Ankle pump-pelvic floor muscle cooperative activation rehabilitation system based on pressure feedback closed loop

Abstract

The application discloses a kind of based on pressure feedback closed loop's ankle pump-pelvic floor muscle synergic activation rehabilitation system, belong to rehabilitation medical technical field, including signal acquisition unit, and the time sequence signal and multichannel electromyogram of collection foot pressure;Synergic control unit includes: phase prediction submodule, for estimating ankle pump movement phase, calculate the predicted peak time of reaching next dorsiflexion peak value;Fatigue estimation submodule is used to extract median frequency, root-mean-square amplitude and peak contraction time Multidimensional features, weighted fusion calculation comprehensive fatigue index;Parameter optimization decision submodule is used to output electric stimulation instruction, optimal stimulation parameter combination and training adjustment instruction;Rehabilitation execution unit includes the pneumatic ankle pump driving device for driving ankle pump movement and the electromyographic stimulation device for applying electric stimulation to pelvic floor muscle.The present application guarantees the accurate synchronization of ankle pump movement and pelvic floor muscle electric stimulation by advance trigger, and guarantees the comfort of patient use by fatigue monitoring.

Description

Technical Field

[0001] This invention relates to the field of rehabilitation medicine technology, and in particular to an ankle pump-pelvic floor muscle synergistic activation rehabilitation system based on a pressure feedback closed loop. Background Technology

[0002] Ankle pump exercises and low-frequency pulsed electrical stimulation (LPS) of the pelvic floor muscles are both well-established clinical rehabilitation methods. Ankle pump exercises drive calf muscle contraction through periodic dorsiflexion (pointing the toes) and plantarflexion (pointing the feet) movements of the ankle joint, accelerating venous blood return in the lower limbs. This is a standard rehabilitation method for preventing deep vein thrombosis and improving lower limb circulation. Low-frequency pulsed electrical stimulation of the pelvic floor muscles applies low-frequency pulsed current (usually 8-80Hz) to the pelvic floor muscle area through external electrodes, inducing passive rhythmic contraction of the pelvic floor muscles. It is used for postpartum pelvic floor function repair and is a first-line treatment for clinical pelvic floor rehabilitation.

[0003] In existing technologies, some rehabilitation devices attempt to combine the above two methods, mainly in the form of: using plantar or dorsolateral pressure sensors to detect the pressure peak of ankle pump movement; triggering the electrical stimulation module when the pressure reaches a preset fixed threshold, and stopping when the pressure falls below the threshold, thereby establishing a basic linkage between ankle pump movement and electrical stimulation. However, the above-mentioned existing combined scheme is essentially a single switch logic triggered by a fixed threshold, which has the following technical defects:

[0004] First, there is an inherent time lag and low synchronization accuracy. Existing solutions employ a passive response logic of "pressure → threshold → initiation stimulus." From the time the pressure signal reaches the threshold to the effective contraction of the pelvic floor muscles, there is a signal acquisition delay of approximately 10-30 ms, a controller processing delay of approximately 5-20 ms, and a neuromuscular conduction delay of approximately 50-150 ms, for a total delay of 65-200 ms. A standard ankle pump movement cycle is only 1.5-3 seconds. The aforementioned delays account for 4%-13% of the single movement window, resulting in a significant time discrepancy between the peak pelvic floor muscle contraction and the peak ankle pump dorsiflexion, making true neuromuscular synchronous coupling impossible.

[0005] Second, the lack of muscle fatigue recognition poses a risk of overtraining. Current methods cannot identify the fatigue state of the pelvic floor muscles, resulting in continuous training at a fixed frequency and duration. When the pelvic floor muscles are fatigued (usually after 5-15 minutes of continuous training), continuing to apply the same intensity of stimulation not only drastically reduces activation efficiency (effective activation rate can decrease by 40%-60% in a fatigued state), but may also cause adverse reactions such as muscle spasms and increased post-training soreness, affecting patient compliance. Summary of the Invention

[0006] The purpose of this invention is to overcome the shortcomings of the prior art, such as low timing synchronization accuracy and inability to identify muscle fatigue, and to provide an ankle pump-pelvic floor muscle synergistic activation rehabilitation system based on pressure feedback closed loop, which has the advantages of precise synchronization between ankle pump movement and pelvic floor muscle electrical stimulation and fatigue adaptive regulation.

[0007] An ankle pump-pelvic floor muscle synergistic activation rehabilitation system based on pressure feedback closed loop includes:

[0008] The signal acquisition unit is used to simultaneously acquire the foot pressure timing signal during ankle pump movement and the multi-channel electromyographic signal during electromyographic stimulation.

[0009] The collaborative control unit, which is communicatively connected to the signal acquisition unit, includes:

[0010] The phase prediction submodule is used to estimate the ankle pump phase based on the temporal characteristics of the foot pressure signal and calculate the predicted peak time to reach the next dorsiflexion peak.

