Pelvic floor recruitment margin and defense timing closed-loop magnetic stimulation method and system

Abstract

The application discloses a pelvic floor recruitment margin and defense timing closed-loop magnetic stimulation method and system, belongs to the technical field of medical electronic equipment and rehabilitation treatment device, and is used for pelvic floor recruitment margin and defense timing closed-loop magnetic stimulation. The method comprises the following steps: injecting gradient detection pulses into the pelvic floor muscle, extracting a stimulation response fingerprint and synthesizing a recruitable margin; synchronously collecting abdominal pressure event multi-mode signals, detecting a defense timing gap, triggering compensation stimulation in a precursor stage by moving forward a total system delay, constructing a six-dimensional state vector containing the recruitable margin, the defense timing gap, an autonomous-evoked coupling index, a fatigue debt, a bias index and a treatment plasticity coefficient, and generating a control instruction through four-level arbitration. The application realizes personalized pelvic floor magnetic stimulation treatment by actively detecting and identifying individual response relations, feeding forward timing compensation and timely closing.

Description

Technical Field

[0001] This invention discloses a closed-loop magnetic stimulation method and system for pelvic floor recruitment margin and defense sequence, belonging to the technical field of medical electronic equipment and rehabilitation therapy devices. Background Technology

[0002] Currently, pelvic floor magnetic stimulation is a common non-invasive method for treating stress urinary incontinence. It typically uses a chair-type or bed-type coil assembly to achieve non-invasive stimulation, and is combined with pelvic floor surface electromyographic biofeedback technology. Existing equipment is basically open-loop or shallowly closed-loop in terms of control: before treatment, the operator presets the stimulation frequency, intensity, and fixed contraction / rest rhythm based on experience or scales. During treatment, the parameters remain unchanged or only increase according to preset steps. Even when equipped with electromyographic feedback, the feedback signal is mainly used for visual guidance, or at most, the stimulation intensity is monotonically adjusted according to the amplitude of electromyography.

[0003] However, this control method suffers from three types of mismatches that are difficult to eliminate through parameter tuning: The first is individual mismatch, as different patients have vastly different pelvic floor muscle strength, fat thickness, geometric distance between the coil and the target muscle, and neural excitation threshold, resulting in different recruitment effects of the same output parameters in different patients by several times; The second is temporal mismatch, as the fixed rhythm does not distinguish the current contraction / relaxation phase of the pelvic floor muscles, and more importantly, it cannot cope with the protective contraction demand during sudden increases in abdominal pressure, such as coughing or sneezing—some patients have normal maximum muscle strength but significant delays in pelvic floor muscle activation, and existing equipment does not measure this temporal quantity at all; The third is state mismatch, as the current scheme uses the spontaneous electromyography amplitude as the only state quantity, while low electromyography amplitude may be caused by a variety of reasons (poor recruitment ability, depletion of recruitable reserves, fatigue, poor probe fit, etc.), and the treatment directions corresponding to different causes are opposite, which cannot be distinguished in a single amplitude dimension. Summary of the Invention

[0004] The purpose of this invention is to provide a closed-loop magnetic stimulation method and system for pelvic floor recruitment margin and defense timing, in order to solve the problem in the prior art that pelvic floor magnetic stimulation devices cannot achieve precise closed-loop control of magnetic stimulation output dose, timing and mode based on individualized stimulation-muscle response relationship and pelvic floor defense timing state.

[0005] Pelvic floor recruitment margin and defense timing closed-loop magnetic stimulation methods include: S1. Acquire pelvic floor electromyography (EMG) signals using a vaginal surface EMG probe. Perform gated sampling based on the magnetic stimulation discharge time, filter and separate artifacts. When the pelvic floor muscles are in a resting state and the fatigue recovery debt index is below the upper limit of the fatigue debt threshold, inject a standardized probe magnetic pulse with an increasing gradient. Measure the evoked EMG incremental response of each pulse, fit the recruitment curve, and extract the minimum recruitment threshold, recruitment gain, saturation residual space and recovery time constant to synthesize the recruitment margin. S2. Simultaneously acquire abdominal surface electromyography, abdominal belt pressure, and trunk triaxial acceleration signals, fuse them to construct an abdominal pressure proxy signal, detect sudden rise in abdominal pressure events and trace their onset time, identify precursor waveforms and calculate precursor confidence; detect the pelvic floor muscle recruitment initiation time, and define the defense timing gap by the difference between the abdominal pressure initiation time and the recruitment initiation time; when the precursor confidence exceeds the threshold and the defense timing gap shows hysteresis, advance the stimulation trigger time to the total system delay, so that the effective stimulation time is before the abdominal pressure rise edge, and determine the compensatory stimulation intensity based on the timing gap and recruitment margin; S3. Superimpose detection pulses on the platform, measure the relationship between autonomous recruitment and the incremental value after superposition, and construct an autonomous-induced coupling index; collect multiple indicators to synthesize instantaneous fatigue load, establish a fatigue debt accumulation and recovery model, and gating stimulation with a debt threshold; extract burst-type and endurance-type functional reserves through rapid and sustained contraction tasks and calculate the bias index; use the regression slope of the recruitable margin before each treatment course as the treatment plasticity coefficient; construct a state vector from the above quantities, and generate control commands through a four-level arbitration of safety veto, mode selection, dose calculation, and timing determination to control the magnetic stimulation output.

[0006] S1 includes, S1.1, using a dual-channel probe inside the vagina to achieve a sampling rate. Acquire raw signals Constructing the gate function Screening for effective raw electromyography sequences: ; In the formula, For a moment, For the first The discharge moment of the secondary magnetic stimulation. For magnetic stimulation index, To determine the duration of the blind spot after stimulation, This serves as a protective interval before the next stimulus. The effective acquisition sequence after gating is: ; In the formula, This is the effective raw electromyographic sequence after gating. For discrete sample numbers; S1 includes, S1.2, using a combination of bandpass filtering and power frequency notch filtering for... By performing frequency band limiting and interference suppression, the purified electromyography sequence was obtained. Adaptive template cancellation processing is used. The residual artifacts from stimulation are subtracted by scaling the pure artifact template to obtain the pure evoked electromyographic waveform. ; Calculate the sliding window RMS envelope and first-order difference slope, and use the resting baseline threshold to determine the four states: ; In the formula, For four-phase labels, In a resting state, It is in a contraction-ascending state. As a platform, It is in a relaxed state. The current electromyographic envelope amplitude. Let the first-order difference of the envelope be the current moment. For absolute values, The resting baseline threshold, This is the slope threshold.

[0007] S1 includes, S1.3, constructing the detection trigger conditions. : ; In the formula, For indicator functions, The duration of the current resting phase is as follows: The resting place determination threshold, To improve the debt recovery index, The upper limit threshold for fatigue debt; Inject when it is 1; Generate a sequence of detection intensities that increase in gradient from low to high: ; In the formula, For the first The set value of magnetic stimulation intensity for the second detection. The initial detection intensity, For intensity step increments, For the index of the number of probe points, Number of detection points; Set the safety clamp: ; In the formula, The upper limit coefficient is the threshold amplitude coefficient. The motion threshold, This represents the upper limit of the patient's subjective tolerance. Set the interval between adjacent detection pulses : ; In the formula, To recover the estimated time constant; The isolated pure evoked waveform is rectified and integrated within the physiological latency window, and the pre-probe baseline is subtracted to obtain the incremental response. ; Check for shrinkage contamination within the detection window; if contamination is found, the current measurement is invalidated and retested; finally, output a validity flag. ; S1 includes S1.4, and the effective pairing set. Fitting the recruitment curve : ; In the formula, For saturation-induced amplitude, For the intensity of magnetic stimulation, For half the fundraising intensity, It is an exponential function. This is the curve steepness parameter; Three fingerprint components are derived from the fitted parameters, including the minimum recruitment threshold. , fundraising gains and saturated remaining space : ; ; ; In the formula, The intensity of the current therapeutic stimulus; The fourth component is obtained by single-exponential fitting of the relaxation segment of the post-stimulation envelope: ; In the formula, Let the envelope change over time as a function of stimulation. This represents the residual electromyographic amplitude after contraction. For the recovery time constant, This represents the resting envelope baseline value. S1 includes, in S1.5, normalizing the three fingerprint components using a reference value and then synthesizing the original margin using geometric weighting. Then, a Bayesian posterior fusion is performed with the prior mean of the population to obtain the final recruitable margin. ; Multiple rounds of probing results within the same treatment session were smoothed using exponential forgetting recursion; the first and last probing results of each treatment, along with the treatment intensity, were entered into the treatment log cohort. .

