Medical data processing system for pelvic floor muscle rehabilitation training

Abstract

This invention relates to the field of healthcare informatics and discloses a medical data processing system for pelvic floor muscle rehabilitation training, comprising: a data acquisition unit, a feature reconstruction unit, a signal decoupling unit, and a feedback decision unit. The feature reconstruction unit calculates the energy attenuation of abdominal electromyographic signals as they pass through the pelvic cavity volume conductor based on a tissue frequency dissipation model, generating an interference reconstruction component. The signal decoupling unit extracts the energy envelope of each channel and identifies conductive compensation data by comparing the high-frequency integrated energy differences. It then uses the interference reconstruction component to perform differential suppression processing on the pelvic floor electromyographic signals, outputting a pelvic floor muscle-derived intrinsic signal vector. The feedback decision unit generates feedback control commands based on the intrinsic signal vector. This invention utilizes the tissue frequency selective dissipation law to establish an asymmetric judgment mechanism, targeting and removing abdominal muscle compensation interference, solving the problem of false negatives of intrinsic features caused by traditional linear filtering, and ensuring the intrinsic authenticity of rehabilitation medical data.

Description

Technical Field

[0001] This invention relates to a medical data processing system for pelvic floor muscle rehabilitation training, belonging to the field of healthcare informatics technology. Background Technology

[0002] In current clinical rehabilitation practice for pelvic floor dysfunction, pelvic floor electromyographic biofeedback training is a core intervention method. By collecting and processing electromyographic signals generated by pelvic floor muscle contractions, a feedback guidance mechanism is established. The current mainstream technology utilizes a multi-channel signal acquisition system and employs differential amplification and fixed-frequency filtering techniques to extract electromyographic features, providing data support for clinical assessment. The fascia and adipose tissue inside the human pelvic cavity constitute an anisotropic volume conductor with physically selective frequency dissipation characteristics. When bioelectrical signals are conducted outward through such soft tissues, high-frequency components are attenuated, while low-frequency components penetrate relatively easily. This determines that the physiological signals collected by the sensor differ from the source signal in frequency domain distribution.

[0003] Given the generally low pelvic floor muscle strength among patients, compensatory movements often occur during rehabilitation training, i.e., subconsciously engaging abdominal muscles in contraction. The resulting compensatory signals are superimposed on the pelvic floor muscle channels, creating physiological data contamination. Existing medical monitoring equipment typically simplifies this aliased data as linearly superimposed noise, failing to consider the non-stationary modulation of conduction impedance caused by tissue deformation due to increased intra-abdominal pressure. To eliminate interference, the industry has attempted to use digital signal processing logic such as adaptive filtering or reference channel subtraction. However, because these methods are based on the assumption that signal and noise are independent, they ignore the frequency distortion during volume conduction. When performing interference signal subtraction, they cannot distinguish the weight differences between compensatory artifacts and intrinsic pelvic floor electrical activity within a specific frequency band, easily leading to the mistakenly suppressing pelvic floor muscle activity while simultaneously undermining interference. The generated weak high-frequency features are filtered out, causing intrinsic distortion of the underlying rehabilitation medical data. In addition to hardware limitations, software control methods also have bottlenecks. For example, Chinese invention patent application CN120477799A discloses a pelvic floor muscle repair method and related equipment, which uses bispectral analysis and deep learning models to analyze muscle activity. In engineering practice, such solutions tend to focus on signal feature statistical classification and black box model fitting, lacking prior modeling of the physical dissipation characteristics of pelvic volume conductors. Since the asymmetric influence of biological tissue on the attenuation rate of different frequency band components is not considered, when patients produce strong abdominal muscle compensation, this method has difficulty distinguishing interference components of different sources at the same frequency in the overlapping signal stream in the frequency domain. This leads to blurred boundaries between intrinsic signals and compensation artifacts in the algorithm, and in non-stationary physiological environments, it is very easy to cause intrinsic distortion of the underlying data due to excessive hedging or misjudgment.

[0004] Therefore, the technical problem to be solved by this invention is how to achieve deep decoupling between pelvic floor electromyographic characteristics and abdominal muscle compensation interference based on the frequency dissipation law of human tissue, and establish a medical data processing closed loop with physiological prior constraints. Summary of the Invention

[0005] To address the problems in the background art, the technical solution of the present invention is as follows: A medical data processing system for pelvic floor muscle rehabilitation training, comprising:

[0006] The data acquisition unit is used to acquire first electromyographic signal data collected in the pelvic floor anatomical region and second electromyographic signal data collected in the abdominal anatomical region.

[0007] The feature reconstruction unit is used to map the second electromyographic signal data to a preset tissue frequency dissipation model, calculate the spectral energy attenuation of the second electromyographic signal data during the transmission of the second electromyographic signal data through the pelvic volume conductor to the pelvic floor anatomical region, and generate an interference reconstruction component characterizing the component of the second electromyographic signal data that is transmitted to the pelvic floor anatomical region. The tissue frequency dissipation model limits the attenuation rate of the high-frequency component of the electromyographic signal to be higher than that of the low-frequency component. The high-frequency component is an electrical signal component with a frequency of not less than 100 Hz.

