Method and device for constructing feature map of uterine electromyographic signal in pregnancy delivery period, equipment and medium
By constructing a feature map of uterine electromyography signals during pregnancy and childbirth, the problem of systematically linking uterine electromyography signals with the state of uterine activity during pregnancy and childbirth was solved, enabling accurate prediction of uterine activity and monitoring support in multiple scenarios.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- GUANGZHOU AITINGBEI TECH CO LTD
- Filing Date
- 2026-05-15
- Publication Date
- 2026-06-16
AI Technical Summary
Existing technologies have failed to establish a unified and standardized systematic correlation and quantitative characterization between uterine electromyography signals and uterine activity during pregnancy and childbirth, resulting in scattered data, non-standard labeling, and difficulty in accurately identifying and predicting uterine activity.
A method for constructing uterine electromyography signal feature maps during pregnancy and childbirth was adopted, including data preprocessing, standardized annotation, feature extraction and screening, to build a multi-dimensional quantitative feature map library, and a deep learning model was used for prediction.
It enables accurate prediction of uterine activity status, provides monitoring and clinical decision support in multiple scenarios, and improves the consistency of data utilization and identification accuracy.
Smart Images
Figure CN122221053A_ABST
Abstract
Description
Technical Field
[0001] This invention belongs to the field of data processing, specifically relating to a method, apparatus, equipment, and medium for constructing a feature map of uterine electromyography signals during pregnancy and childbirth. Background Technology
[0002] Uterine electromyography (UEM) is the electrical signal generated during the contraction of uterine smooth muscle cells. As a key indicator for assessing uterine activity, it has irreplaceable application value in monitoring pregnancy and labor. It is known that various pathological states during pregnancy and labor (such as preterm labor, full-term labor, uterine atony, and uterine hyperactivity) are intrinsically correlated with uterine MMM signals. However, current technologies lack a unified and standardized technical system to achieve a systematic correlation and quantitative characterization between uterine MMM signals and different uterine activity states. Existing research data is mostly limited to comparisons between single pathological states and normal states; the data is scattered and lacks unified annotation standards, making it difficult to form a reusable and scalable uterine MMM feature reference system. This hinders accurate identification and prediction of uterine activity during pregnancy and labor based on uterine MMM signals, and fails to support monitoring and clinical decision support needs in multiple scenarios. Therefore, there is an urgent need in this field for a method and device capable of systematically constructing a feature atlas of uterine MMM signals during pregnancy and labor to solve the technical problems of data dispersion, non-standard annotation, missing feature systems, and insufficient accuracy in state identification in existing technologies. Summary of the Invention
[0003] This invention addresses the problem that existing technologies cannot uniformly correlate pathological states with uterine electromyography (EMG) signals by providing a method, apparatus, device, and medium for constructing a feature map of uterine EMG signals during pregnancy and childbirth.
[0004] To achieve the above objectives, the present invention adopts one or more of the following technical solutions: In a first aspect, the present invention provides a method for constructing a uterine electromyography signal feature map during pregnancy and childbirth, the method comprising the following steps: S1. Collect uterine electromyography (EMG) signal data of different parturients under different uterine activity states, and preprocess the uterine EMG signal data. The uterine activity states include at least normal pregnancy, preterm birth, full-term labor, uterine atony, and uterine hyperactivity. S2. Standardize and label the preprocessed uterine electromyography signal data. The standardization and labeling includes at least a label for basic information, a label for uterine activity status, and a label for signal quality. The label for signal quality is used to characterize the effectiveness of the uterine electromyography signal data and participates in the subsequent feature screening process. S3. Extract features from the standardized and labeled uterine electromyography signal data. The extracted features should include at least time-domain features, frequency-domain features, and nonlinear features. S4. Using the uterine activity status label as the target variable, perform feature screening on the time domain features, frequency domain features, and nonlinear features to obtain multi-dimensional quantitative features that distinguish different uterine activity states. The feature screening includes at least performing correlation analysis on the features to remove redundant features, and combining the signal quality label to screen the features in order to reduce the impact of low-quality signals on feature selection. S5. Construct a uterine electromyography signal feature atlas library based on the multi-dimensional quantitative features. The feature atlas library includes at least the mapping relationship between different uterine activity states and corresponding feature combinations, as well as the feature importance parameters or distribution range of each feature in the corresponding uterine activity state, which are used to characterize the feature patterns of uterine electromyography signals in different uterine activity states.
[0005] Furthermore, the method also includes: S6. Based on the multi-dimensional quantitative features, train a deep learning model, and use the deep learning model to analyze the input uterine electromyography signal data to obtain the corresponding prediction results of uterine activity status during pregnancy or childbirth.
