An electrocardiogram waveform classification method, device, electronic device and storage medium

Abstract

The present invention discloses an electrocardiogram waveform classification method, device, electronic device and storage medium, which are used to solve the technical problems of low classification efficiency and poor accuracy of existing ECG signals. The present invention includes: collecting an electrocardiogram signal according to a preset sampling frequency; performing normalization processing on the electrocardiogram signal to obtain normalized data; converting the normalized data into a pulse signal; inputting the pulse signal into a preset S-LRNN model to output a waveform classification result; the S-LRNN model includes an SRNN model with an input layer of an LSTM module.

Description

Technical Field

[0001] The present invention relates to the technical field of waveform analysis, and in particular to an electrocardiogram waveform classification method, apparatus, electronic device and storage medium. Background Art

[0002] An electrocardiogram (ECG) is a reaction of the potential signals on both sides of the myocardial cell membrane of the human body, which can reflect the health status of the human heart and is one of the important bases for doctors to diagnose and treat heart diseases. For a single normal heartbeat cycle, the ECG signal can be divided into the following waveform morphologies: P wave (a small offset before the QRS complex wave representing atrial depolarization), QRS complex wave (the largest amplitude part of the heartbeat), and T wave (a small offset after the QRS complex wave representing ventricular repolarization). After receiving the electrocardiogram of a patient, a doctor first annotates each waveform trend (i.e., waveform segmentation) in order to judge the state of the heart according to the change trend of the waveform. However, for an ECG signal with a sampling time of about ten minutes, the number of heartbeat cycles reaches thousands. Manually annotating each area of the ECG signal by a doctor is a very time-consuming and boring task and may be mislabeled. Although with the development of artificial intelligence technology, there have emerged devices for automatic analysis and classification of ECG signals, it is also necessary to perform preprocessing such as waveform segmentation on the original ECG signal to improve the efficiency and accuracy of ECG signal classification. Summary of the Invention

[0003] The present invention provides an electrocardiogram waveform classification method, apparatus, electronic device and storage medium, which are used to solve the technical problems of low efficiency and poor accuracy in the existing ECG signal classification.

[0004] The present invention provides an electrocardiogram waveform classification method, including:

[0005] Collect an electrocardiogram signal according to a preset sampling frequency;

[0006] Perform normalization processing on the electrocardiogram signal to obtain normalized data;

[0007] Convert the normalized data into a pulse signal;

[0008] Input the pulse signal into an S-LRNN model to output a waveform classification result; the S-LRNN model includes an SRNN model with an input layer being an LSTM module.

[0009] Optionally, the pulse signal includes an incremental pulse signal and a decremental pulse signal; the step of converting the normalized data into a pulse signal includes:

[0010] Divide the normalized data into increasing data and decreasing data;

[0011] Generate an incremental pulse signal using the said increasing data;

[0012] Generate a decremental pulse signal using the said decreasing data.

[0013] Optionally, the step of generating an incremental pulse signal using the said increasing data includes:

[0014] Obtain the first signal intensity of each said increasing data;

[0015] Calculate the first difference between the first signal intensity of the current increasing data and the first signal intensity of the previous increasing data;

[0016] Determine whether the first difference is greater than a preset threshold;

[0017] If so, generate an incremental pulse signal based on the time of the current increasing data.

[0018] Optionally, the step of generating a decremental pulse signal using the said decreasing data includes:

[0019] Obtain the second signal intensity of each said decreasing data;

[0020] Calculate the second difference between the second signal intensity of the current decreasing data and the second signal intensity of the previous decreasing data;

[0021] Determine whether the second difference is greater than a preset threshold;

[0022] If so, generate a decremental pulse signal based on the time of the current decreasing data.

[0023] The present invention also provides an electrocardiogram waveform classification device, including:

[0024] An electrocardiogram signal acquisition module, configured to acquire an electrocardiogram signal according to a preset sampling frequency;

[0025] A normalization processing module, configured to perform normalization processing on the electrocardiogram signal to obtain normalized data;

[0026] A pulse signal conversion module, configured to convert the normalized data into a pulse signal;

[0027] A waveform classification module, configured to input the pulse signal into a preset S-LRNN model and output a waveform classification result; the S-LRNN model includes an SRNN model with an input layer being an LSTM module.