[0008] The signal decoupling unit is used to extract the energy envelope of the first electromyography signal data and the interference reconstruction component. When the energy envelope of the first electromyography signal data rises and is accompanied by the loss of high-frequency integral energy, the interference reconstruction component is used to perform differential suppression operation on the first electromyography signal data to remove the low-frequency interference signal belonging to the compensatory contraction of the abdominal muscles in the first electromyography signal data and output the pelvic floor muscle-derived intrinsic signal vector.

[0009] The feedback decision unit is used to compare the intrinsic pelvic floor muscle signal vector with preset medical benchmark data, and generate feedback control commands to adjust the intensity of rehabilitation guidance based on the deviation value generated by the comparison.

[0010] Preferably, when calculating the spectral energy attenuation, the feature reconstruction unit performs gain compression on the high-frequency components in the second electromyographic signal data according to the preset tissue conduction impedance parameters, and compensates for the spatial conduction delay of the interference reconstruction component relative to the first electromyographic signal data, so that the interference reconstruction component and the interference component in the first electromyographic signal data achieve phase alignment in the time domain.

[0011] Preferably, the signal decoupling unit consists of a dynamic gain adjustment subunit and a signal subtraction subunit. The dynamic gain adjustment subunit is used to monitor the amplitude change trend of the interference reconstruction component and determine the interference suppression coefficient based on the amplitude change trend. The signal subtraction subunit is used to perform point-to-point subtraction processing on the first electromyographic signal data using the interference reconstruction component and to normalize and scale the operation residual using the interference suppression coefficient to generate muscle explosive force characteristics and muscle endurance characteristics in the first electromyographic signal data.

[0012] Preferably, the system further includes a fatigue assessment unit, which is used to input the intrinsic signal vector of pelvic floor muscle into a preset state evolution probability model to calculate the fatigue drift of pelvic floor muscle function during continuous training periods; the feedback decision unit is also used to reduce the audio-visual guidance frequency corresponding to the feedback control command when the fatigue drift reaches a preset safety threshold.

[0013] Preferably, the feedback decision unit reduces the frequency of audio-visual guidance while adjusting the visual feedback brightness of the rehabilitation terminal to change the subject's muscle contraction threshold and records the baseline recovery rate of the pelvic floor myogenic intrinsic signal vector in the unguided state.

[0014] Preferably, the preset medical benchmark data includes the range of integrated electromyography values ​​and the median frequency distribution interval set for different pelvic floor function levels; the feedback decision unit is used to assess the motor unit recruitment level of pelvic floor muscle fibers by calculating the degree of deviation of the pelvic floor myogenic intrinsic signal vector from the range of integrated electromyography values.

[0015] Preferably, the data acquisition unit consists of a first signal processing circuit and a second signal processing circuit, which are used to perform analog-to-digital conversion with a precision of more than 12 bits on the acquired pelvic floor raw analog signal and abdominal raw analog signal, respectively, to generate a digital signal sequence that conforms to the HL7 protocol standard.

[0016] Preferably, the system also includes a medical record management unit, which is used to encrypt and store the intrinsic pelvic floor muscle signal vector and the corresponding feedback control instructions, and generate a muscle dynamics evolution trend diagram that reflects the rehabilitation process.

[0017] Preferably, the system further includes a data output unit, which transmits feedback control commands to an external biofeedback stimulation terminal to drive the biofeedback stimulation terminal to output a physical stimulation energy level that changes in a positive correlation with the contraction intensity of the pelvic floor muscles.

[0018] Compared with the prior art, the beneficial effects of the present invention are:

[0019] 1. In the medical data processing of pelvic floor muscle rehabilitation training, based on the inherent frequency-selective dissipation characteristics of human pelvic tissue as a volume conductor, the system introduces a frequency domain asymmetric logic mapping mechanism based on tissue dissipation gradient when processing pelvic floor muscle channel signals and abdominal muscle reference channel signals. By comparing the differences in integral energy distribution of different acquisition channels in preset high and low frequency bands in real time, it can objectively identify distal conduction artifacts caused by compensatory abdominal contractions from the mixed physiological signal stream. Thus, while preserving the intrinsic high-frequency electrophysiological characteristics of local pelvic floor muscles, it can accurately remove low-frequency abdominal pressure interference signals, solving the problem of false killing of intrinsic signals and residual interference that is easily caused by traditional linear filtering logic when processing non-stationary physiological data, and ensuring the intrinsic authenticity of the underlying rehabilitation medical data.

[0020] 2. By establishing a nonlinear correlation model between physiological state vectors and clinical rehabilitation standard library, the system maps the decoupled pure electromyographic features to a multidimensional rehabilitation evaluation coordinate system, realizing the logical transformation from physical electrical signals to clinical rehabilitation sites. It uses the dynamic evolution law of integral electromyographic values ​​and median frequency to quantify key medical indicators such as muscle explosive power, endurance, and coordination. This feature extraction and classification logic based on physiological priors improves the consistency between medical intervention strategies and clinical expert diagnostic conclusions, providing professional and consistent data support for the rehabilitation process in family settings.