[0006] Furthermore, in step S1: The preprocessing includes at least one or more of the following: power frequency interference removal, baseline drift correction, motion artifact removal, and signal segmentation.
[0007] Furthermore, step S2 specifically includes: S21. Based on the integrity of the uterine electromyography signal data, label the preprocessed uterine electromyography signal data with signal quality labels; S22. Based on the gestational age, age, BMI and past medical history of the mother, label the pre-processed uterine electromyography signal data with basic information labels; S23. Based on the clinically confirmed uterine activity status, label the pre-processed uterine electromyography signal data with uterine activity status tags; S24. The basic information tags, uterine activity status tags, and signal quality tags are associated with and stored with the corresponding uterine electromyography (EMG) signal data to form a standardized labeled uterine EMG signal dataset.
[0008] Furthermore, the labeling of the signal quality is graded based on the integrity of the uterine electromyography signal data, and the preset first setting value is greater than the second setting value; When the data integrity value is greater than the first set value, the signal quality label is marked as excellent. When the data integrity value is greater than the second set value and less than or equal to the first set value, the signal quality label is marked as medium. When data integrity is less than or equal to the second set value, the signal quality label is marked as poor. The labeling of the uterine activity status uses a two-level labeling method. First, based on the clinical status record, the uterine electromyography signal data is labeled as normal or abnormal as a primary label. Then, for the uterine electromyography signal data with an abnormal primary label, a secondary label corresponding to the status type is further labeled.
[0009] Furthermore, step S3 specifically includes: S31. For the time series of standardized and labeled uterine electromyography signal data, calculate its time domain characteristics, wherein the time domain characteristics include at least one or more of the following: peak value, mean value, contraction interval, contraction duration and kurtosis. S32. The standardized and labeled uterine electromyography signal data is converted into a frequency domain signal using Fourier transform or wavelet transform, and frequency domain features are extracted. The frequency domain features include at least one or more of the following: dominant frequency, effective frequency band energy ratio, and spectral entropy. S33. Extract nonlinear features from the standardized and labeled uterine electromyography signal data, wherein the nonlinear features include at least one or more of fractal dimension, approximate entropy and sample entropy.
[0010] Furthermore, the feature selection in step S4 specifically includes: S41. Perform correlation analysis on the time-domain features, frequency-domain features and nonlinear features, and remove redundant features with a correlation greater than a preset threshold. S42. Based on the signal quality label, the non-redundant features after removing redundant features are screened to remove non-redundant features corresponding to poor-quality signals and retain non-redundant features corresponding to high-quality and medium-quality signals, so as to reduce the impact of low-quality signals on feature selection. S43. Using the uterine activity status label as the dependent variable and the non-redundant features after screening as the input features, train a random forest model, obtain the feature importance score output by the random forest model, and sort them from high to low according to the feature importance score. S44. From the sorted feature importance scores, retain a preset number of features in descending order to obtain multi-dimensional quantitative features that distinguish different uterine activity states. S45. The effectiveness of the multidimensional quantitative features is verified by using one-way ANOVA or Kruskal-Wallis H test.
[0011] Secondly, the present invention provides a device for constructing a uterine electromyography signal feature map during pregnancy and childbirth, comprising: The acquisition module is used to acquire uterine electromyography (EMG) signal data of different postpartum women under different uterine activity states, and to preprocess the uterine EMG signal data. The standardized labeling module is used to standardize and label the preprocessed uterine electromyography signal data. The standardized labeling includes at least labeling basic information labels, uterine activity status labels, and signal quality labels. The extraction and filtering module is used to extract features from standardized and labeled uterine electromyography (EMG) signal data. The extracted features include at least time-domain features, frequency-domain features, and nonlinear features. Using the uterine activity state label as the target variable, the module filters the time-domain features, frequency-domain features, and nonlinear features to obtain multi-dimensional quantitative features that distinguish different uterine activity states. The feature filtering includes at least correlation analysis of the features to remove redundant features, and combining the signal quality label to filter the features to reduce the influence of low-quality signals on feature selection. Based on the multi-dimensional quantitative features, a uterine EMG signal feature atlas library is constructed. The prediction module is used to train a deep learning model based on the multi-dimensional quantitative features, and to analyze the input uterine electromyography signal data using the deep learning model to obtain the corresponding prediction results of uterine activity status during pregnancy or childbirth.
[0012] Thirdly, the present invention provides a device for constructing a uterine electromyography signal feature map during pregnancy and childbirth, comprising: At least one processor; and, A memory communicatively connected to the at least one processor; wherein, The memory has instructions that can be executed by the at least one processor, which, when executed by the at least one processor, enables the at least one processor to perform the above-described method.
[0013] Fourthly, the present invention provides a non-volatile computer storage medium storing computer-executable instructions configured to implement the above-described method.
