An electromyography signal processing method, device, medium, and program product based on baseline modeling and interference ratio reparameterization.
By using baseline modeling and interference ratio reparameterization to process pelvic floor electromyography (EMG) signals, the problems of data dynamism and abdominal muscle interference in existing pelvic floor EMG signal detection technologies are solved, improving the generalization ability and stability of diagnosis and making it suitable for multi-device applications.
Patent Information
- Application Number
- CN202610453790.1
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2026-04-08
- Publication Date
- 2026-11-17
AI Technical Summary
Existing technologies fail to effectively reflect the relationship between data dynamism and changes in electromyography (EMG) signals in pelvic floor muscle detection, neglecting the influence of abdominal muscle interference. This results in insufficient model generalization ability, poor stability, and strong dependence.
An electromyography signal processing method based on baseline modeling and interference ratio reparameterization is adopted. A proportional function is constructed through stage-level segmentation and feature extraction to suppress abdominal muscle interference. A multi-model probabilistic weighted decision mechanism is constructed to introduce cross-device consistency correction.
It improves the generalization and stability of pelvic floor dysfunction diagnosis, making it suitable for multi-device, multi-center clinical applications. It also reduces the impact of abdominal muscle interference on diagnosis, enhancing the system's stability and reliability.
Smart Images

Figure CN123101166A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of intelligent healthcare, and more specifically, to a method, device, medium, and program product for baseline modeling and interference ratio reparameterization of surface electromyography signals based on protocol awareness. Background Technology
[0002] The pelvic floor muscles are the group of muscles that close the bottom of the pelvis. Healthy pelvic floor muscles provide core support for the pelvic and abdominal organs. Pelvic floor dysfunction is a common but easily overlooked functional disorder. Current clinical assessments mainly rely on physician experience, simple threshold analysis, or single electromyographic indicators, which have the following limitations: 1. Modeling dynamic data as static data: The existing modeling process only treats data as static data, without reflecting the dynamic nature of the data, the interrelationships between different data, and ignoring the changes and mutual influences of electromyographic signals during the detection process, resulting in insufficient model generalization.
[0003] 2. Failure to reflect the influence of abdominal muscle interference on electromyographic signals: In actual testing, abdominal muscle activity often significantly interferes with pelvic floor electromyographic signals. However, existing methods often exclude data from subjects involved in abdominal muscle activity during signal processing, resulting in insufficient generalization ability of the model.
[0004] 3. High dependence on case characteristics and insufficient stability: Existing single models are highly dependent on subject characteristics, collection equipment, and collection conditions, but the actual collection equipment and collection conditions vary greatly, which leads to insufficient stability. Summary of the Invention
[0005] This invention aims to at least solve one of the technical problems existing in the prior art. To this end, this invention provides an electromyography signal processing method based on baseline modeling and interference ratio reparameterization, as well as a method, device, medium, and program product for constructing an intelligent diagnostic model of pelvic floor dysfunction based on the data obtained by this processing method, so as to achieve a solution that can explicitly model and detect protocol differences, suppress abdominal muscle interference, and has cross-device stability without changing the original signal acquisition process.
[0006] The first aspect of this application discloses an electromyography signal processing method based on baseline modeling and interference ratio reparameterization, the method comprising: The system receives a dataset of raw electromyographic (EMG) signals, which includes raw pelvic floor muscle EMG signals and raw abdominal muscle EMG signals, as well as the detection phase type corresponding to each signal segment; the detection phase type includes the contraction phase. Based on the detection stage type, the original pelvic floor muscle electromyography (EMG) signal and the original abdominal muscle EMG signal are segmented at the stage level to obtain pelvic floor muscle signal segments and abdominal muscle signal segments corresponding to the detection stage. During at least one contraction phase, phase feature values are extracted from the pelvic floor muscle signal segment and the abdominal muscle signal segment, respectively, to obtain pelvic floor muscle phase feature values and abdominal muscle phase feature values; a proportional function is constructed based on the pelvic floor muscle phase feature values and abdominal muscle phase feature values to generate interference features for characterizing the structural synergistic or interference relationship between the abdominal muscles and pelvic floor muscles.
[0007] The second aspect of this application discloses a method for constructing an intelligent diagnostic model for pelvic floor dysfunction, the method comprising: Obtain the raw electromyographic signal dataset of the training set samples and the corresponding pelvic floor functional state labels of the samples; The original electromyography signal dataset is processed according to the processing method disclosed in the first aspect of this application to obtain a reparameterized feature set; The model is iteratively trained based on training set samples that include reparameterized feature sets and pelvic floor function status labels; when the trained model reaches the training stopping condition, the trained model is used as the final diagnostic model. The third aspect of this application discloses an intelligent diagnostic method for pelvic floor dysfunction based on electromyography signals, the method comprising: Obtain the raw electromyographic signal dataset of the collected test samples; the raw electromyographic signal dataset includes raw pelvic floor muscle electromyographic signals from the pelvic floor muscle channel and raw abdominal muscle electromyographic signals from the abdominal muscle channel. The original electromyography signal dataset is input into the diagnostic model disclosed in the second aspect of this application for processing to obtain the auxiliary prediction result of the pelvic floor functional status of the sample to be tested. The fourth aspect of this application discloses a computer device, the device comprising: a memory and a processor; the memory for storing a computer program; and the processor for executing the computer program to implement the steps of the above-described method.