[0021] 3. The system integrates a state evolution model based on Markov chains to monitor the drift trend of muscle fatigue indicators during continuous training cycles. It constructs a closed-loop collaborative mechanism that evolves from surface physiological characteristics to overall training load regulation. When the fatigue indicator reaches the physiological safety threshold, the decision logic automatically optimizes the biofeedback guidance parameters to physically block the compensatory contraction path in the fatigue state. This avoids the risk of tissue micro-damage that may be induced by erroneous signal guidance, and achieves deep coupling between medical information flow and clinical risk prevention and control. Attached Figure Description

[0022] Figure 1 This is a flowchart of the pelvic floor muscle intrinsic signal decoupling and feedback control logic of the present invention;

[0023] Figure 2 This is a diagram showing the connection and architecture of the various functional modules of the medical data processing system of the present invention.

[0024] The objectives, features, and advantages of this invention will be further explained in conjunction with the embodiments and with reference to the accompanying drawings. Detailed Implementation

[0025] The technical solutions of the embodiments of this application will be clearly described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, not all embodiments. All other embodiments obtained by those skilled in the art based on the embodiments of this application are within the scope of protection of this application.

[0026] A medical data processing system for pelvic floor muscle rehabilitation training includes:

[0027] The data acquisition unit is used to acquire first electromyographic signal data collected in the pelvic floor anatomical region and second electromyographic signal data collected in the abdominal anatomical region.

[0028] The feature reconstruction unit is used to map the second electromyographic signal data to a preset tissue frequency dissipation model, calculate the spectral energy attenuation of the second electromyographic signal data during the transmission of the second electromyographic signal data through the pelvic volume conductor to the pelvic floor anatomical region, and generate an interference reconstruction component characterizing the component of the second electromyographic signal data that is transmitted to the pelvic floor anatomical region. The tissue frequency dissipation model limits the attenuation rate of the high-frequency component of the electromyographic signal to be higher than that of the low-frequency component. The high-frequency component is an electrical signal component with a frequency of not less than 100 Hz.

[0029] The signal decoupling unit is used to extract the energy envelope of the first electromyography signal data and the interference reconstruction component. When the energy envelope of the first electromyography signal data rises and is accompanied by the loss of high-frequency integral energy, the interference reconstruction component is used to perform differential suppression operation on the first electromyography signal data to remove the low-frequency interference signal belonging to the compensatory contraction of the abdominal muscles in the first electromyography signal data and output the pelvic floor muscle-derived intrinsic signal vector.

[0030] The feedback decision unit is used to compare the intrinsic pelvic floor muscle signal vector with preset medical benchmark data, and generate feedback control commands to adjust the intensity of rehabilitation guidance based on the deviation value generated by the comparison.

[0031] Preferably, when calculating the spectral energy attenuation, the feature reconstruction unit performs gain compression on the high-frequency components in the second electromyographic signal data according to the preset tissue conduction impedance parameters, and compensates for the spatial conduction delay of the interference reconstruction component relative to the first electromyographic signal data, so that the interference reconstruction component and the interference component in the first electromyographic signal data achieve phase alignment in the time domain.

[0032] Preferably, the signal decoupling unit consists of a dynamic gain adjustment subunit and a signal subtraction subunit. The dynamic gain adjustment subunit is used to monitor the amplitude change trend of the interference reconstruction component and determine the interference suppression coefficient based on the amplitude change trend. The signal subtraction subunit is used to perform point-to-point subtraction processing on the first electromyographic signal data using the interference reconstruction component and to normalize and scale the operation residual using the interference suppression coefficient to generate muscle explosive force characteristics and muscle endurance characteristics in the first electromyographic signal data.

[0033] Preferably, the system further includes a fatigue assessment unit, which is used to input the intrinsic signal vector of pelvic floor muscle into a preset state evolution probability model to calculate the fatigue drift of pelvic floor muscle function during continuous training periods; the feedback decision unit is also used to reduce the audio-visual guidance frequency corresponding to the feedback control command when the fatigue drift reaches a preset safety threshold.

[0034] Preferably, the feedback decision unit reduces the frequency of audio-visual guidance while adjusting the visual feedback brightness of the rehabilitation terminal to change the subject's muscle contraction threshold and records the baseline recovery rate of the pelvic floor myogenic intrinsic signal vector in the unguided state.

[0035] Preferably, the preset medical benchmark data includes the range of integrated electromyography values ​​and the median frequency distribution interval set for different pelvic floor function levels; the feedback decision unit is used to assess the motor unit recruitment level of pelvic floor muscle fibers by calculating the degree of deviation of the pelvic floor myogenic intrinsic signal vector from the range of integrated electromyography values.

[0036] Preferably, the data acquisition unit consists of a first signal processing circuit and a second signal processing circuit, which are used to perform analog-to-digital conversion with a precision of more than 12 bits on the acquired pelvic floor raw analog signal and abdominal raw analog signal, respectively, to generate a digital signal sequence that conforms to the HL7 protocol standard.