[0014] The technical solution of this invention can achieve the following beneficial effects: (1) By using a standardized labeling system (including basic information, uterine activity status, and signal quality labels), the pain points of existing data being scattered and inconsistently labeled are solved, the consistency of data utilization is improved, and data standardization is achieved; (2) Simultaneously extract three types of features: time domain, frequency domain, and nonlinearity, and then accurately screen them. In the feature screening process, signal quality labels are combined for screening to reduce the impact of low-quality signals on feature selection. This forms a multi-dimensional quantitative feature that can characterize different uterine activity states. The constructed feature map library establishes a systematic association between uterine activity states and uterine electromyography signals. (3) Relying on the atlas library, it can efficiently and accurately predict the uterine activity status during pregnancy and childbirth, providing reliable technical support for clinical decision support in multiple scenarios, and is more practical. Attached Figure Description
[0015] Figure 1 This is a flowchart of a method for constructing a uterine electromyography signal feature map during pregnancy and childbirth, provided by the present invention. Figure 2 This is a schematic diagram of the structure of a device for constructing a uterine electromyography signal feature map during pregnancy and childbirth provided by the present invention; Figure 3 This is a schematic diagram of the system architecture involved in the embodiments of the present invention.
[0016] Figure 4 This is a schematic diagram of the structure of a computer device for constructing a feature map of uterine electromyography signals during pregnancy and childbirth, provided by the present invention. Detailed Implementation
[0017] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application pertains; the terminology used herein in the specification of the application is for the purpose of describing particular embodiments only and is not intended to be limiting of the application; the terms "comprising" and "having," and any variations thereof, in the specification, claims, and foregoing drawings of this application, are intended to cover non-exclusive inclusion. The terms "first," "second," etc., in the specification, claims, or foregoing drawings of this application are used to distinguish different objects, not to describe a particular order.
[0018] In this document, the term "embodiment" means that a particular feature, structure, or characteristic described in connection with an embodiment may be included in at least one embodiment of this application. The appearance of this phrase in various places throughout the specification does not necessarily refer to the same embodiment, nor is it a separate or alternative embodiment mutually exclusive with other embodiments. It will be explicitly and implicitly understood by those skilled in the art that the embodiments described herein can be combined with other embodiments.
[0019] To enable those skilled in the art to better understand the present application, the technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the accompanying drawings.
[0020] Figure 1 The flowchart illustrates a method for constructing a uterine electromyography signal feature map during pregnancy and childbirth, as provided by this invention.
[0021] The method flow steps of this embodiment are as follows: S1. Collect uterine electromyography (EMG) signal data of different mothers under different uterine activity states, and preprocess the uterine EMG signal data.
[0022] Specifically, the uterine activity states include at least normal pregnancy, premature birth, full-term labor, uterine atony, and uterine hyperactivity.
[0023] Specifically, the data source can be clinical samples collected from the obstetrics department of a tertiary hospital after ethical approval. The data should cover mothers of different gestational weeks (20-42 weeks), ages (20-45 years), and BMIs (18.5-30) to ensure data representativeness. The data should also be classified according to uterine activity, including at least 5 core scenarios (normal pregnancy, preterm birth, full-term labor, uterine atony, and uterine hyperactivity). In addition, the data needs to be anonymized (removing identifying information such as names and ID numbers) or supplemented from public databases (TPEHG, ICLeeHG).
[0024] Specifically, the preprocessing includes at least one or more of the following: power frequency interference removal, baseline drift correction, motion artifact removal, and signal segmentation.
[0025] Specifically, the uterine electromyography signal data for each uterine activity state are preprocessed with power frequency interference removal (50Hz notch filter), baseline drift correction (high-pass filter, cutoff frequency 0.5Hz), motion artifact removal (independent component analysis ICA or wavelet threshold denoising), and signal segmentation (divided according to the uterine contraction cycle or fixed time window, such as 10s / segment).
[0026] S2. Standardize and label the preprocessed uterine electromyography (EMG) signal data. The standardization and labeling includes at least basic information labels, uterine activity status labels, and signal quality labels. The signal quality labels are used to characterize the validity of the uterine EMG signal data and participate in the subsequent feature selection process.
[0027] Specifically, step S2 includes the following steps: S21. Based on the integrity of the uterine electromyography signal data, label the preprocessed uterine electromyography signal data with signal quality labels.
[0028] Specifically, the signal quality label is graded based on the integrity of the uterine electromyography signal data, and a first preset value is greater than a second preset value. When the data integrity is greater than the first preset value, the signal quality label is marked as excellent; when the data integrity is greater than the second preset value and less than or equal to the first preset value, the signal quality label is marked as medium; when the data integrity is less than or equal to the second preset value, the signal quality label is marked as poor.