[0008] The fifth aspect of this application discloses a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of the method described above.
[0009] The sixth aspect of this application discloses a computer program product, including a computer program that, when executed by a processor, implements the steps of the above-described method.
[0010] This application has the following beneficial effects: 1. This application explicitly labels and models the detection protocol during the surface electromyography signal acquisition process, reflecting the changes and mutual influences of electromyography signals during the detection process, highlighting the dynamic nature of data and the interrelationships between different data, and improving the diagnostic generalization ability; 2. This application constructs an abdominal muscle interference ratio function at the feature level to suppress abdominal muscle interference without destroying the original signal. This avoids the problem in the prior art where improper signal processing methods (such as directly excluding data involving abdominal muscles, or processing through signal filtering or removing abnormal signals) can easily destroy the original signal structure or cause information loss. 3. This application constructs multiple candidate diagnostic models based on the protocol subspace and uses a probabilistic weighted ensemble decision mechanism to output the final diagnostic results, thereby improving the stability and reliability of the diagnosis through probabilistic ensemble. 4. The proposed solution introduces a cross-device consistency correction mechanism to improve system stability and is suitable for multi-device, multi-center clinical application scenarios. Attached Figure Description
[0011] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying 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 effort.
[0012] Figure 1 This is a schematic diagram of the method flow provided in the first aspect of the present invention; Figure 2 This is a schematic diagram of the method flow provided in the second aspect of the present invention; Figure 3 This is a schematic diagram of the method flow provided in the third aspect of the present invention; Figure 4 This is a schematic diagram of a computer device provided in an embodiment of the present invention; Figure 5 This is a schematic diagram of the architecture of an exemplary computing device provided in an embodiment of the present invention; Figure 6 This is a schematic diagram of the storage medium provided in an embodiment of the present invention; Figure 7 This is a schematic diagram of the process for intelligent diagnosis of pelvic floor dysfunction based on patient data provided in an embodiment of the present invention. Detailed Implementation
[0013] To enable those skilled in the art to better understand the present invention, the technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings.
[0014] In some of the processes described in the specification, claims, and accompanying drawings of this invention, multiple operations appearing in a specific order are included. However, it should be clearly understood that these operations may not be executed in the order they appear herein, or may be executed in parallel. The operation numbers, such as 101, 102, etc., are merely used to distinguish different operations and do not represent any execution order. Furthermore, these processes may include more or fewer operations, and these operations may be executed sequentially or in parallel. It should be noted that the descriptions such as "first," "second," etc., in this document are used to distinguish different messages, devices, modules, etc., and do not represent a sequential order, nor do they limit "first" and "second" to different types.
[0015] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.
[0016] Figure 1 This is a schematic flowchart of an electromyography signal processing method based on baseline modeling and interference ratio reparameterization provided by an embodiment of the present invention. Specifically, the method includes the following steps: S101: Receive the acquired raw electromyography (EMG) signal dataset, which includes raw pelvic floor muscle EMG signals and raw abdominal muscle EMG signals, as well as the detection phase type corresponding to each signal segment; the detection phase type includes the contraction phase; wherein, the raw pelvic floor muscle EMG signals come from the pelvic floor muscle channel, the raw abdominal muscle EMG signals come from the abdominal muscle channel, and the detection phase type is the detection protocol annotation information; In some embodiments, the detection protocol annotation information also includes the detection duration and the detection stage sequence number, which are used to perform stage-level segmentation and semantic constraints on signal segments in the original electromyography signal dataset.
[0017] S102: Based on the detection stage type, the original pelvic floor muscle electromyography signal and the original abdominal muscle electromyography signal are segmented at the stage level to obtain pelvic floor muscle signal segments and abdominal muscle signal segments corresponding to the detection stage. S103: During at least one contraction phase, extract phase feature values from the pelvic floor muscle signal segment and the abdominal muscle signal segment respectively to obtain pelvic floor muscle phase feature values and abdominal muscle phase feature values; construct a proportional function based on the pelvic floor muscle phase feature values and abdominal muscle phase feature values to generate interference features for characterizing the structural synergistic or interference relationship between the abdominal muscles and pelvic floor muscles. The interference features are obtained through the following method: Within any contraction phase s, pelvic floor muscle feature sets are extracted from the pelvic floor muscle signal segments and abdominal muscle signal segments, respectively. and abdominal muscle feature set The feature set includes at least one or more of time-domain features, frequency-domain features, and nonlinear features. In electromyography (EMG) signal analysis, signal amplitude is the basis of the original data, while time-domain, frequency-domain, and nonlinear features are the results obtained after "interpreting" and "refining" these amplitude data from different perspectives. The ratio function, also known as the interference feature, is obtained by calculating one or more interference features for one or more feature pairs respectively. .
