Coronary artery stenosis degree identification method and system based on electrocardio and heart sound signals

Abstract

The invention provides a coronary artery stenosis degree identification method and system based on electrocardio and heart sound signals, and belongs to the technical field of physiological signal analysis. The method comprises the steps that electrocardiosignals and heart sound signals which are to be recognized and synchronously collected are obtained and preprocessed; respectively extracting an amplitude sequence and an interval sequence of the preprocessed electrocardiosignal and heart sound signal; extracting multi-domain features and graphical features based on the amplitude sequence and the interval sequence, and constructing a joint feature set; and inputting the joint feature set into a trained recognition model to obtain a coronary artery stenosis degree recognition result. According to the coronary artery stenosis degree evaluation method, the performance of coronary artery stenosis on two sequences of electrocardio and heart sound signals, amplitude intensity and interval rhythm is fully considered, multi-domain features such as time domain, frequency domain and nonlinearity and detail texture features of a graphical method are extracted, an accurate identification and classification result is obtained, and the coronary artery stenosis degree is accurately evaluated.

Description

Technical Field

[0001] The present invention belongs to the technical field of physiological signal analysis, and relates to a method and system for identifying the degree of coronary stenosis based on electrocardiogram and heart sound signals, and identifying the degree of coronary stenosis by jointly analyzing the amplitude and interval sequences of electrocardiogram and heart sound signals. Background Art

[0002] The statements in this part only provide background technical information related to the present invention and do not necessarily constitute prior art.

[0003] Coronary Artery Disease (CAD) is an important cause leading to the increase in global mortality. Long-term monitoring, early detection, accurate assessment and application of appropriate treatment means can effectively reduce the mortality of CAD. Coronary angiography can accurately measure the degree of coronary stenosis in patients. However, due to its invasive, harmful and expensive drawbacks, it is not suitable for large-scale early screening of CAD. Electrocardiogram and heart sound signals respectively reflect the physiological state of the heart from two aspects of electrical activity and mechanical activity. When coronary stenosis occurs, the above signals will also have pathological changes to varying degrees. Among them, the electrocardiogram signal (ECG) is an electrical signal generated by the depolarization of myocardial cells and contains important physiological information in the human cardiovascular system. In clinical practice, the electrocardiograms of CAD patients often show a series of characteristic changes, including prolongation of QRS complex and QT interval, elevation or depression of ST segment, inversion of T wave, and pathological Q wave, etc. These specific changes not only reflect the pathological process of myocardial ischemia, injury or necrosis, but also provide important reference basis for the clinical diagnosis of CAD. The heart sound signal (PCG) is a mechanical vibration caused by factors such as myocardial contraction and relaxation, opening and closing of heart valves, and blood flow impact. Its intensity and frequency can reflect myocardial contractility and arterial valve changes. For example, the amplitude of the first heart sound is related to the change rate of left ventricular pressure and can reflect the intensity of cardiac contractility. The duration of the heart sound diastolic interval is related to the coronary blood flow reserve and can evaluate whether the blood perfusion time during cardiac diastole is sufficient, etc. In addition, coronary stenosis will cause changes in the blood flow pattern in the coronary artery, resulting in obvious high-frequency murmurs in the diastolic phase of the heart sound signal. Therefore, by analyzing the changes in the interval and amplitude of electrocardiogram and heart sound signals, technical support can be provided for the early non-invasive detection of CAD.

[0004] However, the current analysis of electrocardiogram and heart sound signals mainly focuses on analyzing the changes in the duration of signal intervals, such as heart rate variability analysis based on the electrocardiogram RR interval, etc., missing important complementary information such as the variability characteristics of signal amplitude sequences. In addition, most studies only perform binary classification of disease and non-disease states, while in clinical practice, different treatment strategies need to be adopted according to the specific degree of coronary stenosis of patients. Therefore, it is necessary to make a more detailed classification of different degrees of coronary stenosis. Summary of the Invention

[0005] To overcome the deficiencies of the above-mentioned existing technologies, the present invention provides a method and system for identifying the degree of coronary artery stenosis based on electrocardiogram and heart sound signals, which fully considers the changes in the amplitude intensity and interphase rhythm of the cardiac electrical activity and mechanical activity in the state of coronary heart disease, further excavates the information carried in the two signals, constructs a multi-classification recognition model for coronary artery stenosis of different severities, and improves the comprehensiveness and accuracy of the evaluation.