[0028] Optionally, the pulse signal includes an incremental pulse signal and a decremental pulse signal; the pulse signal conversion module includes:

[0029] A division submodule, configured to divide the normalized data into increasing data and decreasing data;

[0030] an incremental pulse signal generating submodule, configured to generate an incremental pulse signal using the incremental data;

[0031] The decrement pulse signal generating submodule is used to generate a decrement pulse signal using the decrement data.

[0032] Optionally, the incremental pulse signal generating submodule includes:

[0033] a first signal strength acquiring unit, configured to acquire a first signal strength of each incremental data;

[0034] A first difference calculation unit, configured to calculate a first difference between a first signal strength of current incremental data and a first signal strength of previous incremental data;

[0035] a first judging unit, configured to judge whether the first difference is greater than a preset threshold;

[0036] The incremental pulse signal generating unit is configured to generate an incremental pulse signal based on the time of the current incremental data.

[0037] Optionally, the decrement pulse signal generating submodule includes:

[0038] a second signal strength acquiring unit, configured to acquire a second signal strength of each of the decreasing data;

[0039] a second difference calculation unit, configured to calculate a second difference between the second signal strength of the current decrement data and the second signal strength of the previous decrement data;

[0040] a second determining unit, configured to determine whether the second difference is greater than a preset threshold;

[0041] The decrement pulse signal generating submodule is configured to generate a decrement pulse signal based on the time of the current decrement data.

[0042] The present invention further provides an electronic device, comprising a processor and a memory:

[0043] The memory is used to store program code and transmit the program code to the processor;

[0044] The processor is configured to execute any one of the above electrocardiogram waveform classification methods according to instructions in the program code.

[0045] The present invention also provides a computer-readable storage medium for storing program codes for executing the electrocardiogram waveform classification method described in any one of the above.

[0046] As can be seen from the above technical solutions, the present invention has the following advantages: The present invention discloses an electrocardiogram waveform classification method, including: collecting an electrocardiogram signal at a preset sampling frequency; performing normalization processing on the electrocardiogram signal to obtain normalized data; converting the normalized data into a pulse signal; inputting the pulse signal into a preset S-LRNN model to output a waveform classification result. In an embodiment of the present invention, by converting the electrocardiogram signal into a pulse signal for waveform classification, the amount of data processing during waveform classification is reduced, thereby greatly reducing the power consumption of the model, enabling the model to have a broader prospect in the hardware deployment of neural morphologies, and improving the classification efficiency of ECG signals. In addition, by using a preset S-LRNN model to classify the pulse signal, the accuracy of waveform classification is improved. Brief Description of the Drawings

[0047] In order to more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the following will briefly introduce the drawings required for use in the description of the embodiments or the prior art. Obviously, the following drawings are only some embodiments of the present invention. For those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative efforts.

[0048] Figure 1 It is a flowchart of the steps of an electrocardiogram waveform classification method provided by an embodiment of the present invention;

[0049] Figure 2 It is a flowchart of the steps of an electrocardiogram waveform classification method provided by another embodiment of the present invention;

[0050] Figure 3 It is a diagram of the ALIF neuron self-recirculation model provided by an embodiment of the present invention; [[ID=2I]]

[0051] Figure 4 It is a curve graph of the Multi-Gaussian function and several common alternative functions provided by an embodiment of the present invention;

[0052] Figure 5 It is a structural block diagram of an electrocardiogram waveform classification device provided by an embodiment of the present invention. Detailed Embodiments [[ID=|3|]]

[0053] Embodiments of the present invention provide an electrocardiogram waveform classification method, device, electronic device, and storage medium for solving the technical problems of low classification efficiency and poor accuracy of existing ECG signals.

[0054] In order to make the objectives, features, and advantages of the present invention more obvious and understandable, the following will clearly and completely describe the technical solutions in the embodiments of the present invention with reference to the accompanying drawings in the embodiments of the present invention. Obviously, the embodiments described below are only a part of the embodiments of the present invention, rather than all the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present invention.

[0055] Please refer to Figure 1 , Figure 1 which is a flowchart of the steps of a method for classifying electrocardiogram waveforms provided in an embodiment of the present invention.

[0056] A method for classifying electrocardiogram waveforms provided by the present invention specifically may include the following steps:

[0057] Step 101, collect electrocardiogram signals at a preset sampling frequency;

[0058] An electrocardiogram (ECG) is a reaction of the potential signals on both sides of the myocardial cell membrane of the human body, which can reflect the health status of the human heart and is one of the important bases for doctors to diagnose and treat heart diseases. It is a technology that uses an electrocardiograph to record the electroactivity change graph generated by each cardiac cycle from the body surface.