[0018] Based on preset feature mapping rules, from the pelvic floor muscle feature set and abdominal muscle feature set The pelvic floor muscle feature value Px and the abdominal muscle feature value Ax are extracted separately. The pelvic floor muscle feature value Px and the abdominal muscle feature value Ax constitute k feature pairs. Based on the k feature pairs, the interference feature {R}, consisting of k interference features, is calculated. k}, where k is a natural number greater than or equal to 1; Optionally, the preset feature mapping rules are constructed based on the semantic consistency of the protocol stage, including any one or more of the following: matching of the same feature type, matching of the same scale, and matching of the same statistical caliber; that is: if A is taken from a certain type of feature (e.g., fast muscle contraction stage), then P is taken from the corresponding type of feature of the pelvic floor muscle channel in the same detection stage (e.g., fast muscle contraction stage); matching of the same scale and matching of the same statistical caliber refer to different contraction stages of the muscle.
[0019] Optionally, the contraction phase includes a fast muscle contraction phase, a slow muscle contraction phase, and a sustained contraction phase; Optionally, the method further includes: reparameterizing the pelvic floor muscle features based on the interference features to obtain a protocol-related reparameterized interference feature set.
[0020] In some embodiments, the scaling function includes any one or more of the following: a direct ratio relationship, a normalized scaling relationship, or a scaling mapping function that includes a smoothing term.
[0021] In some embodiments, the detection phase type further includes a resting phase, which includes a first resting phase before the pelvic floor muscle contraction phase and a second resting phase after the contraction phase. Specifically, the first resting phase includes, but is not limited to: a pre-resting phase before the fast muscle contraction phase, a resting phase between the fast muscle contraction phase and the slow muscle contraction phase, and a resting phase between the slow muscle contraction phase and the sustained contraction phase. The second resting phase refers to the post-resting phase after the sustained contraction phase. The method further includes: The statistical characteristic values of the pelvic floor muscle electromyographic signal segments during the first resting phase were calculated as the first resting baseline characteristic values. The statistical characteristic values of the pelvic floor muscle electromyographic signal segments during the second resting phase were calculated as the second resting baseline characteristic values. Using the first resting baseline feature value and / or the second resting baseline feature value as reference baselines, the pelvic floor muscle stage feature values in different contraction phases are corrected or normalized at the feature level to obtain the corrected pelvic floor muscle stage feature values. The corrected pelvic floor muscle phase feature value, the first resting baseline feature value or the second resting baseline feature value, and the interference feature are combined to form a protocol-related reparameterized feature set to show the effect of abdominal muscle interference on pelvic floor muscle contraction. Optionally, the reference baseline is a combined baseline function constructed based on a first resting baseline feature value or a second resting baseline feature value.
[0022] In some embodiments, "reparameterization" typically refers to the process of converting raw, high-dimensional, time-varying electromyographic signals into a set of low-dimensional "feature parameters" with clear physical or physiological significance.
[0023] Figure 2 This is a schematic flowchart of a method for constructing an intelligent diagnostic model for pelvic floor dysfunction according to an embodiment of the present invention. Specifically, the method includes the following steps: S201: Obtain the original electromyographic signal dataset of the training set samples and the corresponding pelvic floor function status labels of the samples; S202: Process the original electromyography signal dataset according to the processing method disclosed in the first aspect of this application to obtain a reparameterized feature set; S203: The model is iteratively trained based on training set samples including reparameterized feature sets and pelvic floor function status labels; when the trained model reaches the training stopping condition, the trained model is used as the final diagnostic model; in some embodiments, the model used includes, but is not limited to, machine learning models, neural networks or deep learning models.
[0024] Figure 3 This is a schematic flowchart of an intelligent diagnostic method for pelvic floor dysfunction based on electromyography signals provided by an embodiment of the present invention. Specifically, the method includes the following steps: S301: Acquire the raw electromyographic signal dataset of the collected sample to be tested; the raw electromyographic signal dataset includes the raw pelvic floor muscle electromyographic signal from the pelvic floor muscle channel and the raw abdominal muscle electromyographic signal from the abdominal muscle channel. S302: Input the original electromyography signal dataset into the diagnostic model disclosed in the second aspect of this application for processing to obtain the auxiliary prediction result of the pelvic floor functional status of the sample to be tested.
[0025] More specifically, such as Figure 7 The diagram shown is a flowchart of intelligent diagnosis of pelvic floor dysfunction based on patient data provided in an embodiment of the present invention. The patient data is a multidimensional dataset including the original electromyography signal dataset and other basic clinical data. The other basic clinical data includes, but is not limited to, patient baseline data, clinical symptoms, and relevant data from specialist physical examinations. The dataset is processed according to the diagnostic model disclosed in the second aspect, and the auxiliary prediction result obtained from the patient data is directly output, indicating whether the test sample belongs to normal pelvic floor muscle function or pelvic floor muscle dysfunction.
[0026] In some embodiments, the terms “subject” or “test subject” or “sample to be tested” as used herein refer to any animal (e.g., a mammal), including but not limited to humans, non-human primates, rodents, etc., which will become the recipient of a particular treatment. Generally, the terms “subject” and “patient” are used interchangeably herein when referring to human subjects. Preferably, the sample to be tested is a patient clinically used for prognostic assessment.
