Arrhythmia category identification method based on VAGWO-SVM

Abstract

The invention relates to an arrhythmia identification method. The method comprises the following steps: (1) collecting electrocardiogram data; (2) filtering the ECG signal by adopting complementary ensemble empirical mode decomposition (Complementary ensemble empirical mode decomposition) (Complementary ensemble empirical mode decomposition); (3) main characteristic data P, Q, R, S and T are optimized in combination with an R-wave processing method; (4) dividing the sample data into a training set and a test set by adopting a hold-out method; and (5) learning and training the training set by using the SVM, wherein the test set is used for identifying effect inspection. According to the method, the VAGWO is combined with the SVM to identify arrhythmia types, so that the identification precision and the operation speed are improved, and the method is very suitable for research on arrhythmia.

Description

Technical Field

[0001] The present invention relates to the field of medical and engineering integration, and specifically to a method for identifying arrhythmia categories based on VAGWO-SVM. Background Art

[0002] Arrhythmia is an important group of cardiovascular diseases. It can occur alone or in combination with other cardiovascular diseases. Accurate and rapid diagnosis of arrhythmia is of great significance for the treatment of heart disease. Pattern recognition machine learning algorithms for heart rate identification can effectively make up for the lack of recognition accuracy of traditional diagnostic methods. Complementary EEMD (CEEMD) is an adaptive signal processing method used to decompose complex nonlinear and non-stationary signals into a series of intrinsic mode functions (IMFs) with different frequency characteristics. This technology uses the velocity-assisted grey wolf algorithm (VAGWO) and support vector machine (SVM) algorithm to identify arrhythmia categories, which has broad research significance. Summary of the invention

[0003] The purpose of the present invention is to provide a method for distinguishing arrhythmia classification to achieve reliable identification of arrhythmia. This method has the advantages of being fast and accurate;.

[0004] The present invention achieves the purpose of the invention by adopting the following technical solutions:

[0005] The present invention relates to a method for identifying arrhythmia, comprising the following steps:

[0006] (1) Data preparation: The proportion of ECG data of arrhythmia, heart failure, atrial fibrillation and normal sinus signals is one quarter each;

[0007] (2) Signal preprocessing: ECG signals are filtered using complementary ensemble empirical mode decomposition (CEEND);

[0008] (3) Feature value extraction: Combined with R wave processing method, the main feature data P, Q, R, S, and T are selected;

[0009] (4) Sample set allocation: Use the hold-out method to divide the sample data into training set and test set;

[0010] (5) Establishment of VAGWO-SVM model: Use SVM to learn and train the training set, and use the test set to test the recognition effect. The present invention uses VAGWO combined with SVM to identify arrhythmia types, which improves the recognition accuracy and operation speed, and is very suitable for the study of arrhythmia.

[0011] Preferably, in step (1), the signals used are four different signals, the first being an arrhythmia signal, the second being a heart failure signal, the third being an atrial fibrillation signal, and the fourth being a normal sinus signal, each accounting for one fourth of the total.

[0012] Preferably, in the step (2), the CEEMD decomposition algorithm adopted adds a pair of mutually opposite positive and negative white noises as auxiliary noises into the source signal to eliminate the redundant auxiliary white noise remaining in the reconstructed signal after decomposition by the original EEMD method, while reducing the number of iterations required for decomposition and reducing the computational cost.

[0013] Preferably, in step (4), the training set takes 70% of the total number of samples, and the rest is the test set.

[0014] Preferably, in the step (5), a SVM model is established based on the main information data on the training set, and the arrhythmia recognition effect is analyzed based on the test set.

[0015] Beneficial effects:

[0016] Compared with the prior art, the invention has the following beneficial effects: the CEEMD decomposition algorithm adds a pair of mutually inverse positive and negative white noises as auxiliary noises to the source signal to eliminate the redundant auxiliary white noise remaining in the reconstructed signal after decomposition by the original EEMD method, while reducing the number of iterations required for decomposition and the calculation cost. The invention is combined with the VAGWO-SVM algorithm for the classification of arrhythmias, which improves the recognition accuracy and operation speed, and is very suitable for the research in the field of arrhythmias. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] Figure 1 It is a flow chart of the present invention

[0018] Figure 2 This is the flow chart of VAGWO-SVM. DETAILED DESCRIPTION

[0019] The present invention is further explained below by means of specific examples.