[0029] As a preferred standard, the criteria for a signal quality label of "excellent" are: data integrity ≥ 95% (no missing fields / missing duration ≤ 5% of total duration); the criteria for a signal quality label of "medium" are: data integrity 80%-94% (missing parts do not affect core analysis); and the criteria for a signal quality label of "poor" are: data integrity < 80% (key fields are missing or the collection time is insufficient).
[0030] S22. Based on the mother's gestational age, age, BMI, and past medical history, label the pre-processed uterine electromyography signal data with basic information tags.
[0031] Specifically, gestational age should be accurate to "week + day" (e.g., "28 weeks 3 days"). If it cannot be accurate, an "estimated value" (e.g., "28 weeks ± 1 week") should be marked. For missing values, "gestational age missing" should be marked. Default values should not be entered arbitrarily. Age (years), BMI (body mass index), and medical history (e.g., hypertension, diabetes, preeclampsia, etc.) should be marked.
[0032] S23. Based on the clinically confirmed uterine activity status, label the pre-processed uterine electromyography signal data with uterine activity status tags.
[0033] Specifically, the labeling of the uterine activity status uses a two-level labeling method. First, based on the clinical status record, the uterine electromyography signal data is labeled as normal or abnormal as a primary label. Then, for the uterine electromyography signal data with an abnormal primary label, a secondary label corresponding to the uterine activity status type is further labeled.
[0034] Specifically, the primary label "Normal" indicates no diagnosis of any target disease, and clinical indicators (such as blood glucose and blood pressure) are within the normal range; the primary label "Abnormal" indicates a clinically confirmed diagnosis of at least one target disease (such as gestational diabetes mellitus or preeclampsia); secondary labels strictly follow ICD-10 coding or clinical guidelines (such as gestational diabetes mellitus code O24.4); for undetermined types, the label "Pathological type to be determined" must be added, and arbitrary classification is not allowed. After a single labeler completes the labeling, a second labeler will cross-check it.
[0035] S24. The basic information label, signal quality label, and uterine activity status label are associated and stored with the corresponding uterine electromyography signal data to form a standardized labeled uterine electromyography signal dataset.
[0036] Specifically, uterine electromyography signal data should be uniformly formatted as Excel / CSV or database (such as MySQL), with standardized field naming conventions (such as "sample number", "gestational week", "signal quality label", "primary status label", and "secondary status label").
[0037] S3. Extract features from the standardized and labeled uterine electromyography signal data. The extracted features include at least time-domain features, frequency-domain features, and nonlinear features.
[0038] Specifically, step S3 includes the following steps: S31. For the time series of standardized and labeled uterine electromyography signal data, calculate its time domain characteristics, wherein the time domain characteristics include at least one or more of the following: peak value, mean value, uterine contraction interval, uterine contraction duration, and kurtosis.
[0039] Specifically, the calculation method of time-domain features is shown in Table 1.
[0040] Table 1
[0041] S32. The standardized and labeled uterine electromyography signal data is converted into a frequency domain signal using Fourier transform or wavelet transform, and frequency domain features are extracted. The frequency domain features include at least one or more of the following: dominant frequency, effective frequency band energy ratio, and spectral entropy.
[0042] The specific calculation method for frequency domain features is shown in Table 2.
[0043] Table 2
[0044] S33. Extract nonlinear features from the standardized and labeled uterine electromyography signal data, wherein the nonlinear features include at least one or more of fractal dimension, approximate entropy and sample entropy.
[0045] The specific calculation method for nonlinear characteristics is shown in Table 3.
[0046] Table 3
[0047] S4. Using the uterine activity status label as the target variable, perform feature screening on the time-domain features, frequency-domain features, and nonlinear features to obtain multi-dimensional quantitative features that distinguish different uterine activity states. The feature screening includes at least performing correlation analysis on the features to remove redundant features, and combining the signal quality label to screen the features in order to reduce the impact of low-quality signals on feature selection.
[0048] Specifically, step S4 includes the following steps: S41. Perform correlation analysis on the time-domain features, frequency-domain features, and nonlinear features, and eliminate redundant features with a correlation greater than a preset threshold.
[0049] Specifically, correlation analysis includes linear correlation and nonlinear correlation. Linear correlation uses the Pearson coefficient (suitable for normal distribution characteristics, such as mean and peak value), while nonlinear correlation uses the Spearman rank correlation coefficient (suitable for non-normal distribution characteristics, such as fractal dimension and spectral entropy).