[0008] S2 includes S2.1, which adds three sensors—abdominal electromyography, abdominal belt pressure, and trunk acceleration—to the pelvic floor electromyography system. All of these sensors are driven by the same clock domain acquisition board, and each channel is pre-aligned according to the inherent delay calibrated at the factory. After rectification and filtering, the abdominal electromyography envelope, pressure signal, and acceleration modulus are extracted from each channel. The resting segment is normalized separately, and then the signals are fused together with fixed weights to form a single abdominal pressure proxy signal. When a channel fails, the weights are reallocated. Employing cumulative and change point detection algorithms, real-time monitoring is performed. The rising edge of the alarm signal, after which the accumulated data and the last zero point are traced back to determine the starting moment of the sudden increase in abdominal pressure. Simultaneously, a slope threshold is set to exclude drift values ​​below the slope threshold. S2 includes S2.2, data collected by the system during the calibration phase. The next spontaneous / involuntary cough, take The average waveform of the segment is used as a precursor template: ; In the formula, For length Precursor templates To determine the number of cough events collected during the calibration phase, For the first Abdominal pressure proxy signal of a coughing event For the first The moment when the abdominal pressure suddenly increased during this event. The length of the precursor template; Real-time matched filtering: ; In the formula, For shape matching degree, This represents the mean value within the current analysis window. The mean of the precursor template; Calculate the log-likelihood ratio With the confidence level of the omen : ; ; In the formula, Assuming an impending abdominal pressure event, Assuming no impending abdominal pressure event, for The first-order difference; The instantaneous energy operator was used to process pelvic floor electromyography (EMG), and the initiation transition was detected after short-window smoothing. Confirmation was only made if several consecutive points exceeded a threshold, thus obtaining the recruitment initiation time. .

[0009] S2 includes S2.3, the timing gap of a single event. for: ; In the formula, For the first The moment when pelvic floor muscle recruitment begins in this event; Multi-event robust summation defense against timing gaps for: ; ; In the formula, As a time-series consistency indicator, For a valid set of events, The upper quartile of the timing gap. The lower quartile of the timing gap; Constructing criteria for event validity: ; In the formula, For the first The maximum slope within the current analysis window for this event. The slope threshold for abdominal pressure events. For the first Peak value of pelvic floor electromyography envelope in the next event. The resting envelope standard deviation; Constructing a time-series state hierarchy : ; In the formula, In order to defend in advance, For synchronization, To defend against delay, This is a reference threshold for lead time.

[0010] S2 includes S2.4, where the timing of the compensation stimulus is determined by the total system delay preceding the time when the aura confidence exceeds the limit. ; ; ; ; In the formula, For algorithm detection and decision delay, Delay in preparing for stimulator charging and discharging. To delay the rise of the coil current, For electromechanical delay, The total system delay, To compensate for the timing of stimulus triggering, To trigger the confidence threshold, To compensate for the timing of the stimulus distribution, For the effective moment of stimulation; Constructing advance constraints: ; ; In the formula, This is the expected lead time from the onset of abdominal pressure. This represents the actual lead time for defense. To predict the onset time of abdominal pressure; like No compensation will be issued this time; Calculate the magnitude of the compensatory stimulus: ; ; In the formula, For time series intensity conversion factors, To compensate for the intensity of the stimulus; Compensation pulse train for: ; In the formula, To compensate for the frequency of the pulse train, To compensate for the number of pulses in the pulse train, To compensate for the duration of the pulse train; S2 includes S2.5, which uses a historical event annotation set to define the asymmetric costs of missed detections and false triggers, and searches for the optimal threshold under a hard constraint on the false trigger rate. Alarm threshold in cumulative and change point detection algorithms And adjust online after each treatment; Introducing compensation gain The actual distribution intensity is: ; Evaluate the timing improvement based on event blocks, if and If the target is met, then decrease. If it worsens, it increases; at the end of each block, an uncompensated assessment event is inserted to measure the timing gap without assistance and confirm the improvement of autonomous capabilities.

[0011] S3 includes S3.1, where a standard probe pulse of fixed intensity is superimposed on the voluntary contraction plateau phase to measure the incremental amount of external stimulation that can be recruited after voluntary exertion. and with the level of spontaneous contraction contrast: ; ; ; In the formula, For autonomously induced coupling index, To prevent division by zero, To be non-irritating To maintain a fixed detection intensity; by and Patients were classified into four types, including those with autonomous recruitment disorder. Low fundraising reserves Near saturation Well-functioning : ; In the formula, and This is the partition threshold. Subscription status classification tags.

[0012] S3 includes S3.2, and the calculation platform attenuation rate. Residual ratio after shrinkage Recovery time ratio Repeated response decay and detection response attenuation : ; ; ; ; ; In the formula, For platform envelope function, The maximum spontaneous contraction envelope peak value, To restore the time constant reference value, For the first The same intensity of treatment stimulates a response. This represents the initial recruitable margin prior to this treatment. This represents the available fundraising margin at the current moment; Synthetic instantaneous fatigue load : ; In the formula, , , , , The weighting coefficients for fatigue sub-indicators; Construct a continuous-time first-order model: ; In the formula, This is the fatigue accumulation rate coefficient. To stimulate the switch state, For individual recovery time constant; Calculate fatigue recovery debt : ; In the formula, For discrete iteration time steps, for Moment ; when Stimulation is prohibited when the threshold is above the high threshold, and allowed when the threshold is below the low threshold, maintaining the previous state in between; the remaining rest time is predicted by debt dynamics analysis and a countdown prompt is given.

[0013] S3 includes S3.3, which extracts burst-type functional reserves through rapid short contraction and sustained plateau contraction, respectively. and endurance-type functional reserves Calculate the bias index : ; In the formula, To prepare for explosive functionalities, Reserved for endurance-type functions; From the archive queue Extracting from each treatment session Perform a weighted linear regression with a slope of ; Residual determination platform; Constructing state vectors : ; In the formula, For the treatment of plasticity coefficient; The four-level arbitration generation control command is used, with the first level being secure arbitration. ; In the formula, For impedance, For the contact self-test to pass, High threshold for fatigue debt This represents the lower limit of the saturated remaining space. The second level is mode selection: ; In the formula, It is a timing compensation mode. To facilitate the fundraising model, Endurance maintenance mode It is a self-guided mode. For the variant pattern, As a platform, For a plasticity status label, This is the threshold for bias determination; The third level is the dose continuity law: ; ; ; In the formula, The command intensity after dose-rate calculation. As a reference for dose step size, As the plasticity direction factor, This is the upper limit coefficient of the treatment intensity relative to the exercise threshold. This represents the upper limit of the patient's subjective tolerance. To ensure a rapid response, As a platform, For degeneration; The fourth level is the timing of the stimulus: ; ; In the formula, To stimulate output timing, Enable bit for secure arbitration, For the currently selected stimulation mode, It is in a relaxed state; At the end of treatment, archive and calculate the recommended initial values ​​for the next treatment: ; ; ; In the formula, This is the treatment sequence number. This is the upper limit coefficient for adjustment between treatment courses. The threshold for plasticity significance. For the dose efficiency of each mode, This is the pooling margin available for the final treatment. For this treatment, It is a symbolic function; At the end of treatment, the average intensity and efficiency of each mode are archived. The mode with the highest efficiency is selected as the priority mode for the next treatment. The starting intensity of the next treatment is slightly adjusted according to the plasticity direction based on the current average. At the same time, three trend curves are output: autonomous temporal improvement, plasticity coefficient, and autonomous recruitment efficiency.

[0014] The pelvic floor recruitment margin and defense timing closed-loop magnetic stimulation system includes: A dual-channel surface electromyography probe is installed inside the vagina to collect pelvic floor electromyography signals; Pelvic floor magnetic stimulation coil and its driving circuit are used to output magnetic stimulation pulses; An abdominal pressure event sensing component consisting of abdominal surface electromyography electrodes, an abdominal belt-type pressure sensor, and a trunk triaxial accelerometer; A gated acquisition board that shares the clock domain with the stimulus timing generator; The real-time processing and control host is used to perform probe pulse injection and recruitable margin synthesis, abdominal pressure proxy signal construction and defense timing gap calculation and compensation stimulus triggering, six-dimensional state vector construction and four-level arbitration control command generation; The safety interlocking unit includes a handheld stop switch, electrode contact impedance self-test, and dual-sided output strength clamping. Human-computer interaction unit; each unit is connected via a data bus and its operation is coordinated by the host.

