Uterine contraction reaction detection method, system and device based on surface electromyography signal

Abstract

The present invention provides a method, system and device for detecting uterine contraction reaction based on surface electromyographic signal, and relates to the field of intelligent medical treatment. The method comprises: obtaining the original electromyographic signal data of the uterus on the surface of the subject to be tested; preprocessing the original electromyographic signal data to obtain the preprocessed electromyographic signal features; decomposing the electromyographic signal features using a time series decomposition algorithm, and if there is a feature residual term greater than a first preset threshold, the feature is input into a constructed clustering model to calculate the probability value, and if the probability value is greater than a second preset threshold, the point corresponding to the feature is considered to have a uterine contraction signal. Based on sEMG signals and machine learning methods, the present invention detects in real time whether a pregnant woman has a uterine contraction reaction, converts the electrical signal into a digital signal, and serves as a prompt and alert.

Description

Technical Field

[0001] The present invention relates to the field of intelligent medical treatment, and more specifically, to a method, system, equipment, medium, device and program product for detecting uterine contraction reaction based on surface electromyography signals. Background Art

[0002] Uterine contraction activity detection can identify the risk of premature birth in advance, providing pregnant women and doctors with more time to take intervention measures. The current commonly used method for uterine contraction detection in clinical practice is external uterine contraction pressure monitoring (TOCO), which uses a sensor positioned on the abdominal wall at the fundus of the mother's uterus and fixed with an elastic band to detect changes in the shape and tension of the abdominal wall, indirectly reflecting uterine contractions, and is used to identify when the contraction peak occurs and monitor the contraction frequency. However, TOCO does not monitor the actual intrauterine pressure and is easily affected by the pregnant woman's body position, the thickness of abdominal fat, the tightness of the elastic band, fetal movement, and abdominal wall muscle contraction (such as coughing and vomiting), so it cannot accurately describe the intensity of uterine contractions.

[0003] In recent years, uterine contractions have been monitored by uterine electromyography (uEMG). The principle of uEMG is similar to that of electrocardiogram and electromyography. The excitement of the uterine myometrium produces electrical activity. When uterine contractions increase, the degree of muscle excitement increases, and the number of electrical pulses sent per unit time increases. When bursts of string-like electromyographic waves occur, strong uterine contractions will occur, indicating that uterine myoelectric activity and uterine contractions occur almost simultaneously. Accurate assessment of uterine contractions is crucial for labor management. High-quality research data show that there is a stable baseline before and after the contraction pulse wave displayed by the uterine electromyography, which coincides with the uterine contraction. Motion artifacts can be ignored, and large motion artifacts are easy to distinguish.

[0004] In general, uEMG is a better monitoring method than TOCO. However, so far, most of the related studies on using EHG signals to identify uterine contractions are based on clinical experience, with TOCO pressure signals and records of uterine contractions perceived by pregnant women and doctors as references. In addition, in the process of using uEMG signals to identify uterine contractions, electrode configuration, signal filtering technology, etc. seriously affect the stability of uterine contraction detection, making the method of using uEMG to monitor uterine contractions impossible to apply in clinical practice. Summary of the invention

[0005] The present invention aims to solve at least one of the technical problems existing in the prior art. To this end, the present invention provides a uterine contraction reaction detection method based on surface electromyography signals, which detects whether a pregnant woman has a uterine contraction reaction in real time based on uEMG signals and machine learning methods, converts the electrical signals into digital signals, plays a role in prompting and alerting, and optimizes the stability of uEMG monitoring of uterine contractions.

[0006] The first aspect of the present application discloses a method for detecting uterine contraction reaction based on surface electromyography signals, the method comprising:

[0007] S101, obtaining original electromyographic signal data of the uterus on the body surface of the subject;

[0008] S102, preprocessing the original electromyographic signal data to obtain preprocessed electromyographic signal features;

[0009] S103, using a time series decomposition algorithm to decompose the electromyographic signal features, if there is a feature residual term greater than a first preset threshold, the feature is input into the constructed clustering model to calculate the probability value, if the probability value is greater than a second preset threshold, the point corresponding to the feature is considered to have a uterine contraction signal.