[0050] Then set a threshold: |r|≥0.8 (strong correlation). For each group of strongly correlated features, retain the one with "more explicit clinical significance" (e.g., when "peak value" and "mean value" are strongly correlated, retain "peak value" because the peak value more directly reflects the intensity of uterine contractions). If the correlation coefficient between "band energy percentage (1-2Hz)" and "dominant frequency" is 0.85, retain "dominant frequency" (dominant frequency is an indicator that is easier to understand clinically).
[0051] S42. Based on the signal quality label, the non-redundant features after removing redundant features are screened to remove non-redundant features corresponding to poor-quality signals and retain non-redundant features corresponding to high-quality and medium-quality signals, so as to reduce the impact of low-quality signals on feature selection. S43. Using the uterine activity status label as the dependent variable and the non-redundant features after screening as the input features, train a random forest model, obtain the feature importance score output by the random forest model, and sort them from high to low according to the feature importance score.
[0052] S44. From the sorted feature importance scores, retain a preset number of features in descending order to obtain multi-dimensional quantitative features that distinguish different uterine activity states.
[0053] S45. Validate the effectiveness of the multidimensional quantitative characteristics using one-way ANOVA or the Kruskal-Wallis H test to ensure the clinical reliability of the characteristics.
[0054] For example, for the "top 25 most important" features, statistical tests are performed between the "normal group" and each "abnormal group"; a significance threshold of p<0.05 is set (i.e., the probability that the difference between the two groups is caused by "random factors" is <5%); finally, features that "have p<0.05 between at least one abnormal group and the normal group" and have "clear clinical significance" are selected and retained, ultimately forming 15-20 core feature vectors.
[0055] S5. Construct a uterine electromyography signal feature atlas library based on the multi-dimensional quantitative features. The feature atlas library includes at least the mapping relationship between different uterine activity states and corresponding feature combinations, as well as the feature importance parameters or distribution range of each feature in the corresponding uterine activity state, which are used to characterize the feature patterns of uterine electromyography signals in different uterine activity states.
[0056] S6. Based on the multi-dimensional quantitative features, train a deep learning model, and use the deep learning model to analyze the input uterine electromyography signal data to obtain the corresponding prediction results of uterine activity status during pregnancy or childbirth.
[0057] This invention uses a uterine electromyography (EMG) signal feature atlas library to systematically and applicably correlate different uterine activity states with uterine EMG signals, so as to predict uterine activity states based on uterine EMG signals during different pregnancy and delivery processes.
[0058] Figure 2 This is a schematic diagram of a device for constructing a uterine electromyography signal feature map during pregnancy and childbirth, provided by the present invention.
[0059] The device includes: The acquisition module is used to acquire uterine electromyography (EMG) signal data of different postpartum women under different uterine activity states, and to preprocess the uterine EMG signal data. The standardized labeling module is used to standardize and label the preprocessed uterine electromyography signal data. The standardized labeling includes at least labeling basic information labels, uterine activity status labels, and signal quality labels. The extraction and filtering module is used to extract features from standardized and labeled uterine electromyography (EMG) signal data. The extracted features include at least time-domain features, frequency-domain features, and nonlinear features. Using the uterine activity state label as the target variable, the module filters the time-domain features, frequency-domain features, and nonlinear features to obtain multi-dimensional quantitative features that distinguish different uterine activity states. The feature filtering includes at least correlation analysis of the features to remove redundant features, and combining the signal quality label to filter the features to reduce the influence of low-quality signals on feature selection. Based on the multi-dimensional quantitative features, a uterine EMG signal feature atlas library is constructed. The prediction module is used to train a deep learning model based on the multi-dimensional quantitative features, and to analyze the input uterine electromyography signal data using the deep learning model to obtain the corresponding prediction results of uterine activity status during pregnancy or childbirth.
[0060] Figure 3 This is an exemplary system architecture diagram in which the present invention can be applied.
[0061] System architecture 100 may include terminal devices 101, 102, and 103, a network 104, and a server 105. Network 104 serves as the medium for providing communication links between terminal devices 101, 102, and 103 and server 105. Network 104 may include various connection types, such as wired or wireless communication links or fiber optic cables, etc.
[0062] Users can use terminal devices 101, 102, and 103 to interact with server 105 via network 104 to receive or send messages, etc. Various communication client applications can be installed on terminal devices 101, 102, and 103, such as web browser applications, shopping applications, search applications, instant messaging tools, email clients, social media platform software, etc.
[0063] Terminal devices 101, 102, and 103 can be various electronic devices with displays and support web browsing, including but not limited to smartphones, tablets, e-book readers, MP3 players (Moving Picture Experts Group Audio Layer III), MP4 players (Moving Picture Experts Group Audio Layer IV), laptops, and desktop computers, etc.
[0064] Server 105 can be a server that provides various services, such as a backend server that supports the pages displayed on terminal devices 101, 102, and 103.