[0015] Compared with existing technologies, this invention has the following beneficial effects: Firstly, by actively injecting standardized gradient probe pulses into the pelvic floor muscles and measuring incremental responses, this invention quantifies individual differences into four fingerprint parameters that can be compared across treatment courses: recruitment threshold, recruitment gain, saturation margin, and recovery time constant. This solves the problem of dose mismatch caused by "same electromyographic amplitude but different recruitment potential." Secondly, by simultaneously acquiring multimodal abdominal pressure event signals and identifying precursor waveforms, the effective stimulation time is shifted forward to before the rising edge of abdominal pressure, providing a targeted temporal compensation treatment for the stress urinary incontinence subtype characterized by "normal muscle strength but delayed contraction initiation." Thirdly, by constructing a state-gated scheduling system that incorporates fatigue debt accumulation and recovery dynamics, the stimulation-rest rhythm adapts to the individual's recovery speed, replacing the fixed duty cycle. Fourthly, through a six-dimensional state vector and a four-level arbitration mechanism, safety veto, mode selection, dosage, and timing are coordinated within a unified framework, achieving precise closed-loop control of "adding only when there is leeway, stopping when fatigue occurs, not increasing dosage when the response is good, and not stimulating during the relaxation phase," significantly improving the individual adaptability and safety of treatment. Attached Figure Description

[0016] Figure 1 This is a flowchart of the technology of this invention. Detailed Implementation

[0017] To make the objectives, technical solutions, and advantages of this invention clearer, the technical solutions of this invention are described clearly and completely below. Obviously, the described embodiments are only some, not all, of the embodiments of this invention. All other embodiments obtained by those skilled in the art based on the embodiments of this invention without creative effort are within the scope of protection of this invention.

[0018] Pelvic floor recruitment margin and defense timing closed-loop magnetic stimulation methods include: S1. Acquire pelvic floor electromyography (EMG) signals using a vaginal surface EMG probe. Perform gated sampling based on the magnetic stimulation discharge time, filter and separate artifacts. When the pelvic floor muscles are in a resting state and the fatigue recovery debt index is below the upper limit of the fatigue debt threshold, inject a standardized probe magnetic pulse with an increasing gradient. Measure the evoked EMG incremental response of each pulse, fit the recruitment curve, and extract the minimum recruitment threshold, recruitment gain, saturation residual space and recovery time constant to synthesize the recruitment margin. S2. Simultaneously acquire abdominal surface electromyography, abdominal belt pressure, and trunk triaxial acceleration signals, fuse them to construct an abdominal pressure proxy signal, detect sudden rise in abdominal pressure events and trace their onset time, identify precursor waveforms and calculate precursor confidence; detect the pelvic floor muscle recruitment initiation time, and define the defense timing gap by the difference between the abdominal pressure initiation time and the recruitment initiation time; when the precursor confidence exceeds the threshold and the defense timing gap shows hysteresis, advance the stimulation trigger time to the total system delay, so that the effective stimulation time is before the abdominal pressure rise edge, and determine the compensatory stimulation intensity based on the timing gap and recruitment margin; S3. Superimpose detection pulses on the platform, measure the relationship between autonomous recruitment and the incremental value after superposition, and construct an autonomous-induced coupling index; collect multiple indicators to synthesize instantaneous fatigue load, establish a fatigue debt accumulation and recovery model, and gating stimulation with a debt threshold; extract burst-type and endurance-type functional reserves through rapid and sustained contraction tasks and calculate the bias index; use the regression slope of the recruitable margin before each treatment course as the treatment plasticity coefficient; construct a state vector from the above quantities, and generate control commands through a four-level arbitration of safety veto, mode selection, dose calculation, and timing determination to control the magnetic stimulation output.

[0019] S1 includes, S1.1, using a dual-channel probe inside the vagina to achieve a sampling rate. Acquire raw signals Constructing the gate function Screening for effective raw electromyography sequences: ; In the formula, For a moment, For the first The discharge moment of the secondary magnetic stimulation. For magnetic stimulation index, To determine the duration of the blind spot after stimulation, This serves as a protective interval before the next stimulus. The effective acquisition sequence after gating is: ; In the formula, This is the effective raw electromyographic sequence after gating. For discrete sample numbers; S1 includes, S1.2, using a combination of bandpass filtering and power frequency notch filtering for... By performing frequency band limiting and interference suppression, the purified electromyography sequence was obtained. Adaptive template cancellation processing is used. The residual artifacts from stimulation are subtracted by scaling the pure artifact template to obtain the pure evoked electromyographic waveform. ; Calculate the sliding window RMS envelope and first-order difference slope, and use the resting baseline threshold to determine the four states: ; In the formula, For four-phase labels, In a resting state, It is in a contraction-ascending state. As a platform, It is in a relaxed state. The current electromyographic envelope amplitude. Let the first-order difference of the envelope be the current moment. For absolute values, The resting baseline threshold, This is the slope threshold.

[0020] S1 includes, S1.3, constructing the detection trigger conditions. : ; In the formula, For indicator functions, The duration of the current resting phase is as follows: The resting place determination threshold, To improve the debt recovery index, The upper limit threshold for fatigue debt; Inject when it is 1; Generate a sequence of detection intensities that increase in gradient from low to high: ; In the formula, For the first The set value of magnetic stimulation intensity for the second detection. The initial detection intensity, For intensity step increments, For the index of the number of probe points, Number of detection points; Set the safety clamp: ; In the formula, The upper limit coefficient is the threshold amplitude coefficient. The motion threshold, This represents the upper limit of the patient's subjective tolerance. Set the interval between adjacent detection pulses : ; In the formula, To recover the estimated time constant; The isolated pure evoked waveform is rectified and integrated within the physiological latency window, and the pre-probe baseline is subtracted to obtain the incremental response. ; Check for shrinkage contamination within the detection window; if contamination is found, the current measurement is invalidated and retested; finally, output a validity flag. ; S1 includes S1.4, and the effective pairing set. Fitting the recruitment curve : ; In the formula, For saturation-induced amplitude, For the intensity of magnetic stimulation, For half the fundraising intensity, It is an exponential function. This is the curve steepness parameter; Three fingerprint components are derived from the fitted parameters, including the minimum recruitment threshold. , fundraising gains and saturated remaining space : ; ; ; In the formula, The intensity of the current therapeutic stimulus; The fourth component is obtained by single-exponential fitting of the relaxation segment of the post-stimulation envelope: ; In the formula, Let the envelope change over time as a function of stimulation. This represents the residual electromyographic amplitude after contraction. For the recovery time constant, This represents the resting envelope baseline value. S1 includes, S1.5, normalizing the three fingerprint components using reference values ​​and then synthesizing the original margin using geometric weighting. Then, a Bayesian posterior fusion is performed with the prior mean of the population to obtain the final recruitable margin. ; Multiple rounds of probing results within the same treatment session were smoothed using exponential forgetting recursive smoothing; the first and last probing results of each treatment, along with the treatment intensity, were entered into the treatment record cohort. .

[0021] S2 includes S2.1, which adds three sensors—abdominal electromyography, abdominal belt pressure, and trunk acceleration—to the pelvic floor electromyography system. All of these sensors are driven by the same clock domain acquisition board, and each channel is pre-aligned according to the inherent delay calibrated at the factory. After rectification and filtering, the abdominal electromyography envelope, pressure signal, and acceleration modulus are extracted from each channel. The resting segment is normalized separately, and then the signals are weighted and fused according to fixed weights to form a single abdominal pressure proxy signal. When a channel fails, the weights are reallocated. Employing cumulative and change point detection algorithms, real-time monitoring is performed. The rising edge of the alarm signal, after which the accumulated data and the last zero point are traced back to determine the starting moment of the sudden increase in abdominal pressure. Simultaneously, a slope threshold is set to exclude drift values ​​below the slope threshold. S2 includes S2.2, data collected by the system during the calibration phase. The next spontaneous / involuntary cough, take The average waveform of the segment is used as a precursor template: ; In the formula, For length Precursor templates To determine the number of cough events collected during the calibration phase, For the first Abdominal pressure proxy signal of a coughing event For the first The moment when the abdominal pressure suddenly increased during this event. The length of the precursor template; Real-time matched filtering: ; In the formula, For shape matching degree, This represents the mean value within the current analysis window. This represents the mean of the precursor template; Calculate the log-likelihood ratio With the confidence level of the omen : ; ; In the formula, Assuming an impending abdominal pressure event, Assuming no impending abdominal pressure event, for The first-order difference; The instantaneous energy operator was used to process pelvic floor electromyography (EMG), and the initiation transition was detected after short-window smoothing. Confirmation was only made if several consecutive points exceeded a threshold, thus obtaining the recruitment initiation time. .

[0022] S2 includes S2.3, the timing gap of a single event. for: ; In the formula, For the first The moment when pelvic floor muscle recruitment begins in this event; Multi-event robust summation defense against timing gaps for: ; ; In the formula, As a time-series consistency indicator, For a valid set of events, The upper quartile of the timing gap. The lower quartile of the timing gap; Constructing criteria for event validity: ; In the formula, For the first The maximum slope within the current analysis window for this event. The slope threshold for abdominal pressure events. For the first Peak value of pelvic floor electromyography envelope in the next event. The resting envelope standard deviation; Constructing a time-series state hierarchy : ; In the formula, In order to defend in advance, For synchronization, To defend against delay, This is a reference threshold for lead time.