[0010] In some embodiments, the method for obtaining the first preset threshold value includes: obtaining the original electromyographic signal of the uterus on the body surface of the training set sample, filtering, normalizing, and reducing the dimension of the original electromyographic signal to obtain the independent components of the training set sample; decomposing the independent components of the training set sample by a time series decomposition algorithm and extracting residual terms, and calculating the residual mean value of the abnormal point to obtain the first preset threshold value;

[0011] In some embodiments, the raw electromyographic signal data includes data of a first channel S1, a second channel S2, and a third channel S3; wherein the data of the first channel is obtained by the difference between the electromyographic signal data of position one and position two at the fundus of the uterus, the data of the second channel is obtained by the difference between the electromyographic signal data of position two and position three, and the data of the third channel is obtained by the difference between the electromyographic signal data of position four and position three;

[0012] Optionally, the position 1 and the position 2 are located at any two positions of the fundus of the uterus on the abdominal surface of the subject, preferably symmetrically distributed relative to the navel;

[0013] Optionally, the position three is located on the lower right side of the navel on the abdominal surface of the subject;

[0014] Optionally, the position 4 is located at the lower left side of the navel on the abdominal surface of the subject;

[0015] Optionally, the raw electromyographic signal data is obtained by contacting electrode patches distributed on the abdominal surface of the subject, wherein the electrode patches include a first electrode patch located at position one, a second electrode patch located at position two, a third electrode patch located at position three, and a fourth electrode patch located at position four.

[0016] In some embodiments, the method further includes S1022: dimensionality reduction processing, performing dimensionality reduction processing on the electromyographic signal features to obtain independent components;

[0017] Optionally, the independent components are decomposed using a time series decomposition algorithm. If there is an independent component residual term greater than a first preset threshold, the independent component is input into the constructed clustering model to calculate the probability value. If the probability value is greater than a second preset threshold, the point corresponding to the independent component is considered to have a uterine contraction signal.

[0018] In some embodiments, the preprocessing method includes: filtering processing; filtering the raw electromyographic signal data to obtain smooth electromyographic signal data; extracting any one or more of the smooth electromyographic signal data MAV, MAVS, ZC, SSC, WL, RMS, and IAV to obtain the electromyographic signal features;

[0019] Optionally, the preprocessing method further includes S1021: normalization processing, which is located after the filtering processing and before the feature extraction.

[0020] In some embodiments, the method for constructing the clustering model includes: obtaining the original electromyographic signal of the uterus on the surface of the training set sample, filtering, normalizing, and reducing the dimension of the original electromyographic signal to obtain the independent components of the training set sample; using a machine learning algorithm to perform cluster analysis on the training set sample to obtain the clustering model;

[0021] Optionally, the clustering model construction method further includes: if there is a feature residual term or an independent component residual term that is less than or equal to a first preset threshold in S103, the original electromyographic signal data of the subject to be tested is used as the electromyographic signal data of the training set sample to input into the constructed clustering model to retrain the model, so as to obtain an updated clustering model;

[0022] Optionally, the clustering model construction method further includes: if the probability value is less than or equal to a second preset threshold value in S103, inputting the original electromyographic signal data of the subject to be tested as the electromyographic signal data of the training set sample into the constructed clustering model to re-train the model, so as to obtain an updated clustering model;

[0023] Optionally, the clustering model is a Gaussian mixture model, SVM, ANN, or RF;

[0024] Optionally, the timing decomposition algorithm is preferably an STL decomposition algorithm.

[0025] The second aspect of the present application provides a uterine contraction reaction detection system based on surface electromyography signals, the system comprising:

[0026] An acquisition module is used to acquire the original electromyographic signal data of the uterus on the body surface of the subject;

[0027] A preprocessing module is used to preprocess the original electromyographic signal data to obtain the preprocessed electromyographic signal characteristics;

[0028] The uterine contraction detection module is used to decompose the electromyographic signal features using a time series decomposition algorithm. If there is a feature residual term greater than a first preset threshold, the feature is input into the constructed clustering model to calculate the probability value. If the probability value is greater than a second preset threshold, the point corresponding to the feature is considered to have a uterine contraction signal.

[0029] A third aspect of the present application provides a computer device, the device comprising: a memory and a processor; the memory is used to store program instructions; the processor is used to call program instructions, and when the program instructions are executed, it is used to execute the steps of the method described in the first aspect of the present application.

[0030] A fourth aspect of the present application provides a computer-readable storage medium having a computer program stored thereon, and when the computer program is executed by a processor, the steps of the method described in the first aspect of the present application are implemented.

[0031] In a fifth aspect, the present application provides a uterine contraction reaction detection device based on surface electromyography signals, the device comprising: a signal acquisition module and a logic processing module, the signal acquisition module and the logic processing module are connected, the logic processing module receives the signal from the signal acquisition module and processes it; the signal acquisition module comprises a multiplexer, a signal amplification circuit, a signal filtering circuit, and an analog-to-digital conversion circuit connected in sequence.

[0032] The beneficial effects of this application compared to the prior art are as follows:

[0033] 1. The uterine contraction reaction detection method in this application processes the signal obtained from the body surface through high-pass filtering (DC isolation processing), high-power amplification, low-pass filtering (filtering out high-frequency interference) and other conditioning processes to ensure that the mixed very low frequency (close to DC) and high frequency interference signals in SEMG are processed and the stability of uterine contraction detection is optimized. This method realizes the automatic recognition of uterine contraction signals and provides a new way to monitor uterine activity.