[0020] (1) Data preparation: Four different signals are used: the first is arrhythmia signal, the second is heart failure signal, the third is atrial fibrillation signal, and the fourth is normal sinus signal. The proportion of each signal is one quarter;

[0021] (2) Signal preprocessing: The ECG signal is filtered using complementary ensemble empirical mode decomposition (CEEMD). The CEEMD decomposition algorithm adds a pair of opposite positive and negative white noises as auxiliary noises to the source signal to eliminate the redundant auxiliary white noise remaining in the reconstructed signal after decomposition by the original EEMD method. At the same time, it reduces the number of iterations required for decomposition and reduces the computational cost.

[0022] (3) Feature value extraction: Combined with R wave processing method, the main feature data P, Q, R, S, and T are selected;

[0023] (4) Sample set allocation: Use the hold-out method to divide the sample data into training set and test set;

[0024] (5) Establishment of VAGWO-SVM model: Use SVM to learn and train the training set, and use the test set to test the recognition effect. The present invention uses VAGWO combined with SVM to identify arrhythmia types, which improves the recognition accuracy and operation speed, and is very suitable for the study of arrhythmia.

[0025] First, the velocity term of I wolf guided by alpha wolf, beta wolf and delta wolf is established as follows.

[0026]

[0027] in, Indicates that the update is in progress Under the leadership of ω Wolf The velocity in the dimension, in addition, sgn is the sign function, represents the acceleration term of the wolf, and the calculation formula is as follows:

[0028]

[0029] Among them, w1, w2, w3 are uniformly distributed random numbers generated in the range [0, 1]. In addition, b is a linearly decreasing parameter determined in the following order:

[0030]

[0031] Represents ω wolf pack and The corrected distance between wolves, the distance in each dimension can be calculated as follows:

[0032]

[0033] yes The wolf's position in dimension θ at iteration t, is the position of i in latitude θ at iteration t. is the tth iteration The wolf's position in the θ dimension, is the position of i in latitude θ at the tth iteration. is the key factor multiplied by each leader wolf, used to randomly emphasize or not emphasize the leader wolf, because each There is uncertainty about the fitness of the wolf, especially in the early iterations of the optimization process. These coefficients can take these uncertainties into account and help the algorithm better complete the exploration phase. In VAGWO, new definitions of these coefficients are proposed, and their calculation method is as follows: is a key factor that is multiplied by each leader wolf to randomly emphasize or de-emphasize them, because each There is uncertainty about the fitness of the wolf, especially in the early iterations of the optimization process. These coefficients can take these uncertainties into account and help the algorithm better complete the exploration phase. In VAGWO, new definitions of these coefficients are proposed, and their calculation method is as follows:

[0034]

[0035] Among them, w4, w5, w6 are uniformly distributed random numbers generated in the range of [0, 1], and the adaptive parameters are set as follows:

[0036]

[0037] x is a tuning parameter that acts as an inertia weight, facilitating a properly reliable transition from exploration to exploitation:

[0038]

[0039] Finally, by calculating the three positions of Y, the next position of ω wolf can be updated as follows:

[0040]

[0041] The new position of ω wolf is calculated by averaging the three updated positions shown in equations (15)-(17), as follows:

[0042]

[0043] It will be apparent to those skilled in the art that the invention is not limited to the details of the exemplary embodiments described above and that the invention can be implemented in other specific forms without departing from the spirit or essential features of the invention. Therefore, the embodiments should be considered in all respects as exemplary and non-restrictive, and the scope of the invention is defined by the appended claims rather than the foregoing description, and it is intended that all variations within the meaning and range of equivalents of the claims be included in the invention. Any reference numeral in a claim should not be considered as limiting the claim to which it relates.