[0023] S2 includes S2.4, where the timing of the compensation stimulus is determined by the total system delay preceding the time when the aura confidence exceeds the limit. ; ; ; ; In the formula, For algorithm detection and decision delay, Delay in preparing for stimulator charging and discharging. To delay the rise of the coil current, For electromechanical delay, The total system delay, To compensate for the timing of stimulus triggering, To trigger the confidence threshold, To compensate for the timing of the stimulus distribution, For the effective moment of stimulation; Constructing advance constraints: ; ; In the formula, This is the expected lead time from the onset of abdominal pressure. This represents the actual lead time for defense. To predict the onset time of abdominal pressure; like No compensation will be issued this time; Calculate the magnitude of the compensatory stimulus: ; ; In the formula, For time series intensity conversion factors, To compensate for the intensity of the stimulus; Compensation pulse train for: ; In the formula, To compensate for the frequency of the pulse train, To compensate for the number of pulses in the pulse train, To compensate for the duration of the pulse train; S2 includes S2.5, which uses a historical event annotation set to define the asymmetric costs of missed detections and false triggers, and searches for the optimal threshold under a hard constraint on the false trigger rate. Alarm threshold in cumulative and change point detection algorithms And adjust online after each treatment; Introducing compensation gain The actual distribution intensity is: ; Evaluate the timing improvement based on event blocks, if and If the target is met, then decrease. If it worsens, it increases; at the end of each block, an uncompensated assessment event is inserted to measure the timing gap without assistance and confirm the improvement of autonomous capabilities.

[0024] S3 includes S3.1, where a standard probe pulse of fixed intensity is superimposed on the voluntary contraction plateau phase to measure the incremental amount of external stimulation that can be recruited after voluntary exertion. and with the level of spontaneous contraction contrast: ; ; ; In the formula, For autonomously induced coupling index, To prevent division by zero, To be non-irritating To maintain a fixed detection intensity; by and Patients were classified into four types, including those with autonomous recruitment disorder. Low fundraising reserves Near saturation Well-functioning : ; In the formula, and This is the partition threshold. Subscription status classification tags.

[0025] S3 includes S3.2, and the calculation platform attenuation rate. Residual ratio after shrinkage Recovery time ratio Repeated response decay and detection response attenuation : ; ; ; ; ; In the formula, For platform envelope function, The maximum spontaneous contraction envelope peak value, To restore the time constant reference value, For the first The same intensity of treatment stimulates a response. This represents the initial recruitable margin prior to this treatment. This represents the available fundraising margin at the current moment; Synthetic instantaneous fatigue load : ; In the formula, , , , , The weighting coefficients for fatigue sub-indicators; Construct a continuous-time first-order model: ; In the formula, This is the fatigue accumulation rate coefficient. To stimulate the switch state, For individual recovery time constant; Calculate fatigue recovery debt : ; In the formula, For discrete iteration time steps, for Moment ; when Stimulation is prohibited when the threshold is above the high threshold, and allowed when the threshold is below the low threshold, maintaining the previous state in between; the remaining rest time is predicted by debt dynamics analysis and a countdown prompt is given.

[0026] S3 includes S3.3, which extracts burst-type functional reserves through rapid short contraction and sustained plateau contraction, respectively. and endurance-type functional reserves Calculate the bias index : ; In the formula, To prepare for explosive functionalities, Reserved for endurance-type functions; From the archive queue Extracting from each treatment session Perform a weighted linear regression with a slope of ; Residual determination platform; Constructing state vectors : ; In the formula, For the treatment of plasticity coefficient; The four-level arbitration generation control command is used, with the first level being secure arbitration. ; In the formula, For impedance, For the contact self-test to pass, High threshold for fatigue debt This represents the lower limit of the saturated remaining space. The second level is mode selection: ; In the formula, It is a timing compensation mode. To facilitate the fundraising model, Endurance maintenance mode It is a self-guided mode. For the variant pattern, As a platform, For a plasticity status label, This is the threshold for bias determination; The third level is the dose continuity law: ; ; ; In the formula, The command intensity after dose-rate calculation. As a reference for dose step size, As the plasticity direction factor, This is the upper limit coefficient of the treatment intensity relative to the exercise threshold. This represents the upper limit of the patient's subjective tolerance. To ensure a rapid response, As a platform, For degeneration; The fourth level is the timing of the stimulus: ; ; In the formula, To stimulate output timing, Enable bit for secure arbitration, For the currently selected stimulation mode, It is in a relaxed state; At the end of treatment, archive and calculate the recommended initial values ​​for the next treatment: ; ; ; In the formula, This is the treatment sequence number. This is the upper limit coefficient for adjustment between treatment courses. The threshold for plasticity significance. For the dose efficiency of each mode, This is the pooling margin available for the final treatment. For this treatment, It is a symbolic function; At the end of treatment, the average intensity and efficiency of each mode are archived. The mode with the highest efficiency is selected as the priority mode for the next treatment. The starting intensity of the next treatment is slightly adjusted according to the plasticity direction based on the current average. At the same time, three trend curves are output: autonomous temporal improvement, plasticity coefficient, and autonomous recruitment efficiency.

[0027] The pelvic floor recruitment margin and defense timing closed-loop magnetic stimulation system includes: A dual-channel surface electromyography probe is installed inside the vagina to collect pelvic floor electromyography signals; Pelvic floor magnetic stimulation coil and its driving circuit are used to output magnetic stimulation pulses; An abdominal pressure event sensing component consisting of abdominal surface electromyography electrodes, an abdominal belt-type pressure sensor, and a trunk triaxial accelerometer; A gated acquisition board that shares the clock domain with the stimulus timing generator; The real-time processing and control host is used to perform probe pulse injection and recruitable margin synthesis, abdominal pressure proxy signal construction and defense timing gap calculation and compensation stimulus triggering, six-dimensional state vector construction and four-level arbitration control command generation; The safety interlocking unit includes a handheld stop switch, electrode contact impedance self-test, and dual-sided output strength clamping. Human-computer interaction unit; each unit is connected via a data bus and its operation is coordinated by the host.

[0028] The following description, in conjunction with the accompanying drawings, further illustrates the process of this invention. Figure 1 As shown, the first stage is active detection and fingerprint extraction, which involves injecting gradient detection pulses into the pelvic floor muscles, fitting a recruitment curve, and calculating the recruitment margin. The second stage is timing compensation and aura triggering, which calculates the defense timing gap based on the aura recognition results and implements feedforward compensation accordingly. The third stage is state fusion and four-level arbitration, which integrates six-dimensional state information and generates control commands through four levels of arbitration: safety, mode, dosage, and timing. The above three stages are executed sequentially, and the output of the third stage is fed back to the first stage through a feedback loop to guide the adjustment of subsequent detection parameters. The three core components of the entire process are: gradient detection → recruitment curve → recruitment margin, aura recognition → defense timing gap → feedforward compensation, and six-dimensional state → safety / mode / dosage / timing.

[0029] The following description, in conjunction with an embodiment, further illustrates the use of a vaginal implanted dual-channel Glazer-type sEMG probe with a sampling rate of [missing information]. Acquire raw signals Each discharge of the magnetic stimulation coil generates an electromagnetic artifact on the electromyographic channel with an amplitude far exceeding that of the electromyographic signal itself. Therefore, the acquisition is not continuous, but rather the gating window is controlled by the synchronization signal output by the stimulation timing generator.

[0030] Let the first The discharge time of the secondary magnetic stimulation is Then the gating function is: ; The valid acquisition sequence after gating is: ; Input: The raw dual-channel electromyographic voltage picked up by the probe, in units of... ; Sampling rate, taking (This is to ensure that enough sample points can still be obtained within the 5–10 ms stimulation interval). : No. The discharge timing of each magnetic stimulation pulse is directly given by the stimulation timing generator, without needing to be deduced from electromyography. : Duration of blind zone after stimulation, taken This is used to skip the saturation phase during discharge. The protective interval before the next stimulus is taken as follows: To prevent the leading edge from climbing into the sampling window; : Intermediate variable, gating indicator function, takes the value 0 / 1; Output: The effective raw electromyographic sequence after gating.

[0031] A combination of bandpass filtering and power frequency notch filtering is used to... By performing frequency band limiting and interference suppression, the purified electromyography sequence was obtained. The specific process includes normalizing the transfer function of the bandpass segment (low-pass prototype). ): ; Notch band transfer function: ; Cascaded output: .

[0032] Input: Gated electromyography sequence; : Filter order; , Lower and upper cutoff angular frequencies: suppress motion artifacts and probe displacement drift below 20Hz, and suppress high-frequency noise above 450Hz; : Power frequency normalized angular frequency; The radius of the notch pole determines the bandwidth of the notch; the closer it is to 1, the narrower the bandwidth. Output: purified electromyography sequence.