[0065] It should be noted that the method for constructing a uterine electromyography signal feature map during pregnancy and childbirth provided in this embodiment can be executed by the server 105. Correspondingly, the device for constructing a uterine electromyography signal feature map during pregnancy and childbirth to implement the method can be deployed in terminal devices 101, 102, and 103.
[0066] It should be understood that Figure 3 The number of terminal devices, networks, and servers shown is merely illustrative. Depending on implementation needs, any number of terminal devices, networks, and servers can be included.
[0067] Figure 4 A schematic diagram of a computer device for constructing uterine electromyography signal feature maps during pregnancy and childbirth, provided by the present invention, includes: At least one processor; and, A memory communicatively connected to the at least one processor; wherein, The memory has instructions executable by the at least one processor, which, when executed by the at least one processor, enable the at least one processor to: Uterine electromyography (EMG) signal data of different parturients under different uterine activity states were collected, and the EMG signal data were preprocessed. The uterine activity states included at least normal pregnancy, preterm birth, full-term labor, uterine atony, and uterine hyperactivity. The preprocessed uterine electromyography (EMG) signal data is standardized and labeled. The standardized labeling includes at least a labeling of basic information, a labeling of uterine activity status, and a labeling of signal quality. The labeling of signal quality is used to characterize the effectiveness of the uterine EMG signal data and participate in the subsequent feature selection process. Feature extraction was performed on the standardized and labeled uterine electromyography signal data. The extracted features included at least time-domain features, frequency-domain features, and nonlinear features. Using the uterine activity status label as the target variable, feature screening is performed on the time domain features, frequency domain features, and nonlinear features to obtain multi-dimensional quantitative features that distinguish different uterine activity states. The feature screening includes at least performing correlation analysis on the features to remove redundant features, and combining the signal quality label to screen the features in order to reduce the impact of low-quality signals on feature selection. Based on the multi-dimensional quantitative features, a uterine electromyography signal feature atlas library is constructed. The feature atlas library includes at least the mapping relationship between different uterine activity states and corresponding feature combinations, as well as the feature importance parameters or distribution range of each feature in the corresponding uterine activity state, which is used to characterize the feature patterns of uterine electromyography signals in different uterine activity states. Based on the aforementioned multi-dimensional quantitative features, a deep learning model is trained, and the deep learning model is used to analyze the input uterine electromyography signal data to obtain the corresponding prediction results of uterine activity status during pregnancy or childbirth.
[0068] The computer device 4 includes a memory 41, a processor 42, and a network interface 43 that are interconnected via a system bus. It should be noted that only the computer device 4 with components 41-43 is shown in the figure; however, it should be understood that it is not required to implement all the shown components, and more or fewer components can be implemented alternatively. Those skilled in the art will understand that the computer device described here is a device capable of automatically performing numerical calculations and / or information processing according to pre-set or stored instructions, and its hardware includes, but is not limited to, microprocessors, application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), digital signal processors (DSPs), embedded devices, etc.
[0069] The computer device can be a desktop computer, laptop, handheld computer, or cloud server, etc. The computer device can interact with the user via a keyboard, mouse, remote control, touchpad, or voice control.
[0070] The memory 41 includes at least one type of readable storage medium, including flash memory, hard disk, multimedia card, card-type memory (e.g., SD or DX memory), random access memory (RAM), static random access memory (SRAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), programmable read-only memory (PROM), magnetic memory, disk, optical disk, etc. In some embodiments, the memory 41 may be an internal storage unit of the computer device 4, such as the hard disk or memory of the computer device 4. In other embodiments, the memory 41 may also be an external storage device of the computer device 4, such as a plug-in hard disk, smart media card (SMC), secure digital card (SD), flash card, etc., equipped on the computer device 4. Of course, the memory 41 may also include both the internal storage unit and its external storage device of the computer device 4. In this embodiment, the memory 41 is typically used to store the operating system and various application software installed on the computer device 4, such as the program code for a method of constructing a feature map of uterine electromyography signals during pregnancy and childbirth. In addition, the memory 41 can also be used to temporarily store various types of data that have been output or will be output.
[0071] In some embodiments, the processor 42 may be a central processing unit (CPU), controller, microcontroller, microprocessor, or other data processing chip. The processor 42 is typically used to control the overall operation of the computer device 4. In this embodiment, the processor 42 is used to run program code stored in the memory 41 or process data, for example, to run the program code for the method of constructing a uterine electromyography signal feature map during pregnancy and childbirth.
[0072] The network interface 43 may include a wireless network interface or a wired network interface, which is typically used to establish communication connections between the computer device 4 and other electronic devices.