[0034] 2. The uterine contraction reaction monitoring in this application needs to first perform uterine contraction reaction detection through the model constructed by the time series decomposition algorithm. If the conditions are met, the clustering model is used for detection. Only when two models detect uterine contraction reactions at the same time can the result of identifying uterine contraction reactions be obtained. The above process can ensure the accuracy and stability of uterine contraction reaction detection.

[0035] 3. Using the method disclosed in this application, the four electrodes can be directly placed on the corresponding positions of the pregnant woman's abdominal surface on the belly support belt, avoiding the problem of being unable to accurately describe the strength of uterine contraction due to the influence of the pregnant woman's body position, the thickness of abdominal fat, the tightness of the elastic band, fetal movement and abdominal wall muscle contraction. BRIEF DESCRIPTION OF THE DRAWINGS

[0036] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the drawings required for use in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative work.

[0037] Figure 1 is a schematic diagram of a method flow chart provided by the first aspect of an embodiment of the present invention;

[0038] Figure 2 is a schematic flow chart of a system provided by the second aspect of an embodiment of the present invention;

[0039] Figure 3 is a schematic diagram of a computer device provided by an embodiment of the present invention;

[0040] Figure 4 is a schematic diagram of the hardware design of the device provided in the fifth aspect of the embodiment of the present invention;

[0041] Figure 5 is a schematic diagram of a circuit design of a signal acquisition module provided by an embodiment of the present invention;

[0042] Figure 6 It is a schematic flow chart of a method for detecting uterine contraction reaction provided by an embodiment of the present invention;

[0043] Figure 7 It is a schematic diagram of the position of the electrode patch provided in an embodiment of the present invention designed on a belly support belt (the belly support belt is worn on the abdominal surface of the subject);

[0044] Figure 8 It is a schematic diagram of the surface electromyography data of the subject (pregnant woman) collected and processed in daily life provided by an embodiment of the present invention; wherein, a single collection is 30 minutes, and in order to ensure data quality, only the middle 20 minutes of data are retained for a single experimental data, that is, the experimental data of the first 5 minutes and the last 5 minutes are deleted. DETAILED DESCRIPTION

[0045] In order to enable those skilled in the art to better understand the solutions of the present invention, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the accompanying drawings in the embodiments of the present invention.

[0046] In some of the processes described in the specification and claims of the present invention and the above-mentioned figures, multiple operations that appear in a specific order are included, but it should be clearly understood that these operations may not be executed in the order in which they appear in this article or executed in parallel. The serial numbers of the operations, such as 101, 102, etc., are only used to distinguish different operations, and the serial numbers themselves do not represent any execution order. In addition, these processes may include more or fewer operations, and these operations may be executed in sequence or in parallel. It should be noted that the descriptions of "first", "second", etc. in this article are used to distinguish different messages, devices, modules, etc., do not represent the order of precedence, and do not limit the "first" and "second" to be different types.

[0047] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative work are within the scope of protection of the present invention.

[0048] Figure 1 The present invention provides a schematic flow chart of a method for detecting uterine contraction reaction based on surface electromyographic signals. Specifically, the method comprises the following steps:

[0049] S101: Acquire original electromyographic signal data of the uterus on the body surface of the subject;

[0050] In some embodiments, the term "subject" or "test subject" or "test sample" used herein refers to any animal (e.g., mammal), including but not limited to humans, non-human primates, rodents, etc., which will be the recipient of a specific treatment. Generally, the terms "subject" and "patient" are used interchangeably herein when referring to human subjects.

[0051] Preferably, the subject is a human.

[0052] In some embodiments, the subject is a patient undergoing clinical prognosis assessment.

[0053] In some embodiments, the raw electromyographic signal data includes data of a first channel S1, a second channel S2, and a third channel S3; wherein the data of the first channel is obtained by the difference between the electromyographic signal data of position one and position two at the fundus of the uterus, the data of the second channel is obtained by the difference between the electromyographic signal data of position two and position three, and the data of the third channel is obtained by the difference between the electromyographic signal data of position four and position three;

[0054] Optionally, the position 1 and the position 2 are located at any two positions of the fundus of the uterus on the abdominal surface of the subject, preferably symmetrically distributed relative to the navel;

[0055] Optionally, the position three is located on the lower right side of the navel on the abdominal surface of the subject;

[0056] Optionally, the position 4 is located at the lower left side of the navel on the abdominal surface of the subject;

[0057] Optionally, the raw electromyographic signal data is obtained by contacting electrode patches distributed on the abdominal surface of the subject, wherein the electrode patches include a first electrode patch located at position one, a second electrode patch located at position two, a third electrode patch located at position three, and a fourth electrode patch located at position four.

[0058] In the present invention, the fundus uterus and the lower segment of the uterine body refer to the corresponding parts of the uterus. Under normal circumstances, the uterus is located in the female pelvic cavity, behind the pubic symphysis, and the uterus is divided into two parts: the uterine body and the cervix. The fundus uterus is located in the upper part of the uterine body, that is, at the uppermost part of the uterus, the top of the uterus.