[0011] The fatigue estimation submodule is used to extract multidimensional features of median frequency, root mean square amplitude and peak contraction time from the multi-channel electromyography signal, and to calculate the comprehensive fatigue index by weighted fusion.

[0012] The parameter optimization decision submodule is used to output electrical stimulation instructions, optimal stimulation parameter combinations, and training adjustment instructions based on the predicted peak time and the comprehensive fatigue index.

[0013] The rehabilitation execution unit, which is communicatively connected to the collaborative control unit, includes a pneumatic ankle pump drive device for driving ankle pump movement and an electromyographic stimulation device for applying electrical stimulation to the pelvic floor muscles.

[0014] Preferably, the collaborative control unit further includes a compensation identification submodule for acquiring multi-channel electromyographic signals, including the pelvic floor muscle reference electromyographic signals, abdominal reference electromyographic signals, and thigh adductor muscle reference electromyographic signals. The activation timing and amplitude ratio characteristics of the multi-channel electromyographic signals are used as comparison parameters to identify the compensation force exertion mode by comparing the similarity with a preset template.

[0015] The preset templates include normal activation templates and compensation mode templates. The compensation mode templates include at least abdominal muscle priority activation compensation, thigh muscle synchronous activation compensation, and pelvic floor muscle ineffective activation compensation.

[0016] Among them, similarity comparison and recognition calculates the similarity distance between real-time multi-channel electromyography signals and normal activation templates and various compensation mode templates through dynamic time warping algorithm, and determines the activation mode category of the current cycle based on the principle of minimum distance.

[0017] The identification results are output to the parameter optimization decision submodule.

[0018] Preferably, the compensation identification submodule further includes a correction feedback module, which is configured as follows:

[0019] A compensation detection module is set up to detect the number of times the compensation mode is triggered.

[0020] When a single compensatory exertion is detected, the first corrective instruction is output, including generating a corresponding voice prompt or tactile feedback, and the current contraction cycle is not counted as an effective rehabilitation training session.

[0021] When the number of consecutive compensations exceeds the preset first threshold, a second correction instruction is output, including reducing the ankle pump movement speed and increasing the pre-trigger amount, in order to prolong the time window for the patient to perceive pelvic floor muscle contraction.

[0022] When the number of consecutive compensations exceeds the preset second threshold, a third correction instruction is output, including pausing ankle pump movement and outputting low-intensity electrical stimulation to guide the patient to reconstruct the correct force exertion pattern.

[0023] The second threshold is greater than the first threshold.

[0024] Preferably, the phase prediction submodule includes:

[0025] The sliding window module is used to collect the sliding time window of the most recent N complete ankle pump exercise cycles and extract the dorsiflexion peak time sequence and cycle duration sequence of each cycle;

[0026] The weighted estimation module is used to calculate the weighted average period duration of the period duration series with time-increasing weights.

[0027] The phase estimation module is used to estimate the current motion phase in real time and predict the peak time of the next backbend based on the weighted average period duration and the previous backbend peak time.

[0028] Preferably, the parameter optimization decision submodule calculates the advanced triggering time based on the predicted peak time and the individualized neuromuscular conduction delay, and sends an electrical stimulation triggering command to the electromyography stimulation device to align the peak time of pelvic floor muscle contraction with the peak time of ankle pump dorsiflexion.

[0029] The advanced triggering time is calculated using the following formula:

[0030] ;

[0031] This is the predicted peak time for the next dorsiflexion peak. For individualized neuromuscular conduction delay, Fixed system processing delay;

[0032] The individualized neuromuscular conduction delay is obtained through a delay calibration module, which is configured to: emit test electrical stimulation pulses at fixed times, simultaneously acquire multi-channel electromyographic signals of the pelvic floor muscles, calculate the time interval from the emission of stimulation to the electromyographic amplitude exceeding a preset multiple of the resting baseline, and repeat the process multiple times to obtain the median.

[0033] Preferably, the parameter optimization decision submodule further includes detecting the actual peak contraction time of the multi-channel electromyographic signal of the pelvic floor muscles after electrical stimulation triggering, and calculating the timing synchronization error with the measured dorsiflexion peak.

[0034] If the absolute value of the synchronization error exceeds the preset threshold, the individualized neuromuscular transmission delay is adaptively corrected according to the preset learning rate, so that the corrected delay converges iteratively in subsequent cycles.

[0035] Preferably, the parameter optimization decision submodule outputs training adjustment instructions based on the comprehensive fatigue index, the method including:

[0036] A preset fatigue grading threshold is set. Based on the comparison between the comprehensive fatigue index and the fatigue grading threshold, corresponding training adjustment instructions are output:

[0037] When the overall fatigue index is at a normal level, the output training adjustment instruction is to maintain or increase the intensity of rehabilitation training.