[0037] Preferably, the system also includes a medical record management unit, which is used to encrypt and store the intrinsic pelvic floor muscle signal vector and the corresponding feedback control instructions, and generate a muscle dynamics evolution trend diagram that reflects the rehabilitation process.

[0038] Preferably, the system further includes a data output unit, which transmits feedback control commands to an external biofeedback stimulation terminal to drive the biofeedback stimulation terminal to output a physical stimulation energy level that changes in a positive correlation with the contraction intensity of the pelvic floor muscles.

[0039] Example 1: In a home-based rehabilitation training scenario for postpartum pelvic floor dysfunction patients, when the system encounters a situation where the patient has weak pelvic floor muscles and is accompanied by compensatory abdominal muscle contractions, the sensor collects the first electromyographic signal data. The signal exhibits a non-stationary aliasing state. Due to the frequency-selective dissipation characteristics of the human pelvic tissue as a volume conductor, the original electrophysiological signal generated by abdominal compensatory movements undergoes physical attenuation of its high-frequency components (greater than 100Hz) as it travels through the pelvic fascia and adipose tissue to the pelvic floor acquisition point. Meanwhile, the low-frequency components are superimposed on the pelvic floor muscle monitoring channel with higher fidelity. This asymmetrical frequency distribution causes conventional proportional differential suppression logic to easily produce residual background noise when attempting to subtract interference because it cannot identify the dissipation differences in the signal transmission path, or it may mistakenly filter out the weak intrinsic high-frequency characteristics of the pelvic floor muscles themselves.

[0040] To overcome the technical constraints between deep signal decoupling and feature fidelity, the feature reconstruction unit synchronously acquires the second electromyographic signal data. Mapped to a preset tissue frequency dissipation model, the spectral energy evolution of the second electromyographic signal data within the current sampling window is calculated to generate interference reconstruction components that characterize the components transmitted to the pelvic floor anatomical region. Simultaneously, the signal decoupling unit extracts the energy envelope of each channel and compares the high-frequency integrated energy differences to identify the weight ratio of conductive compensation data in different frequency bands, dynamically adjusting the coordination coefficient. The value enables the system to perform differential suppression processing. ,in, This is the output vector of pelvic floor muscle-derived intrinsic signals. For the synergy coefficient, The time-varying interference component is calculated based on the organization frequency dissipation model. To account for signal propagation delay, this approach precisely isolates low-frequency artifacts attributable to abdominal muscle contraction while maintaining the energy integrity of high-frequency features in the intrinsic electrophysiological signal of the pelvic floor muscles. The pre-defined tissue frequency dissipation model is expressed as a finite impulse response approximation model based on the transfer function, and the frequency domain response function is set as follows: Where f represents the signal frequency. This is a preset tissue frequency attenuation function, used to represent the tissue attenuation coefficient that increases non-linearly with frequency. The system uses a spatially calibrated amount of the thickness of the abdominal wall to the pelvic floor tissues of the subject. The system control module calls a built-in finite impulse response filter to discretize and fit the frequency domain curve. Within an independent time step of each sampling period, the analog-to-digital converted second electromyographic signal data and the filter weight coefficients are convolved and mapped. Based on the calculated physical conduction delay and energy attenuation characteristics, the interference reconstruction component is output in real time. The intrinsic signal vector purified by the above logic is called in real time by the feedback decision unit and compared with the preset clinical rehabilitation medical benchmark data. When the intrinsic contraction intensity of the pelvic floor muscles measured by the system deviates from the preset training target trajectory, feedback control commands are automatically generated to adjust the acoustic guidance frequency or visual feedback gain. This data processing method based on the prior of biological tissue physical dissipation transforms the originally highly interfering compensatory actions into controlled system input parameters, ensuring that rehabilitation medical intervention commands are always based on real and objective physiological data, and guaranteeing the medical safety and data certainty of the closed loop of pelvic floor rehabilitation training.

[0041] Example 2: In a clinical validation environment for pelvic floor electromyography (EMG) signal analysis, the system faces the challenge of volume conduction frequency distortion caused by differences in body fat percentage among subjects. The experimental data originates from EMG sequences collected by a physical rehabilitation testing platform. This platform has a voltage measurement range of 0μV to 500μV with a voltage measurement accuracy of no less than 0.1μV. Its sampling frequency is set to 1000Hz. The setting of the sampling period involves a technical trade-off between signal spectrum bandwidth and real-time processing load of the system. When the main energy distribution of the pelvic floor EMG signal is detected to be between 20Hz and 450Hz, in order to satisfy the sampling theorem and avoid aliasing risk, the sampling period is selected as the reciprocal of the sampling frequency. To simulate electromagnetic interference in a real medical environment, Gaussian white noise with a signal-to-noise ratio of 20dB is actively superimposed on the experimental signal source. During the experiment, the data acquisition unit synchronously acquires the first EMG signal data of the subject under the preset contraction command. With second electromyographic signal data The original The channel exhibits an energy envelope elevation phenomenon at the moment of active abdominal muscle intervention. Its original signal root mean square value changes abruptly from 12.5μV before contraction to 86.3μV. By comparing with the control group without a tissue frequency dissipation model, it was found that the use of linear differential logic resulted in a 45.2% energy loss of high-frequency components greater than 100Hz in the output signal after deducting abdominal interference, causing nonlinear distortion of the true pelvic floor electromyographic feature vector.