[0073] This invention provides a non-volatile computer storage medium storing computer-executable instructions, wherein the computer-executable instructions are configured as follows: Uterine electromyography (EMG) signal data of different parturients under different uterine activity states were collected, and the EMG signal data were preprocessed. The uterine activity states included at least normal pregnancy, preterm birth, full-term labor, uterine atony, and uterine hyperactivity. The preprocessed uterine electromyography (EMG) signal data is standardized and labeled. The standardized labeling includes at least a labeling of basic information, a labeling of uterine activity status, and a labeling of signal quality. The labeling of signal quality is used to characterize the effectiveness of the uterine EMG signal data and participate in the subsequent feature selection process. Feature extraction was performed on the standardized and labeled uterine electromyography signal data. The extracted features included at least time-domain features, frequency-domain features, and nonlinear features. Using the uterine activity status label as the target variable, feature screening is performed on the time domain features, frequency domain features, and nonlinear features to obtain multi-dimensional quantitative features that distinguish different uterine activity states. The feature screening includes at least performing correlation analysis on the features to remove redundant features, and combining the signal quality label to screen the features in order to reduce the impact of low-quality signals on feature selection. Based on the multi-dimensional quantitative features, a uterine electromyography signal feature atlas library is constructed. The feature atlas library includes at least the mapping relationship between different uterine activity states and corresponding feature combinations, as well as the feature importance parameters or distribution range of each feature in the corresponding uterine activity state, which is used to characterize the feature patterns of uterine electromyography signals in different uterine activity states. Based on the aforementioned multi-dimensional quantitative features, a deep learning model is trained, and the deep learning model is used to analyze the input uterine electromyography signal data to obtain the corresponding prediction results of uterine activity status during pregnancy or childbirth.
[0074] Through the above description of the embodiments, those skilled in the art can clearly understand that the methods of the above embodiments can be implemented by means of software plus necessary general-purpose hardware platforms. Of course, they can also be implemented by hardware, but in many cases the former is a better implementation method. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product is stored in a storage medium (such as ROM / RAM, magnetic disk, optical disk) and includes several instructions to cause a terminal device (which may be a mobile phone, computer, server, air conditioner, or network device, etc.) to execute the methods described in the various embodiments of this application.
[0075] Although embodiments of the present invention have been described in conjunction with the accompanying drawings, the patent owner may make various modifications or alterations within the scope of the appended claims, as long as they do not exceed the protection scope described in the claims of the present invention, they shall be within the protection scope of the present invention.
Claims
1. A method for constructing a feature map of uterine electromyography signals during pregnancy and childbirth, characterized in that, The method includes the following steps: S1. Collect uterine electromyography (EMG) signal data of different parturients under different uterine activity states, and preprocess the uterine EMG signal data. The uterine activity states include at least normal pregnancy, preterm birth, full-term labor, uterine atony, and uterine hyperactivity. S2. Standardize and label the preprocessed uterine electromyography signal data. The standardization and labeling includes at least a label for basic information, a label for uterine activity status, and a label for signal quality. The label for signal quality is used to characterize the effectiveness of the uterine electromyography signal data and participates in the subsequent feature screening process. S3. Extract features from the standardized and labeled uterine electromyography signal data. The extracted features should include at least time-domain features, frequency-domain features, and nonlinear features. S4. Using the uterine activity status label as the target variable, perform feature screening on the time domain features, frequency domain features, and nonlinear features to obtain multi-dimensional quantitative features that distinguish different uterine activity states. The feature screening includes at least performing correlation analysis on the features to remove redundant features, and combining the signal quality label to screen the features in order to reduce the impact of low-quality signals on feature selection. S5. Construct a uterine electromyography signal feature atlas library based on the multi-dimensional quantitative features. The feature atlas library includes at least the mapping relationship between different uterine activity states and corresponding feature combinations, as well as the feature importance parameters or distribution range of each feature in the corresponding uterine activity state, which are used to characterize the feature patterns of uterine electromyography signals in different uterine activity states.
2. The method for constructing a uterine electromyography signal feature map according to claim 1, characterized in that, The method further includes: S6. Based on the multi-dimensional quantitative features, train a deep learning model, and use the deep learning model to analyze the input uterine electromyography signal data to obtain the corresponding prediction results of uterine activity status during pregnancy or childbirth.
3. The method for constructing a uterine electromyography signal feature map according to claim 1, characterized in that, In step S1: The preprocessing includes at least one or more of the following: power frequency interference removal, baseline drift correction, motion artifact removal, and signal segmentation.