[0038] When the overall fatigue index is at the mild to moderate fatigue level, the output training adjustment instruction is to prolong the rest time between contraction and relaxation and / or reduce the intensity of stimulation.

[0039] When the overall fatigue index is at the severe fatigue level, the output training adjustment instruction is to suspend rehabilitation training and enter a forced rest period. During the rest period, the overall fatigue index is continuously monitored, and rehabilitation training is automatically resumed when it drops back to the normal level.

[0040] Preferably, the method by which the parameter optimization decision submodule outputs the optimal combination of stimulus parameters is to set up an individual modeling submodule, which includes:

[0041] The sparse sampling module is used to acquire the combination of parameters consisting of stimulation frequency, stimulation intensity and pulse width during the rehabilitation training process, and to record the amplitude of the electromyographic response on the pelvic floor muscle surface under different parameter combinations.

[0042] The response surface fitting module is used to fit an individualized stimulus response function with stimulus frequency, stimulus intensity, and pulse width as independent variables and the amplitude of electromyographic response on the pelvic floor muscle surface as the dependent variable by using polynomial response surface regression that includes independent variable cross terms and quadratic terms.

[0043] Preferably, the individual modeling submodule further includes an online update module, which is configured as follows:

[0044] During each rehabilitation training session, additional sampling was performed on parameter combinations with high prediction uncertainty of the individualized stimulus response function, and the measured data were added to the historical dataset.

[0045] The historical dataset was re-regressed using time decay weights to update the individualized stimulus-response function. The time decay weights were set such that the weight of recent measured data was greater than that of distant data.

[0046] Preferably, the parameter optimization decision submodule further includes a priority arbitration module, used to set the priority of training adjustment instructions to be higher than that of electrical stimulation instructions.

[0047] Compared with the prior art, the present invention has the following beneficial effects:

[0048] 1. This invention estimates the ankle pump motion phase online and predicts the next dorsiflexion peak time through the phase prediction submodule. Combined with the individualized neuromuscular conduction delay, it calculates the advanced trigger time and advances the pelvic floor muscle electrical stimulation trigger time to before the predicted peak arrives, so that the pelvic floor muscle contraction peak and the ankle pump dorsiflexion peak are precisely aligned.

[0049] 2. This invention also proposes real-time fatigue state identification and adaptive protection of training rhythm. The fatigue estimation submodule calculates a comprehensive fatigue index based on the fusion of multi-dimensional features such as median frequency, root mean square amplitude, and peak contraction time. The parameter optimization decision submodule automatically adjusts the training rhythm according to multi-level fatigue grading thresholds, and implements strategies such as maintaining progressive training, extending rest to reduce intensity, and pausing forced rest under normal, mild to moderate, and severe fatigue states, respectively, to eliminate the risk of overtraining.

[0050] 3. This invention uses a compensation identification submodule to identify compensation patterns such as preferential activation of abdominal muscles, synchronous activation of thigh muscles, and ineffective activation of pelvic floor muscles in real time by comparing the similarity between the activation timing and amplitude ratio characteristics of multi-channel electromyography signals and a dynamic time warping algorithm with a preset template through a similarity comparison. It also triggers graded correction feedback to ensure the effectiveness of training. Attached Figure Description

[0051] Figure 1 This is a schematic diagram of the ankle pump-pelvic floor muscle synergistic activation rehabilitation system based on pressure feedback closed loop according to the present invention. Detailed Implementation

[0052] To make the technical means, creative features, objectives and effects of this invention easier to understand, the invention will be further described below in conjunction with specific embodiments.

[0053] like Figure 1 As shown, an ankle pump-pelvic floor muscle synergistic activation rehabilitation system based on pressure feedback closed loop is presented.

[0054] The signal acquisition unit is used to simultaneously acquire the foot pressure timing signal during ankle pump movement and the multi-channel electromyographic signal during electromyographic stimulation.

[0055] The collaborative control unit, which is communicatively connected to the signal acquisition unit, includes:

[0056] The phase prediction submodule is used to estimate the ankle pump phase based on the temporal characteristics of the foot pressure signal and calculate the predicted peak time to reach the next dorsiflexion peak.

[0057] The fatigue estimation submodule is used to extract multidimensional features of median frequency, root mean square amplitude and peak contraction time from the multi-channel electromyography signal, and to calculate the comprehensive fatigue index by weighted fusion.

[0058] The parameter optimization decision submodule is used to output electrical stimulation instructions, optimal stimulation parameter combinations, and training adjustment instructions based on the predicted peak time and the comprehensive fatigue index.

[0059] The rehabilitation execution unit, which is communicatively connected to the collaborative control unit, includes a pneumatic ankle pump drive device for driving ankle pump movement and an electromyographic stimulation device for applying electrical stimulation to the pelvic floor muscles.