[0042] In the verification process of this invention, the feature reconstruction unit extracts features based on the tissue frequency dissipation model. The high-frequency integrated energy in the channel is no less than 100Hz. The calculation results show that the high-frequency attenuation rate of the abdominal muscle signal during the transmission through the pelvic volume conductor to the pelvic floor anatomical region is 8.2dB / cm. The signal decoupling unit dynamically determines the coordination coefficient σ based on the ratio of the high-frequency integrated energy to the preset low-frequency energy. Within the sampling window, when the ratio is 0.15, the coordination coefficient σ is mapped to 0.85, and the system performs dynamic constraint calculations. ,in, The output vector represents the intrinsic pelvic floor muscle signals, where σ is the coordination coefficient. The interference reconstruction component is calculated based on the organization frequency dissipation model. For signal propagation delay, after processing Its signal-to-noise ratio improved from the original 5.2dB to 18.6dB, and the retention rate of intrinsic high-frequency energy of the pelvic floor muscles reached 94.5%. The energy envelope of the first electromyography signal data increased, accompanied by the loss of high-frequency integral energy. The judgment condition was controlled by the sliding window integral comparison logic. In order to overcome the cross-domain misalignment defect between the overall time domain envelope and the surface transient frequency domain distribution in the computational dimension, the logic forced that the extraction actions of these two types of features share the same digital buffer block. For each finite-length discrete analog sequence segment intercepted by the sliding window, the system's underlying data scheduler simultaneously derived two parallel computing power extraction branches: envelope amplitude analysis and fast Fourier high-frequency integral calculation. The system forcibly attached a unified timestamp sequence number label of the current acquisition section to the extracted independent feature values ​​at both ends, thereby ensuring that the two data streams with different physical attributes always remain within the absolutely parallel temporal phase boundary when entering the joint judgment mechanism to execute the logic combination. During the operation of the signal decoupling unit, the data buffer sliding window with a time span of 100 milliseconds continuously calculates the mean square of the current first electromyography signal data. The root eigenvalue is quantitatively compared with the root mean square (RMS) value of the resting baseline obtained by the calibration procedure. When the positive deviation of the current RMS eigenvalue within the buffered sliding window from the resting baseline exceeds 20%, and the ratio of the integrated energy of the frequency band above 100 Hz within the sliding window to the total energy of the entire frequency band decreases by more than 15% compared to the corresponding ratio of the previous adjacent interference-free contraction window, the physical basis for setting the RMS deviation threshold to 20% originates from the elevation of the lowest proprioceptive surface electromyographic action potential induced by the passive traction of the pelvic floor skeletal muscle groups under abdominal pressure. Deviations below this level are filtered out by the system as background muscle fiber tremor noise. The 15% reduction in high-frequency energy is set based on the physical frequency variation law of human volume conductors. It is the lower bound of the natural physical low-pass dissipation attenuation that the compensatory electrophysiological impulse of the rectus abdominis muscle must undergo when penetrating the soft tissue gap of the human body to reach the distal pelvic floor pickup pole. The combination of the dual thresholds eliminates the misjudgment of single motion artifacts. The system then starts the signal subtraction subunit to perform point-by-point subtraction operation on the current signal data stream using the interference reconstruction component.

[0043] Gradient verification of the key parameter boundaries shows that the synergy coefficient The value range of is deterministically related to the conduction impedance distribution characteristics of pelvic tissues. When the abdominal interference intensity is between 1 and 5 times the intrinsic signal of the pelvic floor, the system output The residual error remained below 3.2%; when the interference intensity exceeded 5 times the boundary, the residual error slowly rose to 4.8% due to the nonlinear impedance jump caused by tissue compression. This performance trend verified that the parameter range defined by the present invention covers common compensatory conditions in clinical practice. The above experimental data confirmed that the data processing method based on the frequency dissipation law of biological tissue can effectively achieve deep decoupling of multi-source heterogeneous signals, ensure the authenticity of the underlying rehabilitation medical data in complex electrophysiological environments, and provide an accurate data benchmark for the feedback decision unit to generate feedback control commands to adjust the intensity of rehabilitation guidance.

[0044] Example 3: This example combines Figures 1 to 2 Description of the medical data processing system for pelvic floor muscle rehabilitation training, such as... Figure 1 As shown, the data acquisition unit acquires first and second electromyographic (EMG) signal data, transmits the second EMG signal data to the feature reconstruction unit, and simultaneously transmits the first EMG signal data to the signal decoupling unit. The feature reconstruction unit generates an interference reconstruction component by mapping the tissue model and transmits this interference reconstruction component to the signal decoupling unit. The signal decoupling unit performs differential suppression and outputs an intrinsic signal vector, and transmits the resulting pelvic floor muscle-derived intrinsic signal vector to the feedback decision unit. The feedback decision unit compares the vector with a reference to generate a feedback control command and finally outputs the feedback control command. Figure 2 As shown, the data acquisition unit is connected downward to the feature reconstruction unit, the feature reconstruction unit is connected downward to the signal decoupling unit, the signal decoupling unit is connected downward to the fatigue assessment unit, the feedback decision unit and the medical record management unit, the fatigue assessment unit is connected downward to the feedback decision unit, and the feedback decision unit is connected downward to the data output unit and the medical record management unit.