[0033] Adaptive template cancellation processing The residual artifacts from stimulation are subtracted by scaling the pure artifact template to obtain the pure evoked electromyographic waveform. The specific process includes: Even after gating and filtering, an exponentially decaying residual artifact tail remains immediately following the stimulation pulse, while the induced electromyographic response (M wave) also occurs during this period (latency approximately 10–25 ms), resulting in temporal overlap. This step separates the two.

[0034] Length was taken after each stimulus. Alignment segment (by Alignment), artifact template From the zero output condition (stimulus intensity below the motor threshold) (The data collected only produces artifacts and does not generate electromyographic responses) Taking the median point by point of the segment yields: ; For each actual stimulus, the template scaling factor is estimated using least squares: ; Separated evoked myoelectric potential residuals: ; The template is recursively updated based on exponential forgetting to track coil temperature drift and probe position micro-movements. ; : Input, number Alignment of electromyographic segments after the second stimulation; Alignment segment length, take Corresponding sample number; Pure artifact fitting window, after stimulus... This window is earlier than the M-wave latency period, therefore It is determined solely by artifacts and is not contaminated by electromyographic responses; : Intermediate variable, artifact shape template; : Intermediate variable, the magnitude of the artifact relative to the template, which changes linearly with the intensity of the stimulus; Forgetting factor; The patient's motor threshold stimulus intensity; Output, number Purely evoked electromyographic waveforms of a single stimulus.

[0035] Calculate the sliding window RMS envelope and first-order difference slope, and perform four-state determination in conjunction with the resting baseline threshold. The specific process includes: envelope: ; First-order difference (slope): ; Resting baseline and trigger threshold: ; Phase state machine output: ; RMS window length, take The corresponding number of sample points is used to balance smoothness and response latency. Difference interval, take Corresponding sample number; The resting calibration window is 5 seconds before the start of treatment; Intermediate variables: median and standard deviation of the resting envelope; Threshold multiple; : Slope threshold, take ; Output, electromyographic envelope ( ), update rate ; Output: Quad-phase labels.

[0036] Construct detection trigger conditions : ; Generate a sequence of detection intensities that increase in gradient from low to high: ; Set the safety clamp: ; Set the interval between adjacent detection pulses : ; : Intermediate variable, probe permission flag, injection is allowed when it is 1; The current resting phase has lasted for a certain duration; Ensure that electromyography has returned to baseline; Fatigue—Recover debts. Prohibit detection when fatigue has not been recovered to avoid detection itself causing additional fatigue and contaminating fingerprints; Initial detection intensity, taken as the maximum output of the device. ; Intensity step, take ; : Number of detection points, take ; Motor threshold, determined by the binary search at the time of the first treatment (when the evoked response first exceeds...). (minimum strength). : The above-threshold limit coefficient ensures that the detection intensity does not enter the therapeutic dose range and that the detection does not constitute an additional therapeutic load; The patient's subjective tolerance limit is obtained by using a handheld stop switch and tolerance calibration. : The recovery time constant obtained from the previous fit, making the interval adaptable to the individual recovery rate; Output: Input side of the probe paired dataset. .

[0037] The isolated pure evoked waveform is rectified and integrated within the physiological latency window, and the pre-probe baseline is subtracted to obtain the incremental response. The process involves checking for shrinkage contamination within the detection window; if contamination is found, the current point is discarded and the measurement is repeated. Specifically, this includes analyzing the separated pure induced waveform. (corresponding detection intensity) Integrating within the physiological latency window: ; Subtracting the baseline before detection yields the incremental response: ; Validity criteria (if not met, the detection point is invalidated and must be retested): ; Input, purely evoked electromyography; M-wave integral window, after stimulation ; The number of sample points in this window; : Detect the baseline window before stimulation ; : Intermediate variable, the first The average rectified induced amplitude of each detection point ( ); : Intermediate variable, the level of spontaneous electromyography prior to this detection; : Upper limit of the slope of autonomous activity, take If a patient spontaneously contracts within the detection window, it is considered contamination. Output: Purely exogenous induced recruitment increment; Output: Validity flag.

[0038] For the effective pairing set Fitting the recruitment curve : ; In the formula, For saturation-induced amplitude, For the intensity of magnetic stimulation, For half the fundraising intensity, It is an exponential function. This is the curve steepness parameter; Three fingerprint components are derived from the fitted parameters, including the minimum recruitment threshold. , fundraising gains and saturated remaining space : ; ; ; In the formula, The intensity of the current therapeutic stimulus; The fourth component is obtained by single-exponential fitting of the relaxation segment of the post-stimulation envelope: ; In the formula, Let the envelope change over time as a function of stimulation. This represents the residual electromyographic amplitude after contraction. For the recovery time constant, This represents the resting envelope baseline value.

[0039] After normalizing the three fingerprint components using a reference value, the original margin is synthesized using geometric weighting. Then, a Bayesian posterior fusion is performed with the prior mean of the population to obtain the final recruitable margin. The specific process includes first normalizing the three components to a reference value (eliminating differences in dimensions and equipment): ; Synthetic initial margin: ; By fusing posterior priors with population priors, random errors from single detections can be suppressed. ; The patient's peak value of the periodic voluntary contraction envelope ( ), used to convert induced gain into "a proportion relative to one's own muscle strength"; Reference stimulus intensity increment (take) Device output), enabling Dimensionless; Threshold reference value (take) Device output). The higher The smaller the value, the greater the threshold penalty term. Weight, default It can be configured according to the indication and written into the device parameter table; Intermediate variables: margins and ranges directly calculated in this probe. ; same sex and postpartum period The prior mean and standard deviation are stored in the equipment parameter library; The observation uncertainty of this fitting is composed of the fitting residuals and the number of significant points. estimate; Output: Final value of the fundraising margin.

[0040] Multiple rounds of probing results within the same treatment session were smoothed using exponential forgetting recursion; the first and last probing results of each treatment, along with the treatment intensity, were entered into the treatment log cohort. The specific process includes multiple rounds of probing within the same treatment session: ; Inter-treatment archiving: The fingerprints and margins of the first probe (pre-treatment state) and the last probe (post-treatment state) for each treatment are written into a queue. ; The number of detection rounds during this treatment; : A forgetting factor during the treatment course, balancing smoothing and the speed of tracking changes in fatigue; Treatment sequence number; Number of treatments completed; Output: fingerprint time sequence archive, which is the therapeutic plasticity coefficient. The sole source of data; : No. The intensity of treatment used in this treatment, and Archive them together to ensure that comparisons across treatment courses are conducted under the same detection conditions.

[0041] Based on pelvic floor electromyography (sEMG), three additional sensors—abdominal electromyography (EMG), abdominal belt pressure, and trunk acceleration—are added, all driven by a single clock domain acquisition board. Each channel is pre-aligned according to its factory-calibrated inherent delay. Specifically, in addition to the dual pelvic floor sEMG channels, three event sensing channels are added, all driven by the same clock domain acquisition board, along with abdominal surface EMG. External oblique muscle / rectus abdominis muscle Intra-abdominal pressure proxy : A flexible piezoresistive sensor (or an intravaginal balloon pressure tube) placed inside the patient's abdominal binder. Triaxial acceleration : Fixed to the lower edge of the sternum, It is used to capture the trunk impact of coughing / jumping.

[0042] Unified time base alignment (based on sEMG time base at the bottom of the basin) (Based on) ; : Input, number Raw signal from the road sensor; Original sampling rate for each channel; Unified resampling target rate; : No. The inherent transmission / filter group delay (s) of the channel, factory calibration value, and the fluid delay of the pressure line are approximately... The digital filter delay of the accelerometer is approximately ; Output: a sequence of event channels that are strictly aligned with pelvic floor electromyography.

[0043] After rectification and filtering, the abdominal electromyography envelope, pressure signal, and acceleration modulus are extracted from each channel. The resting segment is normalized separately, and then the signals are fused together with fixed weights to form a single abdominal pressure proxy signal. When a channel fails, weights are reassigned. The specific process includes preprocessing of each channel: rectified low-pass envelope is used for abdominal electromyography, and second-order Butterworth low-pass envelope is used for pressure and acceleration (cutoff). The formula is uniformly expressed as ; Acceleration composition modulus and removal of gravitational component: ; Each channel performs z-normalization on its own resting segment and then weights and fuses them to form an abdominal pressure proxy signal. : ; Intermediate variable, the first Preprocessed signals (abdominal muscle electrical envelope / pressure / acceleration modulus); Low-pass cutoff frequency, preserving the rising edge of cough abdominal pressure (typical rise time) ) and filter out high-frequency jitter; Calibrated value of gravitational modulus in the stationary section; Mean and standard deviation of the quiet breathing segment 10 seconds before the start of treatment; : Blending weights, default The pressure channel has the highest weight; when a channel becomes saturated or falls off (self-test fails), its weight is reset to 0 and the remaining channels are reset to one. Output: Dimensionless abdominal pressure surrogate quantity.