[0027] In some embodiments, the auxiliary prediction results include, but are not limited to, paper or electronic reports. These results are obtained by intelligent machines based on the relevant data of the subjects and are intended only as a reference for medical personnel, not as the final diagnosis results of the subjects.
[0028] Figure 4 This is a schematic diagram of a computer device provided in an embodiment of the present invention, such as... Figure 4 As shown, the device 2000 may include: one or more processors 2010 and one or more memories 2020; wherein the memories store computer-readable code that, when run by the one or more processors, can perform the methods described above.
[0029] The processor in this embodiment can be an integrated circuit chip with signal processing capabilities. The processor can be a general-purpose processor, a digital signal processor (DSP), an application-specific integrated circuit (ASIC), an off-the-shelf programmable gate array (FPGA), or other programmable logic devices, discrete gate or transistor logic devices, or discrete hardware components. It can implement or execute the methods, operations, and logic block diagrams disclosed in this embodiment. The general-purpose processor can be a microprocessor or any conventional processor, and can be based on an x86 or ARM architecture.
[0030] In general, the various exemplary embodiments of this disclosure can be implemented in hardware or dedicated circuitry, software, firmware, logic, or any combination thereof. Some aspects can be implemented in hardware, while others can be implemented in firmware or software that can be executed by a controller, microprocessor, or other computing device. When aspects of embodiments of this disclosure are illustrated or described as block diagrams, flowcharts, or using some other graphical representation, it will be understood that the blocks, apparatuses, systems, techniques, or methods described herein can be implemented as non-limiting examples in hardware, software, firmware, dedicated circuitry or logic, general-purpose hardware or controllers or other computing devices, or some combination thereof.
[0031] For example, the method or apparatus according to embodiments of this disclosure can also be used by means of Figure 5 The architecture of the computing device 3000 shown is used for implementation. For example... Figure 5 As shown, the computing device 3000 may include a bus 3010, one or more CPUs 3020, a read-only memory (ROM) 3030, a random access memory (RAM) 3040, a communication port 3050 connected to a network, an input / output component 3060, a hard disk 3070, etc. The storage devices in the computing device 3000, such as the ROM 3030 or the hard disk 3070, may store various data or files used for processing and / or communication of the methods provided in this disclosure, as well as program instructions executed by the CPU. The computing device 3000 may also include a user interface 3080. Of course, Figure 5 The architecture shown is merely exemplary and can be omitted as needed when implementing different devices. Figure 5 One or more components in the computing device shown.
[0032] This invention also includes a computer-readable storage medium, such as... Figure 6The diagram illustrates a storage medium 4000 provided in an embodiment of the present invention. The computer storage medium 4020 stores computer-readable instructions 4010. When the computer-readable instructions 4010 are executed by a processor, the method described above according to embodiments of the present disclosure can be performed. The computer-readable storage medium in the embodiments of the present disclosure may be volatile memory or non-volatile memory, or may include both volatile and non-volatile memory. Non-volatile memory may be read-only memory (ROM), programmable read-only memory (PROM), erasable programmable read-only memory (EPROM), electrically erasable programmable read-only memory (EEPROM), or flash memory. Volatile memory may be random access memory (RAM), which is used as an external cache. By way of example, but not limitation, many forms of RAM are available, such as Static Random Access Memory (SRAM), Dynamic Random Access Memory (DRAM), Synchronous Dynamic Random Access Memory (SDRAM), Double Data Rate Synchronous Dynamic Random Access Memory (DDRSDRAM), Enhanced Synchronous Dynamic Random Access Memory (ESDRAM), Synchronous Link Dynamic Random Access Memory (SLDRAM), and Direct Memory Bus Random Access Memory (DRRAM). It should be noted that the memory used in the methods described herein is intended to include, but is not limited to, these and any other suitable types of memory.
[0033] This disclosure also provides a computer program product or system, including a computer program that, when executed by a processor, implements the steps of the above-described method. The computer program product or computer program is used to execute an electromyography signal processing system based on baseline modeling and interference ratio reparameterization. The system includes: The first electromyographic signal acquisition module is used or configured to receive the acquired raw electromyographic signal dataset, which includes raw pelvic floor muscle electromyographic signals and raw abdominal muscle electromyographic signals, as well as the detection phase type corresponding to each signal segment; the detection phase type includes the contraction phase. The signal segmentation module is used or configured to perform stage-level segmentation of the original pelvic floor muscle electromyography (EMG) signal and the original abdominal muscle EMG signal based on the detection stage type, so as to obtain pelvic floor muscle signal segments and abdominal muscle signal segments corresponding to the detection stage.
[0034] An interference feature generation module is used or configured to extract phase feature values from the pelvic floor muscle signal segment and the abdominal muscle signal segment respectively during at least one contraction phase to obtain pelvic floor muscle phase feature values and abdominal muscle phase feature values; and to construct a proportional function based on the pelvic floor muscle phase feature values and abdominal muscle phase feature values to generate interference features for characterizing the structural synergistic or interference relationship between the abdominal muscles and pelvic floor muscles.
