feature extraction module 202 is used to extract features from the filtered ECG data to obtain ECG feature data; the ECG feature data includes basic ECG features and target ECG features, and the target ECG features are obtained by extracting features from the filtered ECG data using a normal distribution probability density function and a histogram probability density curve.

[0097] The abnormal ECG recognition module 203 is used to input the ECG feature data into an abnormal ECG recognition model to obtain an abnormal ECG recognition result; the abnormal ECG recognition model is a trained machine learning model.

[0098] In an exemplary embodiment, a computer device is provided. The computer device may be a server or a terminal. The internal structure diagram thereof may be as follows: Figure 4As shown. The computer device includes a processor, a memory, an input / output interface (Input / Output, referred to as I / O) and a communication interface. The processor, the memory and the input / output interface are connected through a system bus, and the communication interface is connected to the system bus through the input / output interface. The processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program and a database. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The database of the computer device is used to store abnormal ECG recognition data. The input / output interface of the computer device is used to exchange information between the processor and an external device. The communication interface of the computer device is used to communicate with an external terminal through a network connection. When the computer program is executed by the processor, an abnormal ECG recognition method is implemented.

[0099] Those skilled in the art will understand that Figure 4 The structure shown in the figure is only a block diagram of a part of the structure related to the solution of the present application, and does not constitute a limitation on the computer device to which the solution of the present application is applied. The specific computer device may include more or fewer components than those shown in the figure, or combine certain components, or have a different arrangement of components. In an exemplary embodiment, a computer device is provided, including a memory and a processor, wherein a computer program is stored in the memory, and the processor implements the steps in the above-mentioned method embodiments when executing the computer program.

[0100] In an exemplary embodiment, a computer-readable storage medium is provided, storing a computer program, and when the computer program is executed by a processor, the steps in the above method embodiments are implemented.

[0101] In an exemplary embodiment, a computer program product is provided, including a computer program, and when the computer program is executed by a processor, the steps in the above method embodiments are implemented.

[0102] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, stored data, displayed data, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties, and the collection, use and processing of relevant data must comply with relevant regulations.

[0103] Those of ordinary skill in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be completed by instructing the relevant hardware through a computer program, and the computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to the memory, database or other medium used in the embodiments provided in the present application can include at least one of non-volatile and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetic random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. By way of illustration and not limitation, RAM may be in various forms, such as static random access memory (SRAM) or dynamic random access memory (DRAM).

[0104] The database involved in each embodiment provided in this application may include at least one of a relational database and a non-relational database. The non-relational database may include a distributed database based on blockchain, etc., but is not limited thereto. The processor involved in each embodiment provided in this application may be a general-purpose processor, a central processing unit, a graphics processor, a digital signal processor, a data processing logic of a programmable logic device, etc., but is not limited thereto.

[0105] The technical features of the above embodiments may be combined arbitrarily. To make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0106] This article uses specific examples to illustrate the principles and implementation methods of this application. The description of the above embodiments is only used to help understand the method and core ideas of this application. At the same time, for those skilled in the art, according to the ideas of this application, there will be changes in the specific implementation methods and application scope. In summary, the content of this specification should not be understood as limiting this application.

Claims

1. A method for identifying abnormal electrocardiogram, characterized in that: The abnormal electrocardiogram recognition method comprises: Performing filtering on the original ECG data to be identified to obtain filtered ECG data; Performing feature extraction on the filtered ECG data to obtain ECG feature data; the ECG feature data includes basic ECG features and target ECG features, and the target ECG features are obtained by performing feature extraction on the filtered ECG data using a normal distribution probability density function and a histogram probability density curve; Inputting the ECG feature data into an abnormal ECG recognition model to obtain an abnormal ECG recognition result; the abnormal ECG recognition model is a trained machine learning model; Perform feature extraction on the filtered ECG data to obtain ECG feature data, which specifically includes: Calculate the normal distribution probability density function of the filtered ECG data; Generate a histogram of the filtered ECG data; Calculating the count of each interval in the histogram and normalizing the probability density of each interval count; Calculate the center position of each interval after the probability density is normalized; According to the counts of each interval and each center position after probability density normalization, a histogram probability density curve is fitted using a linear interpolation method; Calculate the area of ​​the histogram probability density curve under the normal distribution probability density curve, and record it as a first area; Calculate the area of ​​the histogram probability density curve above the normal distribution probability density curve, and record it as the second area; The ratio of the first area to the second area is used as the target ECG feature.