[0006] To achieve the above object, one or more embodiments of the present invention provide the following technical solutions:

[0007] The first aspect of the present invention provides a method for identifying the degree of coronary artery stenosis based on electrocardiogram and heart sound signals;

[0008] A method for identifying the degree of coronary artery stenosis based on electrocardiogram and heart sound signals, comprising:

[0009] Obtain the synchronously collected electrocardiogram signal and heart sound signal to be identified, and perform preprocessing;

[0010] Extract the amplitude sequence and interphase sequence of the preprocessed electrocardiogram signal and heart sound signal respectively;

[0011] Extract multi-domain features and graphical features based on the amplitude sequence and interphase sequence, and construct a joint feature set;

[0012] Input the joint feature set into the trained recognition model to obtain the recognition result of the degree of coronary artery stenosis.

[0013] As a further technical solution, the preprocessing includes filtering and power frequency notch filtering of the synchronously collected electrocardiogram signal and heart sound signal.

[0014] As a further technical solution, the amplitude sequence of the electrocardiogram signal includes R wave amplitude, P wave amplitude, Q wave amplitude, T wave amplitude, ST segment amplitude sequence; the electrocardiogram signal interphase sequence includes RR interval, QT interval, PR interval and QRS interval sequence;

[0015] The amplitude sequence of the heart sound signal includes the maximum amplitude of the S1 segment, the maximum amplitude of the S2 segment, the maximum amplitude ratio of S1 / S2, the average amplitude of the S1 segment, the average amplitude of the S2 segment, the average amplitude of the systolic phase Sys, the average amplitude of the diastolic phase Dia, the S1 / Sys, S2 / Dia average amplitude ratio sequence; the interphase sequence of the heart sound signal includes S1 interval, S2 interval, systolic interval, diastolic interval, systolic-diastolic interval ratio sequence.

[0016] As a further technical solution, the multi-domain features include time-domain features, frequency-domain features and non-linear features.

[0017] As a further technical solution, graphical analysis is adopted to extract graphical features of the amplitude sequence and the interphase sequence, and the graphical analysis includes extracting image features by using a recurrence plot, a Markov transition field, and a Gramian angular field.

[0018] As a further technical solution, the image features include local binary features, gray-level co-occurrence matrix features, and image statistical features.

[0019] As a further technical solution, the process of inputting the joint feature set into the trained recognition model to obtain the recognition result of the coronary artery stenosis degree is as follows:

[0020] Feature recursive elimination method is used to perform feature selection on the joint feature set, and an optimal feature set is constructed according to the feature selection result;

[0021] The optimal feature set is input into the trained multi-class recognition model to achieve multi-level recognition among different coronary artery stenosis degree groups.

[0022] The second aspect of the present invention provides a coronary artery stenosis degree recognition system based on electrocardiogram and heart sound signals.

[0023] A coronary artery stenosis degree recognition system based on electrocardiogram and heart sound signals includes:

[0024] A signal acquisition module, configured to: acquire synchronous electrocardiogram signals and heart sound signals to be recognized, and perform preprocessing;

[0025] A sequence extraction module, configured to: respectively extract the amplitude sequence and the interphase sequence of the preprocessed electrocardiogram signal and heart sound signal;

[0026] A joint feature set construction module, configured to: extract multi-domain features and graphical features based on the amplitude sequence and the interphase sequence, and construct a joint feature set;

[0027] A classification result recognition module, configured to: input the joint feature set into the trained recognition model to obtain the recognition result of the coronary artery stenosis degree.