[0045] Example 4: In the initial stage of rehabilitation intervention for patients with severe pelvic floor dysfunction, the system faces non-stationary jumps in signal conduction impedance due to the thickness of the patient's abdominal subcutaneous fat tissue. Before starting rehabilitation guidance, the system initiates a patient-specific impedance benchmark calibration procedure, and the data acquisition unit acquires the subject's first electromyographic signal data in a resting state. With second electromyographic signal data A baseline noise level for the background electromagnetic environment is established; the feature reconstruction unit extracts the background noise level during the subject's completion of a preset intensity of isolated abdominal muscle contraction. The original electrophysiological characteristics in the channel were used to calculate the attenuation operator reflecting the physical dissipation characteristics of the abdominal wall tissue based on the energy distribution curves of the original electrophysiological characteristics in the frequency domain from 20 Hz to 500 Hz. In the specific mapping calculation, the feature reconstruction unit extracts the integral area of ​​the characteristic frequency band with a frequency greater than 100Hz in the energy distribution curve as the high-frequency energy feature value, and extracts the integral area of ​​the frequency band from 20Hz to 100Hz as the low-frequency energy feature value. It calculates the ratio of the low-frequency energy feature value to the high-frequency energy feature value. The system internally stores a set of human body surface impedance calibration lookup tables, establishing a nonlinear monotonic correspondence between the energy ratio and the absolute value of the attenuation operator. Through table lookup and linear interpolation algorithms, the system deterministically transforms the currently calculated ratio into attenuation operator parameters with physical dimensions. The feature reconstruction unit then converts the attenuation operator... The system injects a transfer function to calculate the proportion of high-frequency energy loss of abdominal muscle interference signals after crossing a specific thickness of soft tissue. The system then compares this data with... The source energy and the aliased component energy conducted to the pelvic floor region determine the interference gain weight within the current sampling window, and the coordination coefficient is calculated. .

[0046] Coordination coefficient The value of is determined by the following formula: ,in, For the synergy coefficient, For the identified conducted interference power, The total power in the first electromyography signal data is the attenuation operator when the subject's subcutaneous tissue thickness increases, leading to an increase in conduction impedance. As the number of trigger signals increases, the decoupling unit lowers the coordination coefficient. In the high-frequency band, the gain is adjusted to avoid excessive interference with the intrinsic signals of the pelvic floor muscles. In this calibrated dynamic processing flow, the signal decoupling unit utilizes the real-time updated coordination coefficients. right The channel data underwent asymmetric filtering, and the system detected abdominal compensatory movements in the subjects during training. The system elevates the pelvic floor muscles, completes phase matching and amplitude compensation of the interference reconstruction components within 10ms, outputs the intrinsic pelvic floor muscle signal vector, and the feedback decision unit receives the intrinsic signal vector and generates adjustment feedback control commands to control the acoustic guidance frequency to be finely adjusted in real time with a step accuracy of 0.5Hz. This guides the subject to reduce abdominal muscle intervention and achieve the rehabilitation goal of isolated contraction of pelvic floor muscles.

[0047] Example 5: In the system initialization scenario of a multi-channel electromyography (EMG) synchronous acquisition unit, when the system faces an initial signal gain imbalance caused by the difference in contact impedance between the electrode pads and the skin, the data acquisition unit initiates a gain balance verification program. A standard sinusoidal test level with a frequency of 50Hz and an amplitude of 100μV is injected into the first and second acquisition channels via a built-in signal generator. The root mean square ratio of the output signals of each channel is calculated to determine the initial gain compensation operator. The initial gain compensation operator The gain balance calibration is superimposed onto the original signal data to eliminate deviations caused by differences in hardware interfaces. It should be noted that the aforementioned gain balance calibration procedure only runs independently during the internal self-test phase when the subject is not connected to the device. The standard sinusoidal test level output by the built-in signal generator is only injected in a closed loop into the operational amplifiers and isolation circuits at the front end of the two acquisition channels. The generated calibration current is completely blocked by the optocoupler isolation barrier at the hardware front end, and no active excitation current is output to the external human body surface. This purely internal circuit hardware gain calibration process physically ensures the passive monitoring attribute of this system for external human body volume conductors, providing a basis for the coordination coefficient in the subsequent signal decoupling unit. The mapping provides normalized data input.