[0059] If a pregnant woman has uterine contractions, they all start from the uterine fundus, and the lower uterine segment refers to the part between the uterine body and the cervix, which is generally called the isthmus of the uterus during non-pregnancy. During pregnancy, the isthmus will gradually lengthen to form the lower segment of the uterus, which can reach 7-10cm by the end of pregnancy. It is called part of the female soft birth canal and is also the most commonly selected site for incision during cesarean section.

[0060] S102: preprocessing the original electromyographic signal data to obtain preprocessed electromyographic signal features;

[0061] In some embodiments, the preprocessing method includes: filtering processing; filtering the raw electromyographic signal data to obtain smooth electromyographic signal data; extracting any one or more of the smooth electromyographic signal data MAV, MAVS, ZC, SSC, WL, RMS, and IAV to obtain the electromyographic signal characteristics; specifically, Mean Absolute Value (MAV): the calculation method of the mean absolute value is:

[0062]

[0063] Where N is the number of surface electromyography data points collected, x iis the i-th data in the signal data sequence, which is often used to determine the intensity amplitude of the arm movement signal. This feature refers to the average value of the absolute value of the amplitude of the sEMG signal within a certain time window, which is usually used to evaluate the intensity and fatigue of muscle contraction. Similar to the root mean square value (RMS), MAV can also be used to estimate the average electrical signal strength during muscle contraction. The difference is that the square of the signal is not considered when MAV is calculated, but the absolute value of the amplitude is directly averaged. Therefore, compared with RMS, MAV is more stable and less affected by high-frequency noise. Generally, when the muscle is subjected to a large load or fatigue, MAV will decrease, and when the muscle is under a lighter load or more relaxed state, MAV will increase. Therefore, by monitoring the changes in the MAV value of the sEMG signal, the degree of muscle fatigue and the intensity of muscle contraction can be evaluated.

[0064] Mean Absolute Value Slope (MAVS) is calculated as:

[0065] MAVS=MAV i+1 -MAV i

[0066] The calculation method of Zero Crossing (ZC) is:

[0067]

[0068]

[0069] Where N is the number of surface electromyography data points collected, and x(i-1), x(i), and x(i+1) are the i-1, i, and i+1 data in the signal data sequence. This feature refers to the number of times the waveform of the sEMG signal crosses the horizontal baseline within a certain time window. It is usually used to evaluate the contraction and relaxation state of the muscle, and can also be used to estimate the contraction state of the muscle and the intensity of the muscle contraction. Usually, when the muscle contracts, the waveform of the sEMG signal becomes more complex, and the number of times it crosses the horizontal baseline will increase, so ZC will also increase. When the muscle is in a relaxed state, the waveform of the sEMG signal will be simple, and the number of times it crosses the horizontal baseline will decrease, so ZC will also decrease.

[0070] Slope Sign Changes(SSC): The slope sign changes are calculated as:

[0071]

[0072] Where N is the number of surface electromyography signal data points collected, and x(i-1), x(i), and x(i+1) are the i-1, i, and i+1 data in the signal data sequence. This feature refers to the number of times the sign of the amplitude change rate changes in the waveform of the sEMG signal, and is often used to indicate that the surface electromyography signal is about to fluctuate.

[0073] Waveform Length (WL) is calculated as follows:

[0074]

[0075] Where N is the number of surface electromyography signal data points collected, and x_i is the i-th data in the signal data sequence. This feature refers to the cumulative sum of the waveform lengths of the sEMG signal within a certain time window, which represents the complexity of the surface electromyography signal and is usually used to evaluate the contraction and relaxation state of the muscle. Usually, when the muscle contracts, the waveform of the sEMG signal becomes more complex, the length of the waveform increases, and therefore WL also increases. When the muscle is in a relaxed state, the waveform of the sEMG signal is simple and the length of the waveform decreases.

[0076] The root mean square (RMS) is calculated as:

[0077]

[0078] Among them, N is the number of surface electromyography signal data points collected, and xi is the i-th data in the signal data sequence, which is used to represent the average power of the collected signal. This feature refers to the square root of the amplitude mean of the sEMG signal within a certain time window, which is usually used to evaluate the intensity and fatigue of muscle contraction. Generally, when the muscle is subjected to a large load or fatigue, the RMS will decrease, and when the muscle is under a lighter load or more relaxed state, the RMS will increase. Therefore, by monitoring the changes in the RMS value of the sEMG signal, the degree of muscle fatigue and the intensity of muscle contraction can be evaluated.

[0079] The calculation method of IntegeratedAbsolute Value (IAV) is:

[0080]

[0081] Optionally, the preprocessing method further includes S1021: normalization processing, which is located after the filtering processing and before the feature extraction.