[0028] The third aspect of the present invention provides a computer-readable storage medium, on which a program is stored, and when the program is executed by a processor, the steps in a coronary artery stenosis degree recognition method based on electrocardiogram and heart sound signals as described in the first aspect of the present invention are implemented.

[0029] The fourth aspect of the present invention provides an electronic device, including a memory, a processor, and a program stored on the memory and executable on the processor, and when the processor executes the program, the steps in a coronary artery stenosis degree recognition method based on electrocardiogram and heart sound signals as described in the first aspect of the present invention are implemented.

[0030] One or more of the above technical solutions have the following beneficial effects:

[0031] 1. The present invention fully considers the manifestation of coronary artery stenosis in both ECG and heart sound signals, and overcomes the limitations of single signal analysis by jointly analyzing the two signals. By extracting the amplitude sequence and interval sequence of ECG and heart sound signals, the activity state of the heart is characterized from the two perspectives of the intensity and rhythm of the electromechanical activity of the heart, providing complementary information for the identification of different degrees of coronary artery stenosis;

[0032] 2. The present invention uses multiple methods such as multi-domain analysis and graphical analysis to extract features from amplitude and interval sequences, thereby mining more extensive detail information in the signal and providing more effective information for the detection of coronary heart disease;

[0033] 4. Based on the traditional binary classification of disease and health, the present invention introduces the accurate identification of different degrees of coronary artery stenosis, which significantly improves the precision of diagnosis and clinical application value.

[0034] Advantages of additional aspects of the present invention will be given in part in the following description, and in part will become obvious from the following description, or will be learned through practice of the present invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0035] The accompanying drawings in the specification, which constitute a part of the present invention, are used to provide a further understanding of the present invention. The exemplary embodiments of the present invention and their descriptions are used to explain the present invention and do not constitute improper limitations on the present invention.

[0036] Figure 1 This is a flow chart of the method of the first embodiment.

[0037] Figure 2 It is a schematic diagram of the construction process of the recursive graph used in the first embodiment of the present invention.

[0038] Figure 3 It is a schematic diagram of the construction process of the Markov transition field used in the first embodiment of the present invention.

[0039] Figure 4 Schematic diagram of the construction process of the Gram angle field used in the first embodiment of the present invention.

[0040] Figure 5 Schematic diagram of the multi-classification decision boundary of a one-to-many support vector machine used in the first embodiment of the present invention.

[0041] Figure 6 It is a system structure diagram of the second embodiment. DETAILED DESCRIPTION

[0042] It should be noted that the following detailed description is exemplary and is intended to provide further illustration of the present invention. Unless otherwise specified, all technical and scientific terms used herein have the same meaning as commonly understood by those of ordinary skill in the technical field to which the present invention pertains.

[0043] It should be noted that the terms used herein are only for describing specific embodiments and are not intended to limit the exemplary embodiments according to the present invention.

[0044] In the case of no conflict, the embodiments in the present invention and the features in the embodiments can be combined with each other.

[0045] Based on the analysis of various amplitude and interval variabilities of beat-by-beat signals during the cardiac cycle, the cardiac state can be reflected from two aspects of the intensity and rhythm of cardiac electro-mechanical activities, providing important physiological information for the systolic and diastolic functions of the heart, the state of neural regulation, and hemodynamic changes, thereby providing strong support for the early diagnosis of diseases and the formulation of personalized treatment plans. Therefore, the present invention provides a method for identifying the degree of coronary artery stenosis based on the joint analysis of amplitude and interval sequences of electrocardiogram and heart sound signals, fully considering the manifestations of coronary artery stenosis in two signals of electrocardiogram and heart sound, two sequences of amplitude intensity and interval rhythm, extracting multi-domain features such as time domain, frequency domain, and non-linearity, as well as detailed texture features of graphical methods, and accurately evaluating the degree of coronary artery stenosis.