[0048] When the system encounters parameter drift in the tissue frequency dissipation model due to differences in the anatomical structures of different subjects, the feature reconstruction unit completes the attenuation operator by retrieving a pre-set attenuation gradient lookup table and performing on-site impulse response tests. The correction uses the abdominal wall tissue thickness data obtained from ultrasound scanning as the input vector to determine the corresponding frequency dissipation benchmark value in the lookup table. Based on the signal characteristics generated by a single abdominal muscle contraction, the deviation between the actual attenuation rate and the benchmark value is calculated, and the high-frequency attenuation slope in the transfer function is updated. High-frequency attenuation slope The calculation formula is as follows: ,in, This is the updated high-frequency attenuation slope. The built-in reference decay rate of the model The impedance transformation factor, In practice, the tissue thickness deviation data is not generated by real-time ultrasound detection emitted by the hardware of this processing system. Instead, it is acquired by an external, independent standard medical ultrasound imaging device during the rehabilitation and record-keeping period. The absolute physical thickness value of the rectus abdominis muscle subcutaneous tissue measured by this external device is imported into this system via a data bus interface conforming to the HL7 protocol standard. The medical record management unit automatically parses it and performs a direct mathematical subtraction operation with the standard healthy population's average abdominal thickness constant built into the system. This generates the tissue thickness deviation input vector required for the above formula calculation. This method allows the tissue frequency dissipation model to be adapted to the tissue characteristics of a specific individual, ensuring the repeatability of the intrinsic signal vector extraction process among subjects with different physiological characteristics.

[0049] Example 6: In a multi-center clinical data validation scenario, when the system encounters raw data gain deviations caused by sensor placement misalignment and inconsistent electrode contact impedance, the data acquisition unit initiates a standardized sensor geometry calibration procedure, which limits the first electromyographic signal data. The vertical distance between the acquisition electrodes and the subject's pubic symphysis was between 3 cm and 5 cm, and the second electromyographic signal data... The acquisition electrode center is located 2 cm below the navel in the rectus abdominis muscle region. Before the rehabilitation training begins, the system automatically performs a 3-second background noise sampling. The real-time environmental noise baseline is determined by calculating the variance of the sampling sequence. ,exist When the voltage exceeds a preset threshold of 5μV, the signal decoupling unit triggers the electrode application quality detection logic until the measured electrode-skin contact impedance is reached. The data processing channel is only opened when the impedance drops below 5kΩ. This engineering calibration method, which targets physical layout parameters, establishes the geometric boundary for the operation of the frequency dissipation model and reduces the amplitude error of the interference reconstructed component caused by the randomness of sensor position.

[0050] When the feature reconstruction unit solidifies the model parameters for the tissue anisotropy characteristics of an individual subject, the system calls the field impedance feature fitting procedure based on standard pulse excitation. By applying a frequency sweep detection signal with a frequency range of 20Hz to 1000Hz to the abdominal region of the subject, the system uses the least squares method to perform polynomial fitting on the collected spectral energy attenuation data to determine the attenuation operator corresponding to the frequency-selective attenuation characteristics in the tissue frequency dissipation model. Calculate the synergy coefficient The sampling window length is set to 256ms and the step size is 64ms. The selection of the sampling window length is controlled by the weight allocation logic between the signal envelope tracking accuracy and the system processing delay. When the rate of change of the first derivative of the signal monitored in real time exceeds the preset jump threshold, the system automatically reduces the step size to 32ms to capture the transient characteristics of compensatory actions. The final output intrinsic signal vector is checked by the amplitude deviation verification program to keep the processing accuracy error of rehabilitation medical data within the range of 1.5%. This realizes the parameter solidification of the decoupling logic of multi-source heterogeneous signals and ensures the physical accuracy of the rehabilitation guidance intensity feedback control command output.

[0051] During the closed-loop feedback-guided rehabilitation phase, the feedback decision unit receives the pelvic floor muscle-derived intrinsic signal vector output by the signal decoupling unit. By performing envelope analysis on the intrinsic signal vector, it establishes the subject's electrophysiologically active components and generates feedback control commands based on preset control logic. The intensity factor of these feedback control commands... The calculation method is as follows ,in, Intensity factor This is the preset linear adjustment coefficient; This represents the root mean square value of the intrinsic signal output by the signal decoupling unit at the current sampling time. To obtain the reference value of the maximum voluntary contraction intensity of the subjects during the system calibration phase, in this closed-loop feedback engineering phase, to quantify the fatigue drift of pelvic floor muscle function, the fatigue assessment unit simultaneously inputs the extracted electrophysiological activity components into a state evolution probability model constructed based on a discrete Markov chain. This model clearly divides the state space of pelvic floor muscle function into three discrete levels: no fatigue, compensatory fatigue, and exhaustive fatigue. The system uses the median frequency of the signal within five consecutive sampling periods as the observation parameter. By calculating the percentage decrease in the current period's median frequency compared to the calibration baseline, the state transition probability matrix is ​​dynamically updated. When the decrease in the median frequency exceeds a 5% quantization step, the algorithm adds a 10% probability increment weight to the probability matrix to transition to the next fatigue state. The system uses the updated probability matrix to deduce the mathematical expectation of the current fatigue level in real time, and uses this specific value as the fatigue drift input into the feedback closed loop. The system compares this value in real time. The degree of deviation between the evolution trend and the preset target amplitude curve is monitored. When the deviation value exceeds the preset threshold of 10%, the feedback output intensity is automatically adjusted to ensure the mapping path from the intrinsic electrophysiological characteristics of the physical layer to rehabilitation decision instructions.