[0059] In the embodiments of the present invention, electrocardiogram signals within a certain period of time can be collected at a preset sampling frequency.

[0060] In one example, the selected sampling frequency may be 250 Hz.

[0061] Step 102, perform normalization processing on the electrocardiogram signals to obtain normalized data;

[0062] Normalization is to limit the data to be processed within a certain required range after processing (through a certain algorithm). Normalization is for the convenience of subsequent data processing and, secondly, to ensure that the program converges faster during operation. The specific function of normalization is to summarize and unify the statistical distribution of samples.

[0063] In the embodiments of the present invention, after obtaining the electrocardiogram signals, normalization processing can be performed on the electrocardiogram signals to obtain normalized data.

[0064] Step 103, convert the normalized data into pulse signals;

[0065] A pulse signal is a discrete signal with various shapes. Compared with ordinary analog signals (such as sine waves), the waveforms are discontinuous on the Y-axis (there are obvious intervals between waveforms), but having a certain periodicity is its characteristic. The most common pulse wave is the rectangular wave (i.e., square wave). Pulse signals can be used to represent information or as carriers, such as pulse code modulation (PCM) and pulse width modulation (PWM) in pulse modulation, etc. They can also be used as clock signals for various digital circuits and high-performance chips.

[0066] After completing the normalization of the data, the normalized data can be preprocessed and converted into a pulse signal.

[0067] Step 104: Input the pulse signal into a preset S-LRNN model and output the waveform classification result; the S-LRNN model includes an SRNN model with an input layer being an LSTM module.

[0068] After obtaining the pulse signal, it can be input into a preset S-LRNN model to output the waveform classification result; the S-LRNN model includes an SRNN model with an input layer being an LSTM module, and this LSTM module has 4 inputs.

[0069] In the embodiment of the present invention, since the electrocardiogram signal has three meaningful parts of the cardiac cycle, namely the P wave, the T wave, and the QRS wave complex. Among them, the QRS wave is composed of the Q wave, the R wave, and the S wave. In the monitoring task, the purpose of the embodiment of the present invention is to continuously identify the current wave type. Therefore, the present invention can identify the characteristics of the electrocardiogram signal through a preset S-LRNN model to obtain the waveform classification result. In one example, according to the waveform composition of the electrocardiogram signal, the input signal of the preset S-LRNN model can be a pulse signal, and the output signal can be at least one of the P wave, the PQ wave, the QR wave, the RS wave, the ST wave, and the TP wave. Among them, the pulse signal is first input into a 4-input LSTM layer and then into a 4-36-6 SRNN model to improve the long-term memory ability of the network. At the same time, this network structure has a better classification effect on the processing of ECG signals.

[0070] The present invention discloses an electrocardiogram waveform classification method, including: collecting an electrocardiogram signal according to a preset sampling frequency; performing normalization processing on the electrocardiogram signal to obtain normalized data; converting the normalized data into a pulse signal; inputting the pulse signal into a preset S-LRNN model to output a waveform classification result. In an embodiment of the present invention, by converting the electrocardiogram signal into a pulse signal for waveform classification, the amount of data processing during waveform classification is reduced, thereby greatly reducing the power consumption of the model, enabling the model to have a broader prospect in the hardware deployment of neural morphology, and improving the classification efficiency of ECG signals. In addition, by using the preset S-LRNN model to classify the waveform of the pulse signal, the accuracy of waveform classification is improved.

[0071] Please refer to Figure 2 , Figure 2 which is a flowchart of the steps of an electrocardiogram waveform classification method provided by another embodiment of the present invention. Specifically, it may include the following steps:

[0072] Step 201, collect an electrocardiogram signal according to a preset sampling frequency;

[0073] Step 202, perform normalization processing on the electrocardiogram signal to obtain normalized data;

[0074] Steps 201-202 are the same as steps 101-102, and the specific description can be referred to the description of steps 101-102, which will not be elaborated here.

[0075] Step 203, divide the normalized data into increasing data and decreasing data;

[0076] In an embodiment of the present invention, after completing the normalization of the electrocardiogram signal, level-crossing coding can be performed on the normalized data to convert the normalized data into two categories: increasing data and decreasing data.