[0046] Embodiment 1

[0047] This embodiment discloses a method for identifying the degree of coronary artery stenosis based on electrocardiogram and heart sound signals;

[0048] As Figure 1 shown, a method for identifying the degree of coronary artery stenosis based on electrocardiogram and heart sound signals includes the following steps:

[0049] Step S1, acquiring the simultaneously collected electrocardiogram signal and heart sound signal to be identified and performing preprocessing;

[0050] In step S1, the electrocardiogram signal and heart sound signal of the subject are simultaneously collected and preprocessed. The preprocessing process includes: using a second-order Butterworth band-pass filter (0.05 - 75 Hz) to remove the noise interference of the electrocardiogram signal, using a polynomial 3rd-order Savitzky-Golay filter to process the baseline drift of the electrocardiogram signal, and using a notch filter (50 Hz) to remove the power frequency interference of the electrocardiogram signal; using a Butterworth high-pass filter (20 Hz) to remove the low-frequency noise interference of the heart sound signal, and using a notch filter (50 Hz) to remove the power frequency interference of the heart sound signal.

[0051] Step S2, respectively extracting the amplitude sequence and interval sequence of the preprocessed electrocardiogram signal and heart sound signal;

[0052] In step S2, for the electrocardiogram signal, the starting point and peak point of the P wave, the starting and ending points and peak point of the QRS wave, and the starting and ending points and peak point of the T wave in the electrocardiogram signal are respectively located, and the amplitude sequence including the R wave amplitude, P wave amplitude, Q wave amplitude, T wave amplitude, and ST segment amplitude sequence, as well as the interval sequence including the RR interval, QT interval, PR interval, and QRS interval sequence are extracted.

[0053] For the heart sound signal, the starting and ending points of the first heart sound (S1), systolic phase, second heart sound (S2), and diastolic phase in each cardiac cycle are located, and the amplitude sequence including the maximum amplitude of the S1 segment, maximum amplitude of the S2 segment, maximum amplitude ratio of S1 / S2, average amplitude of the S1 segment, average amplitude of the S2 segment, average amplitude of the systolic phase Sys, average amplitude of the diastolic phase Dia, S1 / Sys, S2 / Dia average amplitude ratio sequence, and the interval sequence including the S1 interval, S2 interval, systolic interval, diastolic interval, and systolic-diastolic interval ratio sequence are extracted.

[0054] Step S3, extract multi-domain features and graphical features based on the amplitude sequence and interval sequence, and construct a joint feature set;

[0055] Step S31, calculate and extract multi-domain features based on the amplitude sequence and interval sequence, where the multi-domain features include time-domain features, frequency-domain features, and non-linear features. Specifically,

[0056] (1) When extracting time-domain features, calculate the statistical features of the time series and its first-order difference signal, and a total of 13 statistical features including the mean, standard deviation, maximum value, minimum value, difference between the maximum and minimum values, difference between the upper and lower quartiles, coefficient of variation, skewness, kurtosis, difference mean, difference standard deviation, difference maximum value, and difference minimum value are extracted.

[0057] (2) When extracting frequency-domain features, use the Welch periodogram method to correct the power spectral density to calculate the power spectrum of the time series, and extract 19 power spectrum-related features, including the total power of 0 - 1Hz and the equal-interval band power with a window length of 0.1Hz and a step size of 0.05Hz and its ratio to the total power.

[0058] (3) The non-linear features include calculating four non-linear features of the time series, namely sample entropy, fuzzy measure entropy, permutation entropy, and distribution entropy.

[0059] Step S32, calculate and extract graphical features based on the amplitude sequence and interval sequence. Specifically, use recursive graph, Markov transition field, and Gram angle field graphical analysis means to extract the image features of the amplitude sequence and interval sequence, where the image features include local binary features, gray-level co-occurrence matrix features, and image statistical features.

[0060] Further, as Figure 2As shown in the figure, in the process of extracting local binary features using a recurrence plot, first select the embedding dimension \(m = 3\) and the delay time \( = 2\), and map the original time series to the phase space; then calculate the distances between pairwise spatial vectors in the phase space to obtain a distance matrix; directly construct a threshold-free recurrence plot based on the distance matrix, thereby retaining the detailed information in the signal.