[0082] S103: Decomposing the electromyographic signal feature by using a time series decomposition algorithm, if there is a feature residual term greater than a first preset threshold, inputting the feature into the constructed clustering model to calculate a probability value, if the probability value is greater than a second preset threshold, the point corresponding to the feature is considered to have a uterine contraction signal;

[0083] In some embodiments, the method for obtaining the first preset threshold value includes: obtaining the original electromyographic signal of the uterus on the body surface of the training set sample, filtering, normalizing, and reducing the dimension of the original electromyographic signal to obtain the independent components of the training set sample; decomposing the independent components of the training set sample by a time series decomposition algorithm and extracting residual terms, and calculating the residual mean value of the abnormal point to obtain the first preset threshold value;

[0084] Optionally, the second preset threshold is obtained based on grid search, random search, Bayesian optimization and other methods, and the second preset threshold is the optimal threshold obtained.

[0085] In some embodiments, the method further includes S1022: dimensionality reduction processing, performing dimensionality reduction processing on the electromyographic signal features to obtain independent components; dimensionality reduction analysis is a general operation, and PCA can be used for independent principal component decoupling, or independent component analysis can be performed through the ICA algorithm. The main purpose of considering PCA here is to ensure that the units after dimensionality reduction are orthogonal to each other, which is convenient for subsequent STL decomposition, without considering the collinearity relationship between the units after dimensionality reduction. Although other methods can be used, PCA is a more suitable choice in terms of efficiency and theory. Specifically, PCA decomposes: , where W is the projection matrix; all samples are subtracted from the mean to be centered; the sample covariance matrix is ​​calculated; the covariance matrix is ​​decomposed by eigenvalues, and the eigenvectors corresponding to the largest d eigenvalues ​​are taken to form the projection matrix.

[0086] Optionally, the independent components are decomposed using a time series decomposition algorithm. If there is an independent component residual term greater than a first preset threshold, the independent component is input into the constructed clustering model to calculate the probability value. If the probability value is greater than a second preset threshold, the point corresponding to the independent component is considered to have a uterine contraction signal.

[0087] In some embodiments, the method for constructing the clustering model includes: obtaining the original electromyographic signal of the uterus on the surface of the training set sample, filtering, normalizing, and reducing the dimension of the original electromyographic signal to obtain the independent components of the training set sample; using a machine learning algorithm to perform cluster analysis on the training set sample to obtain the clustering model;

[0088] Optionally, the clustering model construction method further includes: if there is a feature residual term or an independent component residual term that is less than or equal to a first preset threshold in S103, the original electromyographic signal data of the subject to be tested is used as the electromyographic signal data of the training set sample to input into the constructed clustering model to retrain the model, so as to obtain an updated clustering model;

[0089] Optionally, the clustering model construction method further includes: if the probability value is less than or equal to a second preset threshold value in S103, inputting the original electromyographic signal data of the subject to be tested as the electromyographic signal data of the training set sample into the constructed clustering model to re-train the model, so as to obtain an updated clustering model;

[0090] Optionally, the clustering model is Gaussian mixture model (GMM), SVM, ANN, RF;

[0091] Optionally, the timing decomposition algorithm is preferably an STL decomposition algorithm.

[0092] In some embodiments, the model training process is as follows:

[0093] The uterine contraction response monitoring model consists of two parts: model training and online detection.

[0094] Model training mainly includes:

[0095] Clustering models such as Gaussian mixture model (GMM) are used to cluster independent components.

[0096] Assume that the data consists of k components with parameters θ k The Gaussian distribution of k A Gaussian mixture model composed of linear combinations,

[0097] The EM algorithm is used to estimate the parameters and obtain α k ,θ k The EM algorithm is mainly used to estimate the parameters of the GMM. The purpose is to obtain the optimal parameters of the GMM model on the training set through iterative optimization of the EM algorithm. It is part of the GMM model. The EM algorithm refers to the maximum expectation algorithm.

[0098] At the same time, the above-mentioned independent components are decomposed by STL, the residual items are extracted, and the residual mean of the abnormal points is calculated.

[0099] x(t)=T(t)+S(t)+R(t), additive model, where R(t) is the residual term, S(t) is the seasonal term, and T(t) is the trend term.

[0100] Online detection such as Figure 6 As shown, it mainly includes:

[0101] Perform filtering, normalization, and PCA dimension reduction preprocessing on the newly collected data;

[0102] Perform STL decomposition on the new data. If there is an independent component whose residual term is significantly greater than the mean, it will enter the abnormality judgment;

[0103] The GMM model obtained by the above training is used to calculate the probability value of the point. If it is greater than the set threshold, it is judged as an abnormal point and included in the contraction interval. The STL decomposition algorithm is mainly used to propose residual units in the signal sequence for preliminary contraction detection, while the GMM algorithm further detects the unclear contraction interval.