[0077] In one example, the normalized data normalized from each electrocardiogram signal can be converted into increasing data and decreasing data. For example, when there is an electrocardiogram signal with two signal channels, two channels of increasing data and two channels of decreasing data can be obtained.

[0078] Increasing data refers to the data where the waveform of the electrocardiogram signal is increasing. Decreasing data refers to the data where the waveform of the electrocardiogram signal is decreasing.

[0079] Step 204, generate an incremental pulse signal using the increasing data;

[0080] In an embodiment of the present invention, after converting the normalized data into increasing data and decreasing data, the increasing data and decreasing data can be sequentially converted into pulse data.

[0081] Among them, the step of generating an incremental pulse signal using incremental data may include the following sub-steps:

[0082] S41, obtain the first signal strength of each incremental data;

[0083] S42, calculate the first difference between the first signal strength of the current incremental data and the first signal strength of the previous incremental data;

[0084] S43, determine whether the first difference is greater than a preset threshold;

[0085] S44, if so, generate an incremental pulse signal based on the time of the current incremental data.

[0086] In a specific implementation, after obtaining multiple incremental data in the electrocardiogram signal, the magnitudes of adjacent two incremental data can be compared in sequence. When the first difference between the first signal strength of the current incremental data and the first signal strength of the previous incremental data is greater than a preset threshold (such as 0.3), an incremental pulse signal can be generated. By counting all the incremental pulse signals, an incremental pulse signal set of the electrocardiogram signal can be obtained.

[0087] Step 205, generate a decremental pulse signal using decremental data;

[0088] Similarly, the embodiments of the present invention can generate a decremental pulse signal through decremental data.

[0089] In an example, the step of generating a decremental pulse signal using decremental data may include the following sub-steps:

[0090] S51, obtain the second signal strength of each decremental data;

[0091] S52, calculate the second difference between the second signal strength of the current decremental data and the second signal strength of the previous decremental data;

[0092] S53, determine whether the second difference is greater than a preset threshold;

[0093] S54, if so, generate a decremental pulse signal based on the time of the current decremental data.

[0094] In a specific implementation, after obtaining multiple decremental data in the electrocardiogram signal, the magnitudes of adjacent two decremental data can be compared in sequence. When the second difference between the second signal strength of the current decremental data and the second signal strength of the previous decremental data is greater than a preset threshold (such as 0.3), a decremental pulse signal can be generated. By counting all the decremental pulse signals, a decremental pulse signal set of the electrocardiogram signal can be obtained.

[0095] Step 206: Input the pulse signal into the preset S-LRNN model to output the waveform classification result. The S-LRNN model includes an SRNN model with an LSTM module as the input layer.

[0096] After obtaining the pulse signal, the pulse signal can be input into the preset S-LRNN model to output the waveform classification result.

[0097] In an embodiment of the present invention, the preset S-LRNN model may include a 4-input LSTM module and an SRNN model.

[0098] LSTM (Long Short-Term Memory) is a type of recurrent neural network designed specifically to address the long-term dependence problem existing in general RNNs (Recurrent Neural Networks).

[0099] In addition to the LSTM module, the SRNN model also includes 46 ALIF (Adaptive Leaky Integrate-and-Fire) neurons, including 4 input layer neurons, 36 hidden layer neurons, and 6 output layer neurons.

[0100] ALIF (Adaptive Leaky Integrate-and-Fire) neurons can improve the performance of its SNN. In adaptive pulse neurons, the threshold increases after each pulse emission and then decays exponentially with a time constant τ adp and is simulated using the forward Euler first-order exponential integrator method for a discrete-time continuous neuron model with dt = 1ms to obtain:

[0101] α = exp(-dt / τ m )

[0102] ρ = exp(-dt / τ adp )

[0103] η t = ρη t-1 + (1 - ρ)S t-1

[0104] θ = b0 + βη t

[0105] u t = αu t-1 + (1 - α)R m I t - S t-1 θ

[0106] where θ is a dynamic threshold composed of a fixed minimum threshold b0 and an adaptive contribution βη t ; ρ uses the time constant τ adpRepresents the single - time - step decay of the threshold. The parameter β is a constant that controls the magnitude of threshold adaptation; we set β to 1.8 as the default value for the adaptive neuron. α represents the single - time - step decay of the membrane potential at the time constant τ m The neuron can be modeled as a self - recurrent neuron model. As Figure 3 shown

[0107] The training process includes two processes: forward propagation and backward propagation.