[0061] As Figure 3 shown in the figure, the process of constructing a Markov transition field is as follows: select the number of quantile intervals \(Q = 5\), discretize the time series into 5 states, calculate the transition probabilities between each state to obtain a Markov transition matrix; count the values of the elements in the Markov transition matrix and perform normalization processing; finally, arrange them in chronological order and expand to generate a Markov transition field.

[0062] As Figure 4 shown in the figure, the process of constructing a Gram angle field is as follows: first scale the time series to \([0,1]\), then convert the data from the Cartesian coordinate system to the polar coordinate system, identify the temporal correlation of different time points by considering the sum / difference of angles between different points, and construct a Gram angle sum field and a Gram angle difference field respectively according to whether it is the sum or difference of angles.

[0063] Furthermore, extract local binary features, gray-level co-occurrence matrix features, and image statistical features from the constructed recurrence plot, Markov transition field, and Gram angle field respectively. Among them,

[0064] For local binary pattern (LBP) features, compare the gray value of the central pixel with the gray values of its surrounding adjacent pixels. When the gray value of the neighborhood pixel is greater than the center pixel, assign a binary 1; otherwise, assign a binary 0. The obtained binary string is the LBP value. When given a pixel point \((x c ,y c ), the local binary pattern can be expressed as:

[0065]

[0066] In the formula, \(R\) is the pattern radius, \(P\) is the number of neighborhood pixels on the circumference with radius \(R\), \(g c is the gray value of the pixel at the center point, \(g p is the gray value of the pixel on the neighborhood, and \(s\) is the Heaviside function.

[0067] By combining the LBP values of all pixels, the LBP features of the entire image can be formed, and then the texture features of the image can be extracted by calculating the histogram of the LBP pattern. The equivalent local binary pattern improves the traditional LBP so that its features remain consistent under rotational changes. At the same time, it is stipulated that there are at most two jumps between 0 and 1 allowed in the binary pattern (i.e., the pattern type \(U\leq2\)), which is defined as:

[0068]

[0069] Among them, the mapping from LBP P,R to (where the superscript ri represents rotation invariance and u2 represents the equivalent pattern U≤2) generates p + 2 different LBP eigenvalue features.

[0070] For the gray-level co-occurrence matrix (GLCM) features, the gray-level co-occurrence matrix describes the dependence of gray values in the spatial neighborhood by statistically analyzing the spatial relationships between different gray levels in the image. First, the gray levels of the image are compressed to L = 4 levels. Then, with the distance d = 1 between adjacent pixel points, four GLCM matrices are calculated in four different directions (0°, 45°, 90°, and 135°), and four texture features are extracted: energy, correlation, contrast, and homogeneity. The calculation formulas are as follows:

[0071]

[0072]

[0073] In the above formula, P(i,j) is the probability value of the gray-level pair (i,j) appearing in the gray-level co-occurrence matrix, μ x and μ y are the means of the rows and columns of the gray-level co-occurrence matrix respectively, and σ x and σ y are the standard deviations of the rows and columns respectively.

[0074] For the features of the first-order statistics (FOS) of the image, a total of 6 statistical features are included, including the mean, variance, skewness, kurtosis, energy, and entropy value features of the image gray values. The calculation formulas are as follows:

[0075]

[0076] Among them, μ is the mean of the image, σ is the standard deviation of the image, I(x i ) is the gray value of the i-th pixel, p(x i ) is the probability distribution of the gray value x i in the image, and N is the total number of pixels in the image.

[0077] Step S32: Integrate the multi-domain features extracted based on the amplitude sequence and the interphase sequence and the local binary features, gray-level co-occurrence matrix features, and first-order statistics features of the image extracted by constructing recurrence graphs, Markov transition fields, and Gram angular fields to construct a joint feature set.

[0078] Step S4: Input the joint feature set into the trained recognition model to obtain the recognition result of the coronary artery stenosis degree.