2. The abnormal ECG identification method according to claim 1, characterized in that: The feature extraction is performed on the ECG data after filtering to obtain ECG feature data, which specifically includes: The filtered ECG data is sampled at a set interval to determine the absolute value of the signal at each sampling point; Count the number of absolute values ​​of the signal at each sampling point that are greater than the average value and less than the maximum value, and record it as the signal statistics value; According to the sampling order, the sampling points whose signal absolute values ​​are less than the set threshold are assigned 0, and the sampling points whose signal absolute values ​​are greater than the set threshold are assigned 1, and a binary string is obtained; Calculate the coefficient of variation of the binary string and the frequency of occurrence of 1; Calculating the normalized complexity of the binary string; The ECG data after filtering is multiplied by a Hamming window and then subjected to discrete Fourier transform, and the amplitude spectrum area of ​​the data after discrete Fourier transform is calculated; Calculate the frequency corresponding to the maximum value of the power spectrum density of the discrete Fourier transformed data within the first set frequency range; Determining a second set frequency range according to the frequency of the maximum value of the power spectrum density; Calculate the amplitude sum of the discrete Fourier transformed data within a second set frequency range; Calculate the standard deviation of the time complexity of the filtered ECG data; The signal statistics, the coefficient of variation, the frequency of occurrence of 1, the normalized complexity, the amplitude spectrum area, the amplitude sum and the standard deviation of the time complexity constitute the basic electrocardiographic characteristics.

3. The abnormal ECG identification method according to claim 2, characterized in that: The first set frequency range is 0.5 Hz to 9 Hz, the second set frequency range is 0.7F to 1.4F, and F is the frequency corresponding to the maximum value of the power spectrum density in the first set frequency range.

4. The abnormal ECG identification method according to claim 1, characterized in that: The original ECG data to be identified is filtered to obtain the filtered ECG data, which specifically includes: The original ECG data to be identified is subjected to bandpass filtering to obtain filtered ECG data; the starting frequency of the bandpass filtering is 0.7 Hz and the cutoff frequency is 30 Hz.

5. The abnormal ECG identification method according to claim 1, characterized in that: The machine learning model is a random forest model, a decision tree model, an XGBoost model, or an LGB model.

6. An abnormal electrocardiogram recognition device, characterized in that: The abnormal electrocardiogram recognition device comprises: A filtering processing module is used to filter the original ECG data to be identified to obtain the ECG data after filtering; A feature extraction module is used to extract features from the filtered ECG data to obtain ECG feature data; the ECG feature data includes basic ECG features and target ECG features, and the target ECG features are obtained by extracting features from the filtered ECG data using a normal distribution probability density function and a histogram probability density curve; An abnormal ECG recognition module, used for inputting the ECG feature data into an abnormal ECG recognition model to obtain an abnormal ECG recognition result; the abnormal ECG recognition model is a trained machine learning model; Perform feature extraction on the filtered ECG data to obtain ECG feature data, which specifically includes: Calculate the normal distribution probability density function of the filtered ECG data; Generate a histogram of the filtered ECG data; Calculating the count of each interval in the histogram and normalizing the probability density of each interval count; Calculate the center position of each interval after the probability density is normalized; According to the counts of each interval and each center position after probability density normalization, a histogram probability density curve is fitted using a linear interpolation method; Calculate the area of ​​the histogram probability density curve under the normal distribution probability density curve, and record it as a first area; Calculate the area of ​​the histogram probability density curve above the normal distribution probability density curve, and record it as the second area; The ratio of the first area to the second area is used as the target ECG feature.

7. A computer device comprising: A memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the abnormal electrocardiogram identification method according to any one of claims 1 to 5.

8. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the abnormal electrocardiogram identification method described in any one of claims 1 to 5 is implemented.

9. A computer program product, comprising a computer program, characterized in that When the computer program is executed by a processor, the abnormal electrocardiogram identification method described in any one of claims 1 to 5 is implemented.

Images