[0108] Forward propagation: Input the encoded spike data into the input neural layer and calculate within the neuron. When the membrane potential in the neuron The neuron fires a spike, the membrane potential is updated, and the spike is sent to the next - layer neuron. During the process of accumulating the membrane potential, the membrane potential within the neuron and the threshold θ decay with α and ρ respectively. Finally, the membrane potential of the last - layer neuron is used as the output.

[0109] Backward propagation: Use the loss function to calculate the gap between the predicted value and the actual value. For traditional neural networks, the next step is to use gradient descent to optimize network parameters such as weights. To train the S - LRNN, the present invention adopts the Backpropagation Through Time (BPTT) algorithm. BPTT is a commonly used training algorithm for RNN networks. In an RNN, the recurrence of the network induces memory in the form of an internal hidden state h t and the time - stepped input x t updates these states. Therefore, BPTT unfolds the network for all input time steps. For spiking neural networks, due to the spiking mechanism of spiking neurons, it brings the problem of non - differentiability of spikes, making it difficult to apply the chain rule to connect the backward - propagation gradient between the neuron output and the neuron input. Therefore, the present invention adopts the surrogate - function method, using a surrogate function to replace the spike to calculate the gradient. Only in this way can gradient descent be used to optimize the network weights, the τ adp of each spiking neuron, τ m and other parameters.

[0110] Surrogate function: Given a loss function L(t|θ) defined on neural activity at a specific time t, the Backpropagation Through Time (BPTT) algorithm calculates the partial gradient using the chain rule and updates the network parameter θ in the direction of minimizing the loss. Here, the parameter θ includes both synaptic weights and the corresponding neural time constants. In a periodically connected network, past neural activations affect the current loss, and by unfolding the network, the contribution of these past activations to the current loss can be explained.

[0111] The discontinuity of the pulse mechanism of spiking neurons makes it difficult to apply the chain rule to connect the backpropagation gradient between neural outputs and neural inputs; in practice, replacing the discontinuous gradient with a smooth gradient function, the "surrogate gradient" has proven effective and has the additional benefit of allowing spiking neural networks to be mapped to recurrent neural networks in an optimized deep learning framework.

[0112] The present invention herein defines a new surrogate gradient for Multi Gaussian consisting of the weighted sum of multiple Gaussian functions where the selection of the hyperparameters h and s makes the multi-Gaussian contain a negative part:

[0113]

[0114] where u t is the membrane potential of the peak neuron and its internal threshold. The negative parts of the ELU and Leaky RELU functions are considered to mitigate the problem of neuron "death", that is, the sum of all inputs of a neuron is negative and the neuron does not actually participate in network calculations. The shapes of the multi-Gaussian function (MG) and various other surrogate gradient functions are as Figure 4 shown.

[0115] The present invention discloses an electrocardiogram waveform classification method, including: collecting an electrocardiogram signal according to a preset sampling frequency; performing normalization processing on the electrocardiogram signal to obtain normalized data; converting the normalized data into a pulse signal; inputting the pulse signal into a preset S-LRNN model to output a waveform classification result. In the embodiments of the present invention, by converting the electrocardiogram signal into a pulse signal for waveform classification, the amount of data processing during waveform classification is reduced, thereby greatly reducing the power consumption of the model, making the model have a broader prospect in the hardware deployment of neural morphology, and improving the classification efficiency of ECG signals. In addition, by using the preset S-LRNN model to classify the pulse signal, the accuracy of waveform classification is improved.

[0116] Please refer to Figure 5 , Figure 5 which is a structural block diagram of an electrocardiogram waveform classification device provided by an embodiment of the present invention.

[0117] Embodiments of the present invention provide an electrocardiogram waveform classification device, including:

[0118] An electrocardiogram signal acquisition module 501, configured to collect an electrocardiogram signal according to a preset sampling frequency;

[0119] A normalization processing module 502, configured to perform normalization processing on the electrocardiogram signal to obtain normalized data;

[0120] The pulse signal conversion module 503 is used to convert the normalized data into pulse signals;

[0121] The waveform classification module 504 is used to input the pulse signals into a preset S-LRNN model and output waveform classification results; the S-LRNN model includes an SRNN model with an LSTM module as the input layer.