[0079] In step S4, the feature recursive elimination method is used for feature selection of the joint feature set, and an optimal feature set is constructed according to the feature selection result. The optimal feature set is input into the multi-classification model of the one-vs-all support vector machine OVA-SVM to achieve multi-level recognition between different coronary artery stenosis degree groups.

[0080] Specifically, OVA-SVM extends the binary support vector machine to a multi-classification task. A binary classification model is trained for each class, regarding this class as the positive class and the remaining classes as the negative class. In the test phase, the input sample will be classified by all binary classification models respectively, and finally the class with the maximum classification score is selected as the final prediction result. The schematic diagram of OVA-SVM is as Figure 5 shown.

[0081] In this embodiment, for the constructed different coronary artery stenosis degree data sets, OVA first regards the samples of severe coronary artery stenosis (sCAD) as the positive class and the samples of the three groups of moderate coronary artery stenosis (moCAD), (mild coronary artery stenosis) miCAD, and healthy people (Health) as the negative class during training, and constructs a training set for classification. Next, by analogy, the moCAD group, miCAD group, and Health group are regarded as the positive class in turn, and the rest are regarded as the negative class, and three other training sets are constructed respectively and the corresponding SVM classifiers are solved, and finally multi-classification is achieved.

[0082] Embodiment 2

[0083] This embodiment discloses a coronary artery stenosis degree recognition system based on electrocardiogram and heart sound signals;

[0084] As Figure 2 shown, a coronary artery stenosis degree recognition system based on electrocardiogram and heart sound signals includes:

[0085] A signal acquisition module, configured to: acquire the synchronously collected electrocardiogram signal and heart sound signal to be recognized, and perform preprocessing;

[0086] A sequence extraction module, configured to: respectively extract the amplitude sequence and the interval sequence of the preprocessed electrocardiogram signal and heart sound signal;

[0087] A joint feature set construction module, configured to: extract multi-domain features and graphical features based on the amplitude sequence and the interval sequence, and construct a joint feature set;

[0088] A classification result recognition module, configured to: input the joint feature set into the trained recognition model to obtain the coronary artery stenosis degree recognition result.

[0089] Embodiment 3

[0090] The purpose of this embodiment is to provide a computer-readable storage medium.

[0091] A computer-readable storage medium stores a computer program, and when the program is executed by a processor, it implements the steps in a method for identifying the degree of coronary artery stenosis based on electrocardiogram and heart sound signals as described in Embodiment 1.

[0092] Embodiment Four

[0093] The purpose of this embodiment is to provide an electronic device.

[0094] An electronic device includes a memory, a processor, and a program stored on the memory and executable on the processor. When the processor executes the program, it implements the steps in a method for identifying the degree of coronary artery stenosis based on electrocardiogram and heart sound signals as described in Embodiment 1.

[0095] The steps involved in the devices in the above Embodiments Two, Three, and Four correspond to those in Method Embodiment 1. For specific implementation manners, reference may be made to the relevant description part of Embodiment 1. The term "computer-readable storage medium" should be understood to include a single medium or multiple media containing one or more instruction sets; it should also be understood to include any medium that can store, encode, or carry an instruction set for execution by a processor and enable the processor to execute any method in the present invention.

[0096] Those skilled in the art should understand that the above-mentioned modules or steps of the present invention can be implemented by a general-purpose computer device. Optionally, they can be implemented by program codes executable by a computing device, so that they can be stored in a storage device and executed by the computing device, or they can be separately fabricated into individual integrated circuit modules, or multiple modules or steps among them can be fabricated into a single integrated circuit module for implementation. The present invention is not limited to any specific combination of hardware and software.

[0097] Although the specific implementation manners of the present invention have been described above in conjunction with the accompanying drawings, this is not a limitation on the protection scope of the present invention. Those skilled in the art should understand that based on the technical solutions of the present invention, various modifications or deformations that can be made without creative efforts by those skilled in the art are still within the protection scope of the present invention.