[0104] Update the model based on existing data. The purpose of data preprocessing in the online detection phase is to ensure 1) that the data distribution is consistent with the training set data, so that the model will not be unavailable on the training set due to data drift. 2) Online data is raw data, and data transformation operations such as dimensionality reduction and feature processing are required to meet the requirements of model input, otherwise the model cannot process the input. It should be noted that unless the data distribution changes significantly and the model effect is seriously degraded, model training is not required. Online detection is equivalent to the use of the model in the inference or deployment phase. When the performance drops significantly, only the new data needs to be fine-tuned without retraining.

[0105] Figure 2 A uterine contraction reaction detection system based on surface electromyographic signals provided by an embodiment of the present invention comprises:

[0106] The acquisition module 201 is used to acquire the original electromyographic signal data of the uterus on the body surface of the subject;

[0107] A preprocessing module 202 is used to preprocess the original electromyographic signal data to obtain preprocessed electromyographic signal features;

[0108] The uterine contraction detection module 203 is used to decompose the electromyographic signal features using a time series decomposition algorithm. If there is a feature residual term greater than a first preset threshold, the feature is input into the constructed clustering model to calculate the probability value. If the probability value is greater than a second preset threshold, the point corresponding to the feature is considered to have a uterine contraction signal.

[0109] Figure 3 A computer device is provided in an embodiment of the present invention, and the device comprises: a memory and a processor; the memory is used to store program instructions; the processor is used to call the program instructions, and when the program instructions are executed, it is used to execute the steps of the above method.

[0110] The embodiment of the present invention further provides a computer-readable storage medium on which a computer program is stored. When the computer program is executed by a processor, the steps of the above method are implemented.

[0111] The embodiment of the present invention further discloses a computer program product, including a computer program, which implements the steps of the above method when executed by a processor.

[0112] An embodiment of the present invention also discloses a uterine contraction reaction detection device based on surface electromyography signals, the device comprising: a signal acquisition module and a logic processing module, the signal acquisition module and the logic processing module are connected, the logic processing module receives the signal from the signal acquisition module and processes it; the signal acquisition module comprises a multiplexer, a signal amplification circuit, a signal filtering circuit, and an analog-to-digital conversion circuit connected in sequence.

[0113] In some embodiments, the device further comprises: a Bluetooth module, a display screen, an LED, a button, and a power management module, wherein the Bluetooth module, the display screen, the LED, the button, and the power management module are respectively connected to the logic processing module; the electrode position and uterine contraction reaction detection device is specifically as follows Figure 4 As shown in the figure, the electrode belt (E1-E2) is fixed above the belly support belt near the uterine fundus. The third electrode (E3) is fixed at the lower right of the navel, and the fourth electrode (E4) is fixed at the lower left of the navel. By calculating the potential difference at different positions, three channels of electromyographic signals can be collected, namely S1 = E2-E1 (first channel), S2 = E2-E3 (second channel), S3 = E4-E3 (third channel). The device is placed at the navel position, as shown in the figure. Figure 7 As shown, E1-4 are specific positions where the electrode patches are fixed on the belly support belt.

[0114] The overall hardware structure of the device consists of separate modules such as signal acquisition module, logic processing MCU (logic control circuit), Bluetooth module, display screen, LED, button and power management. Among them, the signal acquisition module includes a multiplexer (MUX), a signal amplification circuit, a signal filtering circuit, and an analog-to-digital conversion circuit (ADC). The design parameters of the signal acquisition module and the specific implementation process are as follows: Figure 5 As shown, the analog-to-digital conversion circuit is a circuit that converts analog signals into digital quantities. The bandpass filter removes signals with frequencies outside a certain range and is used to filter out abnormal value information. The Bluetooth module is used to transmit data to the mobile phone APP to achieve real-time data display and intelligent analysis. The display screen can draw information such as electromyography curves and uterine contraction curves in real time, and the power management module can intelligently manage the power supply of the device. Among them, the microcontroller unit (MCU) is also called a single-chip microcomputer (Single Chip Microcompter) or a single-chip microcomputer.

[0115] In the present invention, according to neurophysiological knowledge, muscle action potential will generate a potential difference of -90mV to 30mV. Since the human body is a poor conductor of electricity (internal resistance of the order of 1MΩ), only a peak value of about 1mV can be obtained from the patch electrode on the body surface. According to the literature, SEMG is often mixed with very low frequency (close to DC) and high frequency interference signals, and the effective myoelectric signal spectrum is distributed between 10-500Hz. Therefore, the signal detected from the patch electrode needs to undergo signal conditioning processes such as high-pass filtering (DC isolation processing), high-power amplification, and low-pass filtering (filtering out high-frequency interference).