[0122] In the embodiments of the present invention, the pulse signals include incremental pulse signals and decremental pulse signals; the pulse signal conversion module 503 includes:

[0123] The sub-module for division is used to divide the normalized data into increasing data and decreasing data;

[0124] The incremental pulse signal generation sub-module is used to generate incremental pulse signals by using the increasing data;

[0125] The decremental pulse signal generation sub-module is used to generate decremental pulse signals by using the decreasing data.

[0126] In the embodiments of the present invention, the incremental pulse signal generation sub-module includes:

[0127] The first signal strength acquisition unit is used to acquire the first signal strength of each increasing data;

[0128] The first difference calculation unit is used to calculate the first difference between the first signal strength of the current increasing data and the first signal strength of the previous increasing data;

[0129] The first judgment unit is used to judge whether the first difference is greater than a preset threshold;

[0130] The incremental pulse signal generation unit is used to, if so, generate an incremental pulse signal based on the time of the current increasing data.

[0131] In the embodiments of the present invention, the decremental pulse signal generation sub-module includes:

[0132] The second signal strength acquisition unit is used to acquire the second signal strength of each decreasing data;

[0133] The second difference calculation unit is used to calculate the second difference between the second signal strength of the current decreasing data and the second signal strength of the previous decreasing data;

[0134] The second judgment unit is used to judge whether the second difference is greater than a preset threshold;

[0135] The decremental pulse signal generation sub-module is used to, if so, generate a decremental pulse signal based on the time of the current decreasing data.

[0136] The embodiments of the present invention also provide an electronic device, which includes a processor and a memory:

[0137] The memory is used to store program codes and transmit the program codes to the processor;

[0138] The processor is used to execute the electrocardiogram waveform classification method according to the instructions in the program codes of the embodiments of the present invention.

[0139] The embodiments of the present invention also provide a computer-readable storage medium, which is used to store program codes, and the program codes are used to execute the electrocardiogram waveform classification method according to the embodiments of the present invention.

[0140] Those skilled in the art can clearly understand that for the convenience and conciseness of description, the specific working processes of the systems, devices and units described above can refer to the corresponding processes in the foregoing method embodiments, and will not be described herein again.

[0141] Each embodiment in this specification is described in a progressive manner. The key point of each embodiment is to illustrate the differences from other embodiments. The same or similar parts among the various embodiments can be referred to each other.

[0142] Those skilled in the art should understand that the embodiments of the present invention can be provided as a method, a device, or a computer program product. Therefore, the embodiments of the present invention can take the form of an all-hardware embodiment, an all-software embodiment, or an embodiment combining software and hardware aspects. Moreover, the embodiments of the present invention can take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk memories, CD-ROMs, optical memories, etc.) containing computer-usable program codes.

[0143] The embodiments of the present invention are described with reference to the flowcharts and / or block diagrams of the methods, terminal devices (systems), and computer program products according to the embodiments of the present invention. It should be understood that each process and / or block in the flowchart and / or block diagram can be implemented by computer program instructions, and the combination of the processes and / or blocks in the flowchart and / or block diagram can also be implemented by computer program instructions. These computer program instructions can be provided to the processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing terminal devices to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing terminal devices generate a device for implementing the functions specified in Figure 1 one process or multiple processes and / or blocks Figure 1 one block or multiple blocks.

[0144] These computer program instructions can also be stored in a computer-readable memory that can guide a computer or other programmable data processing terminal devices to work in a specific manner, so that the instructions stored in the computer-readable memory generate a manufactured product including an instruction device, and the instruction device implements the functions in the processFigure 1 one process or multiple processes and / or boxes Figure 1 the functions specified in one box or multiple boxes.

[0145] These computer program instructions can also be loaded onto a computer or other programmable data processing terminal device, so that a series of operation steps are executed on the computer or other programmable terminal device to generate a computer-implemented process. Thus, the instructions executed on the computer or other programmable terminal device provide steps for implementing the functions specified in Figure 1 one process or multiple processes and / or boxes Figure 1 one box or multiple boxes.

[0146] Although the preferred embodiments of the embodiments of the present invention have been described, those skilled in the art can make additional changes and modifications once they learn the basic creative concept. Therefore, the appended claims are intended to be construed as including the preferred embodiments and all changes and modifications falling within the scope of the embodiments of the present invention.