[0044] Employing cumulative and change point detection algorithms, real-time monitoring is performed. The rising edge of the alarm signal, after which the accumulated data and the last zero point are traced back to determine the starting moment of the sudden increase in abdominal pressure. Simultaneously, a slope threshold is set to exclude drift values ​​below the slope threshold; the specific process includes accumulating and testing statistics: ; Alarm conditions and event origin tracing: ; Slope confirmation (suppressing false events caused by slow drift): ; Resting phase Mean (approximately 0 after normalization); : Minimum detectable offset, take (i.e., 2 resting standard deviations) determines sensitivity; Alarm threshold, take It is adaptively adjusted according to the false trigger rate constraint; Intermediate variables, cumulative sum statistics; Intermediate variables, alarm sample points; Output: The starting time (s) of the rapid increase in abdominal pressure, obtained by backtracking from the CUSUM zero point, rather than the alarm point itself; : Slope inspection window (take ); : Slope threshold.

[0045] The system collects data during the calibration phase. The next spontaneous / involuntary cough, take The average waveform of the segment is used as a precursor template: ; In the formula, For length Precursor templates To determine the number of cough events collected during the calibration phase, For the first Abdominal pressure proxy signal of a coughing event For the first The moment when the abdominal pressure suddenly increased during this event. Precursor template length ( (corresponding sample number) Real-time matched filtering: ; In the formula, For shape matching degree ( ), This represents the mean value within the current analysis window. This represents the mean of the precursor template; Calculate the log-likelihood ratio With the confidence level of the omen : ; ; In the formula, Assuming an impending abdominal pressure event, Assuming no impending abdominal pressure event, for The first-order difference.

[0046] The instantaneous energy operator was used to process pelvic floor electromyography (EMG), and the initiation transition was detected after short-window smoothing. Confirmation was only made if several consecutive points exceeded a threshold, thus obtaining the recruitment initiation time. The specific process includes directly targeting the envelope. Taking the threshold introduces a fixed lag due to the envelope smoothing window (50ms), so TKEO is used to enhance the initial transition first: ; Its moving average and detection criteria: ; ; Input: Pelvic floor electromyography (without using envelope) As a starting point criterion, it is only used as the phase background. Intermediate variable, TKEO instantaneous energy, is sensitive to the combined abrupt change in signal amplitude and frequency, and can amplify the electromyographic initiation edge; TKEO smooth window, take It is much shorter than the envelope window, thus reducing the start point determination lag from about 25ms to about 5ms; : Mean and standard deviation of TKEO at rest (from rest calibration window) ); Threshold multiple; : Number of consecutively confirmed sample points (corresponding to) ), to prevent misjudgment of single-point spikes; Output: The moment (s) at which the pelvic floor muscles begin to recruit.

[0047] S2 includes S2.3, the timing gap of a single event. for: ; Multi-event robust summation defense against timing gaps for: ; ; Constructing criteria for event validity: ; Constructing a time-series state hierarchy : ; : Input, number The next event moment; : Intermediate variable, a positive value indicates that the pelvic floor muscles start before abdominal pressure (there is a margin of defense), and a negative value indicates that the pelvic floor muscles contract only after the abdominal pressure has increased (defense lag). : Valid event set, the three conditions are used to exclude three types of contamination: "insufficiently intense abdominal pressure event", "no response from pelvic floor muscles", and "abnormal detection (outside the physiological range)"; Lead time reference threshold, take ; Intermediate variable, temporal consistency index: a large interquartile range indicates unstable temporal control in patients, and this value is used as one of the withdrawal criteria.

[0048] The timing of the compensation stimulus is determined by the total system delay preceding the time when the aura confidence exceeds the limit: ; ; ; ; Constructing advance constraints: ; ; like No compensation will be issued this time; Calculate the magnitude of the compensatory stimulus: ; ; The compensation pulse train uses short, high-frequency packets: ; : Trigger threshold; Algorithm detection and decision delay (approximately) ); Stimulator charging and discharging preparation delay (approximately) ); : Coil current rise delay (approximately ); Electromechanical delay, i.e., the physiological delay from the onset of electromyography to the increase in tone (approximately... (by periodic on-site measurement) Intermediate variable, total system delay, is the amount that must be feedforward compensated in this step; Intermediate variable: the expected lead time from the onset of prodromal symptoms to the onset of abdominal pressure; sampling period. ; : Intermediate variable, the actual defensive lead time obtained, if No compensation will be issued this time (issuing it would be meaningless and would only increase fatigue). Timing-Intensity Conversion Factor (Unit: Device Output % / ms), Default ; The intensity of this compensatory stimulus; Stimulus waveform parameters.

[0049] Using a historical event annotation set, we define the asymmetric costs of missed detections and false triggers, and search for the optimal threshold under a hard constraint on the false trigger rate. and cumulative and threshold And adjust it online recursively after each treatment; the specific process includes defining a cost function using a labeled set consisting of a calibration period and historical events: ; Solving for the operating point under the false trigger rate constraint: ; Online push update (after each treatment): ; The false negative rate and false trigger rate are calculated from the labeled events. Cost weight, default (The cost of false stimulation is higher than that of missed detection, because false stimulation directly consumes fatigue budget). : Intermediate variable, number of false triggers per unit time (times / minute); : Maximum allowed; Threshold learning rate; Treatment sequence number; The write-back threshold forms an internal secondary closed loop.

[0050] Introducing compensation gain The actual distribution intensity was changed to: ; Gradual withdrawal rule (every) (This event is an evaluation block) ; Confirmation of autonomous advance capability (an "assessment event" without compensation is inserted at the end of each block): ; Intermediate variable, used to compensate for gain, initial value 1.0; Adjust the step size for each block; : Evaluate the number of events within the block; Block number; , Lead time and consistency achievement threshold; Intermediate variable, the time gap under uncompensated conditions, reflects the patient's own defense time capability; Improvement in autonomous defense timing (ms), and the therapeutic plasticity coefficient Parallel, as a time-series indicator of treatment efficacy, it is written into the record queue. ; Evaluate the timing improvement based on event blocks, if and If the target is met, then decrease. If it worsens, it increases; at the end of each block, an uncompensated assessment event is inserted to measure the timing gap without assistance and confirm the improvement of autonomous capabilities.

[0051] During the autonomous contraction plateau phase, a standard probe pulse of fixed intensity is superimposed to measure the incremental amount of external stimulation that can be recruited after autonomous exertion. and with the level of spontaneous contraction contrast: ; ; ; In the formula, For autonomously induced coupling index, To prevent division by zero, To be non-irritating To maintain a fixed detection intensity; by and Patients were classified into four types, including those with autonomous recruitment disorder. Low fundraising reserves Near saturation Well-functioning : ; ( range ), : Partition threshold, the factory default value can be configured according to the population group; : Sub-labels for fundraising status.

[0052] Calculate the platform attenuation rate Residual ratio after shrinkage Recovery time ratio Repeated response decay and detection response attenuation : ; ; ; ; ; In the formula, For platform envelope function, The maximum spontaneous contraction envelope peak value, To restore the time constant reference value, For the first The same intensity of treatment stimulates a response. This represents the initial recruitable margin prior to this treatment. This represents the available fundraising margin at the current moment; Synthetic instantaneous fatigue load : ; In the formula, , , , , The weighting coefficients for fatigue sub-indicators are set by default. Repeated response decay and plateau decay have the highest weights (both are most sensitive to fatigue and the most stable to measure). Construct a continuous-time first-order model: ; In the formula, This is the fatigue accumulation rate coefficient. The stimulation switch is in state (1 = stimulating). Let be the individual recovery time constant, and take . ( (This is the macro-to-micro recovery scale conversion factor, default 20), updated with each round of exploration; Calculate fatigue recovery debt : ; In the formula, For discrete iteration time steps, for Moment .

[0053] when Stimulation is prohibited when the threshold is above a high threshold and permitted when the threshold is below a low threshold, maintaining the previous state in between; the remaining rest time is predicted by debt dynamics analysis, providing a countdown prompt; the specific process includes using a dual-threshold hysteresis loop for stimulus permission to avoid frequent on / off cycles near the threshold: ; Predicting remaining rest time using analytical solutions of debt dynamics (for interface countdown and compliance tips): ; Upper limit of stimulation duration per group (to prevent excessively long stimulation sessions) (Not exceeding the limit): ; , Stop and allow thresholds, hysteresis width 0.4; Output: Predicted remaining rest time (s); Absolute duration limit (take) ), safety net; The average fatigue load of the last three groups; Output: The longest allowed stimulus duration for this group; : Stimulating enable site.