[0052] It will be apparent to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above, and that the present invention can be implemented in other specific forms without departing from the spirit or essential characteristics of the present invention.

[0053] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention.

Claims

1. A medical data processing system for pelvic floor muscle rehabilitation training, characterized in that, The system includes: The data acquisition unit is used to acquire first electromyographic signal data collected in the pelvic floor anatomical region and second electromyographic signal data collected in the abdominal anatomical region. The feature reconstruction unit is used to map the second electromyographic signal data to a preset tissue frequency dissipation model, calculate the spectral energy attenuation of the second electromyographic signal data during the transmission of the second electromyographic signal data through the pelvic volume conductor to the pelvic floor anatomical region, and generate an interference reconstruction component characterizing the component of the second electromyographic signal data that is transmitted to the pelvic floor anatomical region. The tissue frequency dissipation model limits the attenuation rate of the high-frequency component of the electromyographic signal to be higher than that of the low-frequency component. The high-frequency component is an electrical signal component with a frequency of not less than 100 Hz. The signal decoupling unit is used to extract the energy envelope of the first electromyography signal data and the interference reconstruction component. When the energy envelope of the first electromyography signal data rises and is accompanied by the loss of high-frequency integral energy, the interference reconstruction component is used to perform differential suppression operation on the first electromyography signal data to remove the low-frequency interference signal belonging to the compensatory contraction of the abdominal muscles in the first electromyography signal data and output the pelvic floor muscle-derived intrinsic signal vector. The feedback decision unit is used to compare the intrinsic pelvic floor muscle signal vector with preset medical benchmark data, and generate feedback control commands to adjust the intensity of rehabilitation guidance based on the deviation value generated by the comparison.

2. The medical data processing system for pelvic floor muscle rehabilitation training according to claim 1, characterized in that, When calculating the spectral energy attenuation, the feature reconstruction unit performs gain compression on the high-frequency components in the second electromyographic signal data according to the preset tissue conduction impedance parameters, and compensates for the spatial conduction delay of the interference reconstruction component relative to the first electromyographic signal data, so that the interference reconstruction component and the interference component in the first electromyographic signal data achieve phase alignment in the time domain.

3. The medical data processing system for pelvic floor muscle rehabilitation training according to claim 1, characterized in that, The signal decoupling unit consists of a dynamic gain adjustment subunit and a signal subtraction subunit. The dynamic gain adjustment subunit is used to monitor the amplitude change trend of the interference reconstruction component and determine the interference suppression coefficient based on the amplitude change trend. The signal subtraction subunit is used to perform point-to-point subtraction on the first electromyography signal data using the interference reconstruction component, and to normalize and scale the operation residual using the interference suppression coefficient to generate muscle explosive force characteristics and muscle endurance characteristics in the first electromyography signal data.

4. The medical data processing system for pelvic floor muscle rehabilitation training according to claim 1, characterized in that, The system also includes a fatigue assessment unit, which is used to input the pelvic floor muscle-derived intrinsic signal vector into a preset state evolution probability model to calculate the fatigue drift of pelvic floor muscle function during continuous training periods. The feedback decision unit is also used to reduce the audio-visual guidance frequency corresponding to the feedback control command when the fatigue drift reaches a preset safety threshold.

5. The medical data processing system for pelvic floor muscle rehabilitation training according to claim 4, characterized in that, While reducing the frequency of audio-visual guidance, the feedback decision unit changes the subject's muscle contraction threshold by adjusting the visual feedback brightness of the rehabilitation terminal, and records the baseline recovery rate of the pelvic floor myogenic intrinsic signal vector in the unguided state.

6. The medical data processing system for pelvic floor muscle rehabilitation training according to claim 1, characterized in that, The preset medical benchmark data includes the range of integrated electromyographic values ​​and the median frequency distribution interval set for different pelvic floor function levels; The feedback decision unit is used to assess the motor unit recruitment level of pelvic floor muscle fibers by calculating the degree of deviation of the pelvic floor myogenic intrinsic signal vector from the range of integral electromyography values.

7. The medical data processing system for pelvic floor muscle rehabilitation training according to claim 1, characterized in that, The data acquisition unit consists of a first signal processing circuit and a second signal processing circuit, which are used to perform analog-to-digital conversion with a precision of more than 12 bits on the acquired raw analog signals of the pelvic floor and the raw analog signals of the abdomen, respectively, to generate a digital signal sequence that conforms to the HL7 protocol standard.

8. The medical data processing system for pelvic floor muscle rehabilitation training according to claim 1, characterized in that, The system also includes a medical record management unit, which is used to encrypt and store the intrinsic pelvic floor muscle signal vectors and corresponding feedback control commands, and generate a muscle dynamics evolution trend diagram that reflects the rehabilitation process.

9. A medical data processing system for pelvic floor muscle rehabilitation training according to claim 1, characterized in that, The system also includes a data output unit, which transmits feedback control commands to an external biofeedback stimulation terminal to drive the biofeedback stimulation terminal to output a physical stimulation energy level that changes in a positive correlation with the contraction intensity of the pelvic floor muscles.

Images