[0035] In some embodiments, the system further includes: The resting phase baseline modeling module is used or configured to calculate the statistical characteristic values of the pelvic floor muscle electromyography signal segments in the first resting phase as the first resting baseline characteristic values; and / or, calculate the statistical characteristic values of the pelvic floor muscle electromyography signal segments in the second resting phase as the second resting baseline characteristic values. The feature correction module is used or configured to use the first resting baseline feature value and / or the second resting baseline feature value as a reference baseline to perform feature-level correction or normalization processing on the pelvic floor muscle stage feature values in different contraction phases to obtain corrected pelvic floor muscle stage feature values. The feature combination module is used or configured to combine the corrected pelvic floor muscle phase feature value, the first resting baseline feature value or the second resting baseline feature value, and the interference feature to form a protocol-related reparameterized feature set.
[0036] In some embodiments, this embodiment also discloses a system for constructing an intelligent diagnostic model for pelvic floor dysfunction, the system comprising: The second electromyographic signal acquisition module is used or configured to acquire the raw electromyographic signal dataset of the training set samples and the corresponding pelvic floor functional state labels of the samples; A feature set processing module is used or configured to process the original electromyographic signal dataset according to the processing method disclosed in the first aspect of this application to obtain a reparameterized feature set; The model training module is used or configured to iteratively train the model based on training set samples including reparameterized feature sets and pelvic floor functional state labels; when the trained model reaches the training stopping condition, the trained model is used as the final diagnostic model; in some embodiments, the model used includes, but is not limited to, machine learning models, neural networks or deep learning models.
[0037] In some embodiments, this embodiment also discloses an intelligent diagnostic system for pelvic floor dysfunction based on electromyography signals, the system comprising: The third electromyographic signal acquisition module is used or configured to acquire the raw electromyographic signal dataset of the collected sample to be tested; the raw electromyographic signal dataset includes raw pelvic floor muscle electromyographic signals from the pelvic floor muscle channel and raw abdominal muscle electromyographic signals from the abdominal muscle channel. The result prediction module is used or configured to input the raw electromyographic signal dataset into the diagnostic model disclosed in the second aspect of this application for processing, and obtain auxiliary prediction results of the pelvic floor functional status of the sample to be tested.
[0038] In this embodiment, in order to more clearly describe the acquisition and processing of surface electromyography signals, the auxiliary diagnosis of pelvic floor muscle dysfunction is used as an example for exemplary description. However, this description is only one application scenario and is not intended to limit the scope of protection of this data processing method. Those skilled in the art will know that the content of this embodiment can be extended to other functional status assessment and rehabilitation scenarios based on electromyography signals, in addition to being used for the auxiliary diagnosis of pelvic floor dysfunction.
[0039] For example, the methods and systems of the present invention can be used for functional status assessment of different muscle groups, motor control ability analysis, or functional recovery level determination, and can be further applied to efficacy assessment, training stage determination, and auxiliary decision support in the rehabilitation training process.
[0040] Furthermore, this invention can also serve as an intelligent analysis module in a rehabilitation management system, combined with rehabilitation training equipment, wearable monitoring devices, or telemedicine systems, to continuously assess changes in the functional status of subjects and provide a reference for adjusting rehabilitation plans. Specific examples are described below: 1. Acquisition and protocol annotation of surface electromyography signals Using at least one surface electromyography (EMG) acquisition device, multi-channel EMG signals are collected from the subject under a preset detection procedure. This procedure comprises two parts: 1. Acquiring surface EMG signals of the pelvic floor muscles using the standard Glazer method; 2. Acquiring surface EMG signals of the abdominal muscles at corresponding body surface locations of the rectus abdominis and / or external oblique muscles. The time-series aspects of this detection procedure include, but are not limited to: • Pre-resting phase (1 minute) • Fast-twitch muscle contraction phase (5 repetitions), rest for 1 minute. • Slow contraction phase, followed by a 1-minute rest. • Continuous contraction phase (1 minute) • Resting phase (1 minute) During the acquisition process, corresponding detection protocol labels are synchronously recorded for each segment of electromyography (EMG) signal, including the detection phase type, duration, and phase sequence number, forming a protocol-annotated EMG dataset. These protocol labels are used for phase-level segmentation and semantic constraints of signal segments during subsequent feature extraction, interference modeling, baseline correction, and protocol subspace modeling.
[0041] 2. Protocol-Aware Feature Extraction and Reparameterization Based on the detection protocol annotation information obtained in step 1, the surface electromyography signals of the pelvic floor muscle channel and the abdominal muscle channel are segmented according to the detection stage, and feature quantization processing is performed on the signals in each detection stage.
[0042] The feature quantification process includes: calculating statistical feature values for the surface electromyography signals of the corresponding channels within each detection stage. The statistical feature values are the average values of signal amplitude or energy-related indicators, which are used to characterize the overall electromyographic activity level within that stage.