[0054] S3 includes S3.3, which extracts burst-type functional reserves through rapid short contraction and sustained plateau contraction, respectively. and endurance-type functional reserves Calculate the bias index The specific process includes the system guiding the patient to complete two standard tasks in sequence, and extracting features respectively (without making any declaration of type I / II muscle fiber attribution): Task A (rapid short contraction, ): ; ; Task B (Continuous platform contraction) ): ; ; Bias Index: ; The starting time is given by the TKEO criterion; Intermediate variable, rate of increase in electromyography (EMG) ); Peak time (s); Intermediate variable, median frequency, derived from sEMG power spectrum. according to This is obtained only as an empirical indicator of the downward shift of the spectrum during fatigue, and is not used to infer muscle fiber type; The percentile normalization operator for the population reference distribution outputs... ; Weighting index, default values ​​for each ; Output, burst and endurance-type functional reserves ( ); : Output, bias exponent. More explosive power than endurance; on the contrary.

[0055] From the archive queue Extracting from each treatment session Perform a weighted linear regression with a slope of The residual determination platform is used for calculation; the specific process includes, From the archive queue Take the same detection conditions (same) Sequence, same Pre-treatment margin sequence Perform weighted least squares regression: ; Synchronous calculation residual judgment platform: ; Treatment records; Treatment sequence number; Number of times completed ( Only output ); Sample weights, representing the number of valid points detected in this probe. This automatically reduces the weight of low-quality detections; Output, plasticity coefficient ( (Unit / time), the physical meaning is the rate at which the stimulus-muscle transmission relationship improves under the same detection intensity; Significance threshold, which is twice the standard error of the slope, to ensure that the judgment is not driven by noise; Output: Plasticity status label.

[0056] Constructing state vectors : ; In the formula, For the treatment of plasticity coefficient; The four-level arbitration generation control command is used, with the first level being secure arbitration. ; In the formula, For impedance, For the contact self-test to pass, High threshold for fatigue debt This is the lower limit of the saturated remaining space. ; The second level is mode selection: ; In the formula, It is a timing compensation mode. To facilitate the fundraising model, Endurance maintenance mode It is a self-guided mode. For the variant pattern, As a platform, For a plasticity status label, The bias determination threshold is... ; The third level is the dose continuity law: ; ; ; In the formula, The command intensity after dose-rate calculation. As a reference for dose step size, As the plasticity direction factor, This is the upper limit coefficient of the treatment intensity relative to the exercise threshold. , This represents the upper limit of the patient's subjective tolerance. The fourth level is the timing of the stimulus: ; ; In the formula, To stimulate the timing of output; Build the total output : .

[0057] At the end of treatment, the average intensity and efficiency of each mode are archived. The mode with the highest efficiency is selected as the priority mode for the next treatment. The starting intensity for the next treatment is finely adjusted based on the current average in the direction of plasticity. At the same time, three trend curves are output: autonomous temporal improvement, plasticity coefficient, and autonomous recruitment efficiency. The specific process includes: at the end of treatment, archiving and calculating the suggested initial values ​​for the next treatment. ; ; : Maximum adjustment range between treatment courses (5%), limiting single jumps; : Intermediate variable, the margin improvement produced per unit stimulation time, i.e., the dose efficiency of the model; Output: The initial intensity and priority mode for the next treatment, written to the patient's file; the report is also output. , , Three trend curves.

[0058] 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 the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some or all of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present invention.

Claims

1. A closed-loop magnetic stimulation method for pelvic floor recruitment margin and defense sequence, characterized in that, include: S1. Acquire pelvic floor electromyography signals using a vaginal surface electromyography probe, perform gated sampling based on the magnetic stimulation discharge time, filter and separate artifacts; When the pelvic floor muscles are in a resting state and the fatigue recovery debt index is below the upper limit threshold of fatigue debt, a standardized probe magnetic pulse with an increasing gradient is injected, the induced electromyographic incremental response of each pulse is measured, the recruitment curve is fitted, and the minimum recruitment threshold, recruitment gain, saturation residual space and recovery time constant are extracted to synthesize the recruitment margin. S2. Simultaneously acquire abdominal surface electromyography, abdominal belt pressure, and trunk triaxial acceleration signals, fuse them to construct an abdominal pressure proxy signal, detect sudden rise in abdominal pressure events and trace their onset time, identify precursor waveforms and calculate precursor confidence; detect the pelvic floor muscle recruitment initiation time, and define the defense timing gap by the difference between the abdominal pressure initiation time and the recruitment initiation time; when the precursor confidence exceeds the threshold and the defense timing gap shows hysteresis, advance the stimulation trigger time to the total system delay, so that the effective stimulation time is before the abdominal pressure rise edge, and determine the compensatory stimulation intensity based on the timing gap and recruitment margin; S3. Superimpose detection pulses on the platform, measure the relationship between autonomous recruitment and the incremental value after superposition, and construct an autonomous-induced coupling index; collect multiple indicators to synthesize instantaneous fatigue load, establish a fatigue debt accumulation and recovery model, and gating stimulation with a debt threshold; extract burst-type and endurance-type functional reserves through rapid and sustained contraction tasks and calculate the bias index; use the regression slope of the recruitable margin before each treatment course as the treatment plasticity coefficient; construct a state vector from the above quantities, and generate control commands through a four-level arbitration of safety veto, mode selection, dose calculation, and timing determination to control the magnetic stimulation output.

2. The pelvic floor recruitment margin and defense time-series closed-loop magnetic stimulation method according to claim 1, characterized in that, S1 includes, S1.1, using a dual-channel probe inside the vagina to achieve a sampling rate. Acquire raw signals Constructing the gate function Screening for effective raw electromyography sequences: ； In the formula, For a moment, For the first The discharge moment of the secondary magnetic stimulation. For magnetic stimulation index, To determine the duration of the blind spot after stimulation, This serves as a protective interval before the next stimulus. The effective acquisition sequence after gating is: ； In the formula, This is the effective raw electromyographic sequence after gating. For discrete sample numbers; S1 includes, S1.2, using a combination of bandpass filtering and power frequency notch filtering for... By performing frequency band limiting and interference suppression, the purified electromyography sequence was obtained. ; Adaptive template cancellation processing The residual artifacts from stimulation are subtracted by scaling the pure artifact template to obtain the pure evoked electromyographic waveform. ; Calculate the sliding window RMS envelope and first-order difference slope, and use the resting baseline threshold to determine the four states: ； In the formula, For four-phase labels, In a resting state, It is in a contraction-ascending state. As a platform, It is in a relaxed state. The current electromyographic envelope amplitude. Let the first-order difference of the envelope be the current moment. For absolute values, The resting baseline threshold, This is the slope threshold.

3. The pelvic floor recruitment margin and defense time-series closed-loop magnetic stimulation method according to claim 2, characterized in that, S1 includes, S1.3, constructing the detection trigger conditions. : ； In the formula, For indicator functions, The duration of the current resting phase is as follows: The resting place determination threshold, To improve the debt recovery index, The upper limit threshold for fatigue debt; Inject when it is 1; Generate a sequence of detection intensities that increase in gradient from low to high: ； In the formula, For the first The set value of magnetic stimulation intensity for the second detection. The initial detection intensity, For intensity step increments, For the index of the number of probe points, Number of detection points; Set the safety clamp: ； In the formula, The upper limit coefficient is the threshold amplitude coefficient. The motion threshold, This represents the upper limit of the patient's subjective tolerance. Set the interval between adjacent detection pulses : ； In the formula, To recover the estimated time constant; The isolated pure evoked waveform is rectified and integrated within the physiological latency window, and the pre-probe baseline is subtracted to obtain the incremental response. ; Check for shrinkage contamination within the detection window; if contamination is found, the current measurement is invalidated and retested; finally, output a validity flag. ; S1 includes S1.4, and the effective pairing set. Fitting the recruitment curve : ； In the formula, For saturation-induced amplitude, For the intensity of magnetic stimulation, For half the fundraising intensity, It is an exponential function. This is the curve steepness parameter; Three fingerprint components are derived from the fitted parameters, including the minimum recruitment threshold. , fundraising gains and saturated remaining space : ； ； ； In the formula, The intensity of the current therapeutic stimulus; The fourth component is obtained by single-exponential fitting of the relaxation segment of the post-stimulation envelope: ； In the formula, Let the envelope change over time as a function of stimulation. This represents the residual electromyographic amplitude after contraction. For the recovery time constant, This represents the resting envelope baseline value. S1 includes, S1.5, normalizing the three fingerprint components using reference values ​​and then synthesizing the original margin using geometric weighting. Then, a Bayesian posterior fusion is performed with the prior mean of the population to obtain the final recruitable margin. ; Multiple rounds of probing results within the same treatment session were smoothed using exponential forgetting recursive smoothing; the first and last probing results of each treatment, along with the treatment intensity, were entered into the treatment record cohort. .