4. The method for constructing a uterine electromyography signal feature map according to claim 1, characterized in that, Step S2 specifically includes: S21. Based on the integrity of the uterine electromyography signal data, label the preprocessed uterine electromyography signal data with signal quality labels; S22. Based on the gestational age, age, BMI and past medical history of the mother, label the pre-processed uterine electromyography signal data with basic information labels; S23. Based on the clinically confirmed uterine activity status, label the pre-processed uterine electromyography signal data with uterine activity status tags; S24. The basic information tags, uterine activity status tags, and signal quality tags are associated with and stored with the corresponding uterine electromyography (EMG) signal data to form a standardized labeled uterine EMG signal dataset.
5. The method for constructing a uterine electromyography signal feature map according to claim 4, characterized in that, The signal quality label is set in a graded manner based on the integrity of the uterine electromyography signal data, and the preset first setting value is greater than the second setting value. When the data integrity value is greater than the first set value, the signal quality label is marked as excellent. When the data integrity value is greater than the second set value and less than or equal to the first set value, the signal quality label is marked as medium. When data integrity is less than or equal to the second set value, the signal quality label is marked as poor. The labeling of the uterine activity status uses a two-level labeling method. First, based on the clinical status record, the uterine electromyography signal data is labeled as normal or abnormal as a primary label. Then, for the uterine electromyography signal data with an abnormal primary label, a secondary label corresponding to the status type is further labeled.
6. The method for constructing a uterine electromyography signal feature map according to claim 1, characterized in that, Step S3 specifically includes: S31. For the time series of standardized and labeled uterine electromyography signal data, calculate its time domain characteristics, wherein the time domain characteristics include at least one or more of the following: peak value, mean value, contraction interval, contraction duration and kurtosis. S32. The standardized and labeled uterine electromyography signal data is converted into a frequency domain signal using Fourier transform or wavelet transform, and frequency domain features are extracted. The frequency domain features include at least one or more of the following: dominant frequency, effective frequency band energy ratio, and spectral entropy. S33. Extract nonlinear features from the standardized and labeled uterine electromyography signal data, wherein the nonlinear features include at least one or more of fractal dimension, approximate entropy and sample entropy.
7. The method for constructing a uterine electromyography signal feature map according to claim 1, characterized in that, The feature selection in step S4 specifically includes: S41. Perform correlation analysis on the time-domain features, frequency-domain features and nonlinear features, and remove redundant features with a correlation greater than a preset threshold. S42. Based on the signal quality label, the non-redundant features after removing redundant features are screened to remove non-redundant features corresponding to poor-quality signals and retain non-redundant features corresponding to high-quality and medium-quality signals, so as to reduce the impact of low-quality signals on feature selection. S43. Using the uterine activity status label as the dependent variable and the non-redundant features after screening as the input features, train a random forest model, obtain the feature importance score output by the random forest model, and sort them from high to low according to the feature importance score. S44. From the sorted feature importance scores, retain a preset number of features in descending order to obtain multi-dimensional quantitative features that distinguish different uterine activity states. S45. The effectiveness of the multidimensional quantitative features is verified by using one-way ANOVA or Kruskal-Wallis H test.
8. A device for constructing a uterine electromyography signal feature map during pregnancy and childbirth, used to implement the method as described in any one of claims 1-7, characterized in that, The device includes: The acquisition module is used to acquire uterine electromyography (EMG) signal data of different postpartum women under different uterine activity states, and to preprocess the uterine EMG signal data. The standardized labeling module is used to standardize and label the preprocessed uterine electromyography signal data. The standardized labeling includes at least labeling basic information labels, uterine activity status labels, and signal quality labels. The extraction and filtering module is used to extract features from standardized and labeled uterine electromyography (EMG) signal data. The extracted features include at least time-domain features, frequency-domain features, and nonlinear features. Using the uterine activity state label as the target variable, the module filters the time-domain features, frequency-domain features, and nonlinear features to obtain multi-dimensional quantitative features that distinguish different uterine activity states. The feature filtering includes at least correlation analysis of the features to remove redundant features, and combining the signal quality label to filter the features to reduce the influence of low-quality signals on feature selection. Based on the multi-dimensional quantitative features, a uterine EMG signal feature atlas library is constructed. The prediction module is used to train a deep learning model based on the multi-dimensional quantitative features, and to analyze the input uterine electromyography signal data using the deep learning model to obtain the corresponding prediction results of uterine activity status during pregnancy or childbirth.
9. A device for constructing a uterine electromyography signal feature map during pregnancy and childbirth, characterized in that, include: At least one processor; as well as, A memory communicatively connected to the at least one processor; wherein, The memory has instructions executable by the at least one processor, which, when executed by the at least one processor, enables the at least one processor to perform the method as described in any one of claims 1-7.
10. A non-volatile computer storage medium, characterized in that, The device stores computer-executable instructions configured to implement the method as described in any one of claims 1-7.