[0116] Figure 8 Schematic diagram of surface electromyography data collected from a subject (pregnant woman) in daily life after processing according to an embodiment of the present invention; wherein, a single collection lasts 30 minutes, and in order to ensure data quality, only the middle 20 minutes of data are retained for a single experimental data, that is, the experimental data of the first 5 minutes and the last 5 minutes are deleted; the data is preprocessed, and the processed data is as follows Figure 8 As shown;

[0117] We compared the performance of the uterine contraction reaction detection method proposed in this application with other machine learning methods. The evaluation indicators include: sensitivity (Sen), specificity (Spe), accuracy (Acc) and area under the ROC curve (AUC), which are calculated as follows:

[0118]

[0119]

[0120]

[0121] AUC=∫Sen(T)(1-Spe)'(T)dT

[0122] Sen Spc Acc AUC Ours 93％ 95％ 96％ 0.98 SVM 68％ 93％ 85％ 0.91 ANN 83％ 92％ 88％ 0.90 RF 72％ 91％ 92％ 0.94

[0123] From the above experimental results, we can see that our method is superior to other methods.

[0124] The validation results of this validation example show that assigning intrinsic weights to indications can moderately improve the performance of the method relative to the default setting.

[0125] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working processes of the systems, devices and units described above can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0126] In the several embodiments provided in the present application, it should be understood that the disclosed systems, devices and methods can be implemented in other ways. For example, the device embodiments described above are only schematic. For example, the division of the units is only a logical function division. There may be other division methods in actual implementation, such as multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be an indirect coupling or communication connection through some interfaces, devices or units, which can be electrical, mechanical or other forms.

[0127] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0128] In addition, each functional unit in each embodiment of the present invention may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit. The above-mentioned integrated unit may be implemented in the form of hardware or in the form of software functional units.

[0129] A person of ordinary skill in the art can understand that all or part of the steps in the various methods of the above embodiments can be completed by instructing related hardware through a program, and the program can be stored in a computer-readable storage medium, and the storage medium may include: a read-only memory (ROM), a random access memory (RAM), a disk or an optical disk, etc.

[0130] A person skilled in the art can understand that all or part of the steps in the above-mentioned embodiment method can be completed by instructing related hardware through a program, and the program can be stored in a computer-readable storage medium. The above-mentioned storage medium can be a read-only memory, a disk or an optical disk, etc.

[0131] The above is a detailed introduction to a computer device provided by the present invention. For a person skilled in the art, according to the concept of the embodiments of the present invention, there may be changes in the specific implementation method and application scope. In summary, the content of this specification should not be understood as limiting the present invention.

Claims

1. A method for detecting uterine contraction reaction based on surface electromyographic signals, characterized in that: The method comprises: S101, obtaining the original electromyographic signal data of the uterus on the body surface of the subject; the original electromyographic signal data includes data of the first channel S1, the second channel S2 and the third channel S3; wherein the data of the first channel is obtained by the difference of the electromyographic signal data of the position one and the position two at the bottom of the uterus, the data of the second channel is obtained by the difference of the electromyographic signal data of the position two and the position three, and the data of the third channel is obtained by the difference of the electromyographic signal data of the position four and the position three; the position one and the position two are located at any two positions of the bottom of the uterus on the body surface of the abdomen of the subject; the position three is located at the lower right of the navel on the body surface of the abdomen of the subject; the position four is located at the lower left of the navel on the body surface of the abdomen of the subject; S102, preprocessing the original electromyographic signal data to obtain preprocessed electromyographic signal features; S103, using a time series decomposition algorithm to decompose the electromyographic signal feature, if there is a feature residual term greater than a first preset threshold, the feature is input into the constructed clustering model to calculate the probability value, if the probability value is greater than a second preset threshold, the point corresponding to the feature is considered to have a uterine contraction signal; the method for obtaining the first preset threshold includes: obtaining the original electromyographic signal of the uterus on the surface of the training set sample, filtering, normalizing, and reducing the dimension of the original electromyographic signal to obtain the independent components of the training set sample; decomposing the independent components of the training set sample through a time series decomposition algorithm and extracting the residual term, and calculating the residual mean value of the abnormal point to obtain the first preset threshold.

2. The method for detecting uterine contraction reaction based on surface electromyographic signal according to claim 1, characterized in that: The position one and the position two are symmetrically distributed relative to the navel.

3. The method for detecting uterine contraction reaction based on surface electromyographic signal according to claim 1, characterized in that: The raw electromyographic signal data is obtained by contacting electrode patches distributed on the abdominal surface of the subject, wherein the electrode patches include a first electrode patch located at position one, a second electrode patch located at position two, a third electrode patch located at position three, and a fourth electrode patch located at position four.

4. The method for detecting uterine contraction reaction based on surface electromyographic signal according to claim 1, characterized in that: The method further includes S1022: dimensionality reduction processing, performing dimensionality reduction processing on the electromyographic signal features to obtain independent components.