[0044] In addition, it should be understood that although the present specification is described according to implementation modes, not every implementation mode contains only one independent technical solution. This description of the specification is only for the sake of clarity. Those skilled in the art should regard the specification as a whole. The technical solutions in each embodiment may also be appropriately combined to form other implementation modes that can be understood by those skilled in the art.

Claims

1. A method for distinguishing arrhythmia categories based on VAGWO-SVM, characterized in that: The following steps are involved: (1) Data preparation: The ECG data of arrhythmia, heart failure, atrial fibrillation and normal sinus signals account for one quarter each; (2) Signal preprocessing: ECG signals are filtered using complementary ensemble empirical mode decomposition (Complementary EEMD); (3) Feature value extraction: Combined with R wave processing method, the main feature data P, Q, R, S, and T are selected; (4) Sample set allocation: Use the hold-out method to divide the sample data into training set and test set; (5) Establishment of VAGWO-SVM model: Use SVM to learn and train the training set, and use the test set to test the recognition effect. The present invention uses VAGWO combined with SVM to identify arrhythmia types, which improves the recognition accuracy and operation speed, and is very suitable for the study of arrhythmia.

2. The method for distinguishing the type of arrhythmia according to claim 1, characterized in that: In the step (1), four different signals are used, the first is an arrhythmia signal, the second is a heart failure signal, the third is an atrial fibrillation signal, and the fourth is a normal sinus signal, and the proportion of each is one quarter.

3. The method for distinguishing arrhythmia types according to claim 1, characterized in that: In the step (2), the CEEMD decomposition algorithm adopts a pair of mutually opposite positive and negative white noises as auxiliary noises to be added to the source signal to eliminate the redundant auxiliary white noise remaining in the reconstructed signal after the original EEMD method decomposition, while reducing the number of iterations required for decomposition and reducing the computational cost.

4. The method for distinguishing arrhythmia types according to claim 1, characterized in that: In the step (3), the main characteristic data are optimized by combining the R-wave processing method.

5. The method for distinguishing arrhythmia types according to claim 1, characterized in that: In step (4), the training set takes 75% of the total number of samples, and the rest is the test set.

6. The method for distinguishing arrhythmia types according to claim 1, characterized in that: In the step (5), the main information data is used to establish an SVM model on the training set, and the arrhythmia recognition effect is analyzed based on the test set, wherein the principle of the speed-assisted gray wolf algorithm is as follows: First, the velocity term of I wolf guided by alpha wolf, beta wolf and delta wolf is established as follows. in, Indicates that the update is in progress Under the leadership of ω Wolf The velocity in the dimension, in addition, sgn is the sign function, represents the acceleration term of the wolf, and the calculation formula is as follows: Among them, w1, w2, w3 are uniformly distributed random numbers generated in the range [0, 1]. In addition, b is a linearly decreasing parameter determined in the following order: Represents ω wolf pack and The corrected distance between wolves, the distance in each dimension can be calculated as follows: yes The wolf is in the tth iteration The location in the dimension, is the number of i in the tth iteration Location in latitude. is the tth iteration Wolf in The position in the dimension. is the key factor multiplied by each leader wolf, used to randomly emphasize or not emphasize the leader wolf, because each There is uncertainty about the fitness of the wolf, especially in the early iterations of the optimization process. These coefficients can take these uncertainties into account and help the algorithm better complete the exploration phase. In VAGWO, new definitions of these coefficients are proposed, and their calculation method is as follows: is a key factor that is multiplied by each leader wolf to randomly emphasize or de-emphasize them, because each There is uncertainty about the fitness of the wolf, especially in the early iterations of the optimization process. These coefficients can take these uncertainties into account and help the algorithm better complete the exploration phase. In VAGWO, new definitions of these coefficients are proposed, and their calculation method is as follows: Among them, w4, w5, w6 are uniformly distributed random numbers generated in the range of [0, 1], and the adaptive parameters are set as follows: x is a tuning parameter that acts as an inertia weight, facilitating a properly reliable transition from exploration to exploitation: Finally, by calculating the three positions of Y, the next position of ω wolf can be updated as follows: The new position of ω wolf is calculated by averaging the three updated positions shown in equations (15)-(17), as follows:

Images