[0043] 2.1 To quantitatively characterize the degree of interference between abdominal muscles and pelvic floor muscle signals, this invention constructs an abdominal muscle interference ratio function R at the feature level. The interference ratio feature R is a proportional function constructed based on the abdominal muscle feature value Ax and the pelvic floor muscle feature value Px. This proportional function is used to characterize the structural synergistic or interference relationship between the abdominal muscles and pelvic floor muscles. The proportional function includes, but is not limited to, direct ratio relationships, normalized ratio relationships, or proportional mapping functions containing smoothing terms. When it is a direct ratio relationship, R = Ax / (Px + ε), where ε is a smoothing constant to prevent the denominator from being zero. The abdominal muscle feature value (A) and pelvic floor muscle feature value (P) are derived from the feature set after the stage division of the protocol awareness. Within any protocol stage, the feature set of the corresponding stage can be extracted from the abdominal muscle channel and the pelvic floor muscle channel, respectively. The feature set may include at least one of time-domain features, frequency-domain features, or nonlinear features.
[0044] A correspondence is established between abdominal muscle feature values and pelvic floor muscle feature values through a preset feature mapping mechanism. The mapping mechanism can be constructed based on feature type consistency, statistical scale consistency, or protocol stage semantic consistency.
[0045] The scaling function can be constructed for one or more feature pairs to generate interference-sensitive features, which together with the pelvic floor muscle features constitute the feature representation after interference modeling.
[0046] Based on the above ratio function, the pelvic floor muscle features are reparameterized to form a protocol-related reparameterized feature set.
[0047] illustrate: This invention does not modify or filter the original electromyographic signal, but models the interference at the feature level, thus preserving the integrity of the original signal.
[0048] 2.2 Baseline Modeling and Characteristic Correction Processing of Pre-Restful Potential In the detection protocol described in step 1, during the pre-resting phase, the average statistical characteristic of the surface electromyography (EMG) signal of the pelvic floor muscle channel during the calculation phase is used as the pre-resting baseline characteristic value; during the post-resting phase, the average statistical characteristic of the surface EMG signal of the pelvic floor muscle channel during the calculation phase is used as the post-resting baseline characteristic value.
[0049] In the feature processing during the non-resting detection phase, the average value of the pelvic floor muscle baseline features obtained in the pre-resting phase is used as an individualized reference baseline. The average values of the pelvic floor muscle statistical features in different contraction phases are then corrected or normalized at the feature level. For example, the pre-resting baseline feature value is used as the reference baseline for the fast-twitch contraction phase; the average value of the pre-resting and post-resting baseline feature values is used as the reference baseline for the slow-twitch contraction and sustained contraction phases. Optionally, the baseline-corrected features can be represented as difference features, proportional features, or nonlinear mapping features constructed based on the contraction phase features and the baseline function. The baseline-corrected pelvic floor muscle features are further combined with the abdominal muscle interference ratio features described in step 2.1 to form a protocol-related reparameterized feature representation.
[0050] illustrate: The processing of resting potentials in this invention is limited to the feature level and does not involve filtering, cropping, or reconstructing the original surface electromyography signals.
[0051] 3. Protocol subspace modeling and candidate model construction 3.1 Modeling Input Construction Using Feature Engineering Data to Replace Original Data Based on the protocol annotation in step 1 and the reparameterized feature set obtained in step 2, this invention no longer uses the original surface electromyography signal as the model input in the modeling stage, but instead uses the reparameterized features as the modeling input data.
[0052] Specifically, the reparameterized features corresponding to each subject in each contraction phase (or protocol phase) are organized to form modeling samples; each modeling sample includes at least: Pelvic floor muscle characteristic values (which can be statistical averages within a period); Quantitative characteristics of abdominal muscle interference (e.g., interference ratio characteristics); Feature representation after resting baseline correlation correction; Each modeling sample is associated with a pelvic floor functional status label to form a feature dataset for training.
[0053] Technical effects: By replacing the original signal with reparameterized features as the modeling input, protocol semantics, abdominal muscle interference modeling and baseline correction information are explicitly introduced without changing the acquisition process, which improves the stability and interpretability of feature representation and reduces the impact of original signal noise and individual differences on model training.
[0054] 3.2 Dataset Balancing Mechanism (Before Modeling) To address the issue of uneven distribution of sample numbers for different pelvic floor functional states in the training dataset, a dataset balancing process is performed on the feature dataset before training the candidate model.
[0055] The dataset balancing process takes the feature dataset and its class labels as input, and includes, but is not limited to, the following methods: Oversampling of minority categories; Undersampling is performed on samples from most categories; Alternatively, samples can be reconstructed / reweighted according to a preset ratio.
[0056] The resulting balanced training dataset is used for subsequent candidate model training.
[0057] Technical effect: By balancing the class imbalance, the training bias caused by class imbalance is reduced, the bias of the model towards the majority class is decreased, and the ability and stability of identifying pelvic floor dysfunction in the minority class are improved.
[0058] 3.3 Candidate Model Training and Output (Modeling) After obtaining a balanced training dataset, construct one or more candidate diagnostic models and train them.
[0059] The candidate model training process includes: Input: reparameterized feature vectors in the balanced training dataset; Training: using pelvic floor functional state labels as supervision signals, training the candidate model to learn the mapping relationship between features and functional states; Output: the trained candidate model outputs the prediction result or prediction probability of pelvic floor functional state for the test sample.