Claims

1. A method for identifying the degree of coronary artery stenosis based on electrocardiogram and heart sound signals, characterized in that: Acquire synchronously collected electrocardiogram signals and heart sound signals to be identified, and perform preprocessing; Extracting the amplitude sequence and interval sequence of the preprocessed ECG signal and heart sound signal respectively; Extracting multi-domain features and graphical features based on the amplitude sequence and the interval sequence, and constructing a joint feature set; The combined feature set is input into the trained recognition model to obtain a coronary artery stenosis degree recognition result.

2. The method for identifying the degree of coronary artery stenosis based on electrocardiogram and heart sound signals as claimed in claim 1, characterized in that: The preprocessing includes filtering and power frequency notching processing on the synchronously collected electrocardiogram signals and heart sound signals.

3. The method for identifying the degree of coronary artery stenosis based on electrocardiogram and heart sound signals as claimed in claim 1, characterized in that: The amplitude sequence of the electrocardiogram signal includes an R wave amplitude, a P wave amplitude, a Q wave amplitude, a T wave amplitude, and an ST segment amplitude sequence; The electrocardiogram signal interval sequence includes RR interval, QT interval, PR interval and QRS interval sequence; The amplitude sequence of the heart sound signal includes the maximum amplitude of the S1 segment, the maximum amplitude of the S2 segment, the maximum amplitude ratio of S1 / S2, the average amplitude of the S1 segment, the average amplitude of the S2 segment, the average amplitude of the systolic period Sys, the average amplitude of the diastolic period Dia, and the average amplitude ratio sequence of S1 / Sys and S2 / Dia; The interval sequence of the heart sound signal includes an S1 interval, an S2 interval, a systolic interval, a diastolic interval, and a systolic-diastolic interval ratio sequence.

4. The method for identifying the degree of coronary artery stenosis based on electrocardiogram and heart sound signals as claimed in claim 1, characterized in that: The multi-domain features include time domain features, frequency domain features and nonlinear features.

5. The method for identifying the degree of coronary artery stenosis based on electrocardiogram and heart sound signals as claimed in claim 1, characterized in that: Graphical analysis is used to extract graphical features of amplitude sequences and interval sequences, and the graphical analysis includes extracting graphical features using recursive graphs, Markov transition fields, and Gram's angle fields.

6. The method for identifying the degree of coronary artery stenosis based on electrocardiogram and heart sound signals as claimed in claim 5, characterized in that: The imaging features include local binary features, gray-level co-occurrence matrix features and image statistical features.

7. The method for identifying the degree of coronary artery stenosis based on electrocardiogram and heart sound signals as claimed in claim 1, characterized in that: The process of inputting the joint feature set into the trained recognition model to obtain the recognition result of the degree of coronary artery stenosis is as follows: The feature recursive elimination method is used to select the joint feature set, and the optimal feature set is constructed according to the feature selection results; The optimal feature set is input into the trained multi-classification recognition model to achieve multi-level recognition between groups with different degrees of coronary artery stenosis.

8. A system for identifying the degree of coronary artery stenosis based on electrocardiogram and heart sound signals, characterized in that: include: The signal acquisition module is configured to: acquire the synchronously collected ECG signal and heart sound signal to be identified, and perform preprocessing; The sequence extraction module is configured to: extract the amplitude sequence and the interval sequence of the preprocessed electrocardiogram signal and the heart sound signal respectively; A joint feature set construction module is configured to: extract multi-domain features and graphical features based on the amplitude sequence and the interval sequence, and construct a joint feature set; The classification result recognition module is configured to: input the joint feature set into the trained recognition model to obtain the recognition result of the degree of coronary artery stenosis.

9. A computer-readable storage medium having a program stored thereon, characterized in that: When the program is executed by a processor, the steps of a method for identifying the degree of coronary artery stenosis based on electrocardiogram and heart sound signals as described in any one of claims 1 to 7 are implemented.

10. An electronic device comprising a memory, a processor, and a program stored in the memory and executable on the processor, characterized in that: When the processor executes the program, the steps in the method for identifying the degree of coronary artery stenosis based on electrocardiogram and heart sound signals as described in any one of claims 1 to 7 are implemented.

Images