5. The method for detecting uterine contraction reaction based on surface electromyographic signal according to claim 4, characterized in that: The independent components are decomposed using a time series decomposition algorithm. If there is an independent component residual term greater than a first preset threshold, the independent component is input into the constructed clustering model to calculate the probability value. If the probability value is greater than a second preset threshold, the point corresponding to the independent component is considered to have a uterine contraction signal.

6. The method for detecting uterine contraction reaction based on surface electromyographic signal according to claim 1, characterized in that: The preprocessing method includes: filtering processing; filtering the original electromyographic signal data to obtain smooth electromyographic signal data; extracting any one or more of the smooth electromyographic signal data MAV, MAVS, ZC, SSC, WL, RMS, and IAV to obtain the electromyographic signal characteristics.

7. The method for detecting uterine contraction reaction based on surface electromyographic signal according to claim 1, characterized in that: The preprocessing method further includes S1021: normalization processing, which is located after the filtering processing and before the feature extraction.

8. The method for detecting uterine contraction reaction based on surface electromyographic signal according to claim 1, characterized in that: The method for constructing the clustering model includes: obtaining the original electromyographic signal of the uterus on the surface of the training set sample, filtering, normalizing and reducing the dimension of the original electromyographic signal to obtain the independent components of the training set sample; and using a machine learning algorithm to perform cluster analysis on the training set sample to obtain the clustering model.

9. The method for detecting uterine contraction reaction based on surface electromyographic signal according to claim 1, characterized in that: The method for constructing the clustering model also includes: if there is a feature residual term or an independent component residual term that is less than or equal to a first preset threshold in S103, the original electromyographic signal data of the subject is input into the constructed clustering model as the electromyographic signal data of the training set sample to re-train the model to obtain an updated clustering model.

10. The method for detecting uterine contraction reaction based on surface electromyographic signal according to claim 1, characterized in that: The clustering model construction method also includes: in S103, if the probability value is less than or equal to a second preset threshold, the original electromyographic signal data of the subject is used as the electromyographic signal data of the training set sample and input into the constructed clustering model to re-train the model to obtain an updated clustering model.

11. The method for detecting uterine contraction reaction based on surface electromyographic signal according to claim 1, characterized in that: The clustering models are Gaussian mixture model, SVM, ANN, and RF.

12. The method for detecting uterine contraction reaction based on surface electromyographic signal according to claim 1, characterized in that: The time series decomposition algorithm is an STL decomposition algorithm.

13. A uterine contraction reaction detection system based on surface electromyography signal, characterized in that: The system comprises: The acquisition module is used to acquire the original electromyographic signal data of the uterus on the body surface of the subject; the original electromyographic signal data includes data of the first channel S1, the second channel S2 and the third channel S3; wherein the data of the first channel is obtained by the difference of the electromyographic signal data of the position one and the position two at the bottom of the uterus, the data of the second channel is obtained by the difference of the electromyographic signal data of the position two and the position three, and the data of the third channel is obtained by the difference of the electromyographic signal data of the position four and the position three; the position one and the position two are located at any two positions of the uterine bottom on the abdominal body surface of the subject; the position three is located at the lower right of the navel on the abdominal body surface of the subject; the position four is located at the lower left of the navel on the abdominal body surface of the subject; A preprocessing module is used to preprocess the original electromyographic signal data to obtain the preprocessed electromyographic signal characteristics; The uterine contraction detection module is used to decompose the electromyographic signal features using a time series decomposition algorithm. If there is a feature residual term greater than a first preset threshold, the feature is input into a constructed clustering model to calculate a probability value. If the probability value is greater than a second preset threshold, the point corresponding to the feature is considered to have a uterine contraction signal. The method for obtaining the first preset threshold includes: obtaining the original electromyographic signal of the uterus on the surface of the training set sample, filtering, normalizing, and reducing the dimension of the original electromyographic signal to obtain the independent components of the training set sample; decomposing the independent components of the training set sample through a time series decomposition algorithm and extracting the residual term, and calculating the residual mean value of the abnormal point to obtain the first preset threshold.

14. A computer device, characterized in that: The device comprises: a memory and a processor; the memory is used to store program instructions; the processor is used to call the program instructions, and when the program instructions are executed, it is used to execute the steps of the method described in any one of claims 1 to 12.

15. A computer-readable storage medium, characterized in that: A computer program is stored thereon, and when the computer program is executed by a processor, the steps of the method described in any one of claims 1 to 12 are implemented.

16. A uterine contraction reaction detection device based on surface electromyographic signals, characterized in that: The device comprises: a signal acquisition module and a logic processing module, wherein the signal acquisition module and the logic processing module are connected, and the logic processing module receives and processes the signal of the signal acquisition module; the signal acquisition module comprises a multiplexer, a signal amplification circuit, a signal filtering circuit, and an analog-to-digital conversion circuit connected in sequence.

Images