[0060] Candidate models can be existing machine learning models, neural networks, or deep learning models, and the specific model type does not constitute a limitation on the present invention.
[0061] 4. Probabilistic weighted ensemble decision making The diagnostic performance of candidate models on the validation dataset is evaluated, and performance weights are assigned to different models according to preset evaluation metrics.
[0062] The predicted probabilities output by each model are weighted and fused according to their performance weights to obtain the final integrated diagnostic probability result, and a diagnostic conclusion of pelvic floor dysfunction is output based on the result.
[0063] 5. Cross-device stability maintenance mechanism By performing consistency correction on the protocol-sensing feature distribution under different EMG acquisition devices or acquisition conditions, the diagnostic model output remains stable. The resting baseline modeling and phased baseline correction are used to reduce the impact of individual differences and device amplitude scale differences on the feature distribution. The interference ratio feature is used to provide a relatively scale-invariant interference representation under different device gain conditions, thereby improving the applicability of the system in multi-device, multi-center scenarios.
[0064] Results Explanation: The method was validated on a real clinical dataset, and the results showed that: 1. After introducing protocol-aware feature reparameterization, the diagnostic model significantly outperforms the control method without this mechanism in terms of overall diagnostic accuracy and stability across multiple validation datasets. 2. The abdominal muscle interference ratio modeling mechanism effectively reduces the impact of abdominal muscle activity on pelvic floor function assessment, and shows more consistent diagnostic output at different testing stages; 3. Compared with a single model, the probabilistic weighted ensemble strategy achieves higher comprehensive diagnostic performance metrics on both the test set and multiple independent validation sets. Specifically, using the proposed method to build a model on a dataset of 4019 cases, the diagnostic performance evaluation metrics on the internal validation dataset and the external validation dataset were as follows: AUC (0.83 vs 0.83), recall (0.83 vs 0.76), specificity (0.70 vs 0.76), and F1-score (0.78 vs 0.67), which fully demonstrates the clinical effectiveness of the proposed method.
[0065] 4. The method of the present invention maintains high consistency and robustness under different electromyography (EMG) acquisition devices, verifying its cross-device application capability.
[0066] It should be noted that the flowcharts and block diagrams in the accompanying drawings illustrate the architecture, functionality, and operation of possible implementations of systems, methods, and computer program products according to various embodiments of this disclosure. In this regard, each block in a flowchart or block diagram may represent a module, segment, or portion of code containing one or more executable instructions for implementing a specified logical function. It should also be noted that in some alternative implementations, the functions indicated in the blocks may occur in a different order than those indicated in the drawings. For example, two consecutively indicated blocks may actually be executed substantially in parallel, and they may sometimes be executed in reverse order, depending on the functions involved. It should also be noted that each block in the block diagrams and / or flowcharts, and combinations of blocks in the block diagrams and / or flowcharts, can be implemented using a dedicated hardware-based system that performs the specified function or operation, or using a combination of dedicated hardware and computer instructions.
[0067] In general, the various exemplary embodiments of this disclosure can be implemented in hardware or dedicated circuitry, software, firmware, logic, or any combination thereof. Some aspects can be implemented in hardware, while others can be implemented in firmware or software that can be executed by a controller, microprocessor, or other computing device. When aspects of embodiments of this disclosure are illustrated or described as block diagrams, flowcharts, or using some other graphical representation, it will be understood that the blocks, apparatuses, systems, techniques, or methods described herein can be implemented as non-limiting examples in hardware, software, firmware, dedicated circuitry or logic, general-purpose hardware or controllers or other computing devices, or some combination thereof.
[0068] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working processes of the systems, devices, and units described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here.
[0069] In the several embodiments provided in this application, it should be understood that the disclosed systems, apparatuses, and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be an indirect coupling or communication connection between apparatuses or units through some interfaces, and may be electrical, mechanical, or other forms.
[0070] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.
[0071] Furthermore, the functional units in the various embodiments of the present invention can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit.
[0072] The exemplary embodiments of this disclosure described in detail above are merely illustrative and not restrictive. Those skilled in the art will understand that various modifications and combinations can be made to these embodiments or their features without departing from the principles and spirit of this disclosure, and such modifications should fall within the scope of this disclosure.
Claims
1. A method for electromyographic signal processing based on baseline modeling and interference ratio reparameterization, characterized in that, include: Receive the acquired raw electromyography (EMG) signal dataset, which includes raw pelvic floor muscle EMG signals and raw abdominal muscle EMG signals, as well as the detection phase type corresponding to each signal segment; The detection phase type includes a contraction phase; Based on the detection stage type, the original pelvic floor muscle electromyography (EMG) signal and the original abdominal muscle EMG signal are segmented at the stage level to obtain pelvic floor muscle signal segments and abdominal muscle signal segments corresponding to the detection stage. During at least one contraction phase, phase feature values are extracted from the pelvic floor muscle signal segment and the abdominal muscle signal segment, respectively, to obtain pelvic floor muscle phase feature values and abdominal muscle phase feature values. A proportional function is constructed based on the pelvic floor muscle stage feature values and the abdominal muscle stage feature values to generate interference features for characterizing the structural synergistic or interference relationship between the abdominal muscles and the pelvic floor muscles.