[0147] Finally, it should also be noted that in this article, relational terms such as first and second are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the term "comprising", "including" or any other variant thereof is intended to cover non-exclusive inclusion, so that a process, method, article or terminal device comprising a series of elements not only includes those elements, but also includes other elements not expressly listed, or also includes elements inherent to such process, method, article or terminal device. Without further limitation, an element defined by the statement "comprising one..." does not exclude the presence of additional identical elements in the process, method, article or terminal device comprising the said element.

[0148] As described above, the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it; although the present invention has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that they can still modify the technical solutions recorded in the foregoing embodiments, or perform equivalent replacements on some of the technical features; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for classifying electrocardiogram waveforms, characterized in that: include: Collecting electrocardiogram signals according to a preset sampling frequency; performing normalization processing on the electrocardiogram signal to obtain normalized data; converting the normalized data into a pulse signal; The pulse signal is input into a preset S-LRNN model to output a waveform classification result; the S-LRNN model includes an SRNN model whose input layer is an LSTM module; the SRNN model includes 46 adaptive pulse neurons ALIF, including 4 input layer neurons, 36 hidden layer neurons and 6 output layer neurons; The pulse signal includes an incremental pulse signal and a decremental pulse signal; and the step of converting the normalized data into a pulse signal includes: dividing the normalized data into increasing data and decreasing data; generating an incremental pulse signal using the incremental data; A decrement pulse signal is generated using the decrement data.

2. The method according to claim 1, characterized in that The step of generating an incremental pulse signal using the incremental data comprises: Acquire a first signal strength of each of the incremental data; Calculating a first difference between a first signal strength of current incremental data and a first signal strength of previous incremental data; Determining whether the first difference is greater than a preset threshold; If so, an incremental pulse signal is generated based on the time of the current incremental data.

3. The method according to claim 1, characterized in that The step of generating a decrement pulse signal using the decrement data comprises: Acquire a second signal strength of each of the decreasing data; Calculating a second difference between the second signal strength of the current decreasing data and the second signal strength of the previous decreasing data; Determining whether the second difference is greater than a preset threshold; If so, a decrement pulse signal is generated based on the time of the current decrement data.

4. An electrocardiogram waveform classification device, characterized in that: include: An electrocardiogram signal acquisition module, used to acquire electrocardiogram signals according to a preset sampling frequency; A normalization processing module, used for performing normalization processing on the electrocardiogram signal to obtain normalized data; A pulse signal conversion module, used for converting the normalized data into a pulse signal; A waveform classification module is used to input the pulse signal into a preset S-LRNN model and output a waveform classification result; the S-LRNN model includes an SRNN model whose input layer is an LSTM module; the SRNN model includes 46 adaptive pulse neurons ALIF, including 4 input layer neurons, 36 hidden layer neurons and 6 output layer neurons; The pulse signal includes an increment pulse signal and a decrement pulse signal; the pulse signal conversion module includes: A division submodule, configured to divide the normalized data into increasing data and decreasing data; an incremental pulse signal generating submodule, configured to generate an incremental pulse signal using the incremental data; The decrement pulse signal generating submodule is used to generate a decrement pulse signal using the decrement data.

5. The device according to claim 4, characterized in that The incremental pulse signal generating submodule includes: a first signal strength acquiring unit, configured to acquire a first signal strength of each incremental data; A first difference calculation unit, configured to calculate a first difference between a first signal strength of current incremental data and a first signal strength of previous incremental data; a first judging unit, configured to judge whether the first difference is greater than a preset threshold; The incremental pulse signal generating unit is configured to generate an incremental pulse signal based on the time of the current incremental data.

6. The device according to claim 4, characterized in that The decrement pulse signal generating submodule includes: a second signal strength acquiring unit, configured to acquire a second signal strength of each of the decreasing data; A second difference calculation unit, configured to calculate a second difference between the second signal strength of the current decreasing data and the second signal strength of the previous decreasing data; a second determining unit, configured to determine whether the second difference is greater than a preset threshold; The decrement pulse signal generating submodule is configured to generate a decrement pulse signal based on the time of the current decrement data.

7. An electronic device, characterized in that: The device includes a processor and a memory: The memory is used to store program code and transmit the program code to the processor; The processor is configured to execute the electrocardiogram waveform classification method according to any one of claims 1 to 3 according to instructions in the program code.

8. A computer-readable storage medium, characterized in that The computer-readable storage medium is used to store program code, and the program code is used to execute the electrocardiogram waveform classification method according to any one of claims 1 to 3.

Images