4. The pelvic floor recruitment margin and defense time-series closed-loop magnetic stimulation method according to claim 3, characterized in that, S2 includes S2.1, which adds three sensors—abdominal electromyography, abdominal belt pressure, and trunk acceleration—to the pelvic floor electromyography system. All of these sensors are driven by the same clock domain acquisition board, and each channel is pre-aligned according to the inherent delay calibrated at the factory. After rectification and filtering, the abdominal electromyography envelope, pressure signal, and acceleration modulus are extracted from each channel. The resting segment is normalized separately, and then the signals are weighted and fused according to fixed weights to form a single abdominal pressure proxy signal. When a channel fails, the weights are reallocated. Employing cumulative and change point detection algorithms, real-time monitoring is performed. The rising edge of the alarm signal, after which the accumulated data and the last zero point are traced back to determine the starting moment of the sudden increase in abdominal pressure. Simultaneously, a slope threshold is set to exclude drift values ​​below the slope threshold. S2 includes S2.2, data collected by the system during the calibration phase. The next spontaneous / involuntary cough, take The average waveform of the segment is used as a precursor template: ； In the formula, For length Precursor templates To determine the number of cough events collected during the calibration phase, For the first Abdominal pressure proxy signal of a coughing event For the first The moment when the abdominal pressure suddenly increased during this event. The length of the precursor template; Real-time matched filtering: ； In the formula, For shape matching degree, This represents the mean value within the current analysis window. This represents the mean of the precursor template; Calculate the log-likelihood ratio With the confidence level of the omen : ； ； In the formula, Assuming an impending abdominal pressure event, Assuming no impending abdominal pressure event, for The first-order difference; The instantaneous energy operator was used to process pelvic floor electromyography (EMG), and the initiation transition was detected after short-window smoothing. Confirmation was only made if several consecutive points exceeded a threshold, thus obtaining the recruitment initiation time. .

5. The pelvic floor recruitment margin and defense time-series closed-loop magnetic stimulation method according to claim 4, characterized in that, S2 includes S2.3, the timing gap of a single event. for: ； In the formula, For the first The moment when pelvic floor muscle recruitment begins in this event; Multi-event robust summation defense against timing gaps for: ； ； In the formula, As a time-series consistency indicator, For a valid set of events, The upper quartile of the timing gap. The lower quartile of the timing gap; Constructing criteria for event validity: ； In the formula, For the first The maximum slope within the current analysis window for this event. The slope threshold for abdominal pressure events. For the first Peak value of pelvic floor electromyography envelope in the next event. The resting envelope standard deviation; Constructing a time-series state hierarchy : ； In the formula, In order to defend in advance, For synchronization, To defend against delay, This is a reference threshold for lead time.

6. The pelvic floor recruitment margin and defense time-series closed-loop magnetic stimulation method according to claim 5, characterized in that, S2 includes S2.4, where the timing of the compensation stimulus is determined by the total system delay preceding the time when the aura confidence exceeds the limit. ； ； ； ； In the formula, For algorithm detection and decision delay, Delay in preparing for stimulator charging and discharging. To delay the rise of the coil current, For electromechanical delay, The total system delay, To compensate for the timing of stimulus triggering, To trigger the confidence threshold, To compensate for the timing of the stimulus distribution, For the effective moment of stimulation; Constructing advance constraints: ； ； In the formula, This is the expected lead time from the onset of abdominal pressure. This represents the actual lead time for defense. To predict the onset time of abdominal pressure; like No compensation will be issued this time; Calculate the magnitude of the compensatory stimulus: ； ； In the formula, For time series intensity conversion factors, To compensate for the intensity of the stimulus; Compensation pulse train for: ； In the formula, To compensate for the frequency of the pulse train, To compensate for the number of pulses in the pulse train, To compensate for the duration of the pulse train; S2 includes S2.5, which uses a historical event annotation set to define the asymmetric costs of missed detections and false triggers, and searches for the optimal threshold under a hard constraint on the false trigger rate. Alarm threshold in cumulative and change point detection algorithms And adjust online after each treatment; Introducing compensation gain The actual distribution intensity is: ； Evaluate the timing improvement based on event blocks, if and If the target is met, then decrease. If it worsens, it increases; at the end of each block, an uncompensated assessment event is inserted to measure the timing gap without assistance and confirm the improvement of autonomous capabilities.

7. The pelvic floor recruitment margin and defense time-series closed-loop magnetic stimulation method according to claim 6, characterized in that, S3 includes S3.1, where a standard probe pulse of fixed intensity is superimposed on the voluntary contraction plateau phase to measure the incremental amount of external stimulation that can be recruited after voluntary exertion. and with the level of spontaneous contraction contrast: ； ； ； In the formula, For autonomously induced coupling index, To prevent division by zero, To be non-irritating To maintain a fixed detection intensity; by and Patients were classified into four types, including those with autonomous recruitment disorder. Low fundraising reserves Near saturation Well-functioning : ； In the formula, and This is the partition threshold. Subscription status classification tags.

8. The pelvic floor recruitment margin and defense time-series closed-loop magnetic stimulation method according to claim 7, characterized in that, S3 includes S3.2, and the calculation platform attenuation rate. Residual ratio after shrinkage Recovery time ratio Repetitive response decay and detection response attenuation : ； ； ； ； ； In the formula, For platform envelope function, The maximum spontaneous contraction envelope peak value, To restore the time constant reference value, For the first The same intensity of treatment stimulates a response. This represents the initial recruitable margin prior to this treatment. This represents the available fundraising margin at the current moment; Synthetic instantaneous fatigue load : ； In the formula, , , , , The weighting coefficients for fatigue sub-indicators; Construct a continuous-time first-order model: ； In the formula, This is the fatigue accumulation rate coefficient. To stimulate the switch state, For individual recovery time constant; Calculate fatigue recovery debt : ； In the formula, For discrete iteration time steps, for Moment ; when Stimulation is prohibited when the threshold is above the high threshold, and allowed when the threshold is below the low threshold, maintaining the previous state in between; the remaining rest time is predicted by debt dynamics analysis and a countdown prompt is given.

9. The pelvic floor recruitment margin and defense time-series closed-loop magnetic stimulation method according to claim 8, characterized in that, S3 includes S3.3, which extracts burst-type functional reserves through rapid short contraction and sustained plateau contraction, respectively. and endurance-type functional reserves Calculate the bias index : ； In the formula, To prepare for explosive functionalities, Reserved for endurance-type functions; From the archive queue Extracting from each treatment session Perform a weighted linear regression with a slope of ; Residual determination platform; Constructing state vectors : ； In the formula, For the treatment of plasticity coefficient; The four-level arbitration generation control command is used, with the first level being secure arbitration. ； In the formula, For impedance, For the contact self-test to pass, High threshold for fatigue debt This represents the lower limit of the saturated remaining space. The second level is mode selection: ； In the formula, It is a timing compensation mode. To facilitate the fundraising model, Endurance maintenance mode, It is a self-guided mode. For the variant pattern, As a platform, For a plasticity status label, This is the threshold for bias determination; The third level is the dose continuity law: ； ； ； In the formula, The command intensity after dose-rate calculation. As a reference for dose step size, As the plasticity direction factor, This is the upper limit coefficient of the treatment intensity relative to the exercise threshold. This represents the upper limit of the patient's subjective tolerance. To ensure a rapid response, As a platform, For degeneration; The fourth level is the timing of the stimulus: ； ； In the formula, To stimulate output timing, Enable bit for secure arbitration, For the currently selected stimulation mode, It is in a relaxed state; At the end of treatment, archive and calculate the recommended initial values ​​for the next treatment: ； ； ； In the formula, This is the treatment sequence number. This is the upper limit coefficient for adjustment between treatment courses. The threshold for plasticity significance. For the dose efficiency of each mode, This is the pooling margin available for the final treatment. For this treatment, It is a symbolic function; At the end of treatment, the average intensity and efficiency of each mode are archived. The mode with the highest efficiency is selected as the priority mode for the next treatment. The starting intensity of the next treatment is slightly adjusted according to the plasticity direction based on the current average. At the same time, three trend curves are output: autonomous temporal improvement, plasticity coefficient, and autonomous recruitment efficiency.

10. A closed-loop magnetic stimulation system for pelvic floor recruitment margin and defense timing, characterized in that, include: A dual-channel surface electromyography probe is installed inside the vagina to collect pelvic floor electromyography signals; Pelvic floor magnetic stimulation coil and its driving circuit are used to output magnetic stimulation pulses; An abdominal pressure event sensing component consisting of abdominal surface electromyography electrodes, an abdominal belt-type pressure sensor, and a trunk triaxial accelerometer; A gated acquisition board that shares the clock domain with the stimulus timing generator; The real-time processing and control host is used to perform probe pulse injection and recruitable margin synthesis, abdominal pressure proxy signal construction and defense timing gap calculation and compensation stimulus triggering, six-dimensional state vector construction and four-level arbitration control command generation; The safety interlocking unit includes a handheld stop switch, electrode contact impedance self-test, and dual-sided output strength clamping. Human-computer interaction unit; each unit is connected via a data bus and its operation is coordinated by the host.

Images