2. The electromyographic signal processing method based on baseline modeling and interference ratio reparameterization according to claim 1, characterized in that, The interference features are obtained through the following method: Within any contraction phase s, pelvic floor muscle feature sets are extracted from the pelvic floor muscle signal segments and abdominal muscle signal segments, respectively. and abdominal muscle feature set The feature set includes at least one or more of time-domain features, frequency-domain features, and nonlinear features; Based on preset feature mapping rules, from the pelvic floor muscle feature set and abdominal muscle feature set The pelvic floor muscle feature value Px and the abdominal muscle feature value Ax were extracted respectively. The pelvic floor muscle feature value Px and the abdominal muscle feature value Ax constitute k feature pairs. Based on the k feature pairs, the interference feature {R}, consisting of k interference features, is calculated. k }, where k is a natural number greater than or equal to 1; Optionally, the preset feature mapping rules include any one or more of the following: matching of the same feature type, matching of the same scale, and matching of the same statistical caliber; Optionally, the contraction phase includes a fast muscle contraction phase, a slow muscle contraction phase, and a sustained contraction phase; Optionally, the method further includes: reparameterizing the pelvic floor muscle features based on the interference features to obtain a protocol-related reparameterized interference feature set.
3. The electromyographic signal processing method based on baseline modeling and interference ratio reparameterization according to claim 2, characterized in that, The proportional function includes any one or more of the following: direct ratio relationship, normalized proportional relationship, or proportional mapping function containing a smoothing term.
4. The electromyographic signal processing method based on baseline modeling and interference ratio reparameterization according to claim 1, characterized in that, The detection phase type also includes a resting phase, which includes a first resting phase before the pelvic floor muscle contraction phase and a second resting phase after the contraction phase. The method further includes: Calculate the statistical characteristic value of the pelvic floor muscle electromyography signal segment during the first resting phase as the first resting baseline characteristic value; and / or, calculate the statistical characteristic value of the pelvic floor muscle electromyography signal segment during the second resting phase as the second resting baseline characteristic value; Using the first resting baseline feature value and / or the second resting baseline feature value as reference baselines, the pelvic floor muscle stage feature values in different contraction phases are corrected or normalized at the feature level to obtain the corrected pelvic floor muscle stage feature values. The corrected pelvic floor muscle phase feature value, the first resting baseline feature value or the second resting baseline feature value, and the interference feature are combined to form a protocol-related reparameterized feature set. Optionally, the reference baseline is a combined baseline function constructed based on a first resting baseline feature value or a second resting baseline feature value.
5. The electromyographic signal processing method based on baseline modeling and interference ratio reparameterization according to claim 4, characterized in that, The original abdominal muscle electromyographic signals were obtained by collecting data from the corresponding body surface locations of the rectus abdominis and / or external oblique muscles.
6. A method for constructing an intelligent diagnostic model for pelvic floor dysfunction, characterized in that, The method includes: Obtain the raw electromyographic signal dataset of the training set samples and the corresponding pelvic floor functional state labels of the samples; The processing method according to any one of claims 2-5 processes the original electromyography signal dataset to obtain a reparameterized feature set; The model is iteratively trained based on the training set samples, which include the reparameterized feature set and the pelvic floor functional status labels; when the trained model reaches the training stopping condition, the trained model is used as the final diagnostic model. Optionally, the method further includes: determining the number of different state labels in the training set samples based on the pelvic floor function state labels, determining whether there is an uneven distribution of the number of samples, and if so, balancing the training set samples.
7. A smart diagnostic method for pelvic floor dysfunction based on electromyographic signals, characterized in that, The method includes: Acquire the raw electromyographic signal dataset of the collected sample to be tested; the raw electromyographic signal dataset includes raw pelvic floor muscle electromyographic signals from the pelvic floor muscle channel and raw abdominal muscle electromyographic signals from the abdominal muscle channel; The original electromyography signal dataset is input into the diagnostic model as described in claim 6 for processing to obtain the auxiliary prediction result of the pelvic floor function status of the sample to be tested; Optionally, the number of diagnostic models is N, where N is a natural number greater than or equal to 1. When N is 1, the auxiliary prediction result of the pelvic floor function status of the sample to be tested is obtained directly based on the output result of the diagnostic model. When N is greater than or equal to 2, the auxiliary prediction results output by different diagnostic models are weighted and fused according to performance weights to obtain the auxiliary prediction result of the pelvic floor functional state of the sample to be tested.
8. A computer device, characterized in that, The device includes: a memory and a processor; the memory is used to store a computer program; the processor executes the computer program to implement the steps of the method according to any one of claims 1-7.
9. A computer-readable storage medium, characterized in that, It stores a computer program that, when executed by a processor, implements the steps of the method as described in any one of claims 1-7.
10. A computer program product, comprising a computer program, characterized in that, When executed by a processor, the computer program implements the steps of the method described in any one of claims 1-7.