[0060] The three functional levels—signal acquisition unit, collaborative control unit, and rehabilitation execution unit—are described below:

[0061] The signal acquisition unit is responsible for synchronously acquiring multi-channel physiological signals. The acquisition is based on deployed sensors, including:

[0062] A flexible pressure sensor for the dorsum of the foot, located on the inner side of the dorsum of the foot air bladder, is used to collect the timing signal of foot pressure during ankle pump movements. A flexible pressure sensor for the plantar foot, located on the inner side of the plantar air bladder, is used for auxiliary verification of movement amplitude. A main acquisition electrode for the pelvic floor muscles, located above the pubic symphysis, is used for pelvic floor muscle activation detection and fatigue estimation. An abdominal muscle reference electrode, located 3 cm below the navel next to the linea alba, is used for abdominal muscle compensation identification. A reference electrode for the adductor muscles of the thigh, located in the middle of the inner thigh, is used for thigh muscle compensation identification. The above six channels of signals are amplified and filtered by a signal conditioning circuit, and then transmitted to the collaborative control unit through an analog-to-digital converter at a sampling rate of not less than 200Hz.

[0063] The collaborative control unit, serving as the core computing and control hub of the system, is deployed in an embedded processor or edge computing platform. It includes a phase prediction submodule, a fatigue estimation submodule, a compensation identification submodule, an individual modeling submodule, and a parameter optimization decision-making submodule. All submodules share a unified time base and sensor data, working collaboratively in parallel.

[0064] The rehabilitation execution unit includes a pneumatic ankle pump drive and an electromyography (EMG) stimulation device. The pneumatic ankle pump drive receives training adjustment instructions from the parameter optimization decision-making submodule and controls the driving frequency and amplitude of the ankle pump movement by adjusting the opening of the pneumatic valve. The EMG stimulation device receives electrical stimulation trigger instructions and the optimal combination of stimulation parameters, applying low-frequency pulsed current to the pelvic floor muscle area. The pneumatic ankle pump drive uses a flexible airbag structure controlled by a pneumatic valve, driving the patient's ankle joint to perform periodic dorsiflexion-plantar flexion movements by adjusting the air intake flow and exhaust sequence. The EMG stimulation device uses a constant current output mode and also has open / short circuit detection and overcurrent protection functions to ensure safe use.

[0065] The various sub-modules under the collaborative control unit will now be described in detail:

[0066] The phase prediction submodule is used to solve the problem of delayed triggering of electrical stimulation in existing technologies. Its principle is to estimate the ankle pump motion phase online based on the temporal characteristics of foot pressure signals and calculate the predicted peak time to reach the next dorsiflexion peak, thereby achieving advanced triggering.

[0067] The phase prediction submodule includes a sliding window module, a weighted estimation module, and a phase inference module;

[0068] The phase prediction submodule includes:

[0069] The sliding window module is used to acquire sliding time windows of the most recent N complete ankle pump exercise cycles, extracting the dorsiflexion peak time sequence and cycle duration sequence for each cycle; the value of N is preferably 3 to 5 cycles, and for each complete cycle, the module extracts the dorsiflexion peak time sequence. and periodic duration series ,in ;

[0070] The weighted estimation module is used to calculate the weighted average period duration of the period duration series using time-series increasing weights. Since the time-series increasing weights increase with time, the latest period has the highest weight. Therefore, the formula for calculating the weighted average duration is as follows:

[0071] ;

[0072] in, For the first The time-increasing weights of each period duration satisfy the following conditions: ,and ;

[0073] The phase estimation module is used to estimate the current motion phase in real time and predict the peak time of the next backbend based on the weighted average period duration and the previous backbend peak time. The current time is set as... The previous peak time of dorsiflexion was Estimate the current motion phase using the following formula. :

[0074] ;

[0075] The formula for predicting the peak time is as follows:

[0076] .

[0077] To achieve proactive triggering, the system needs to pre-calibrate individualized neuromuscular conduction delay. The delay calibration module is integrated into the parameter optimization decision submodule. Since the parameter optimization decision submodule is responsible for making decisions, the phase prediction submodule mainly outputs the current motion phase. The peak time is predicted, but the advanced triggering is still achieved by the electrical stimulation command output by the parameter optimization decision submodule. The parameter optimization decision submodule calculates the advanced triggering time based on the predicted peak time and the individualized neuromuscular conduction delay, and sends an electrical stimulation triggering command to the electromyographic stimulation device to align the peak time of pelvic floor muscle contraction with the peak time of ankle pump dorsiflexion.

[0078] The advanced triggering time is calculated using the following formula:

[0079] ;

[0080] This is the predicted peak time for the next dorsiflexion peak. For individualized neuromuscular conduction delay, The system's fixed processing delay is determined by factory calibration, with a typical value of approximately 15ms.

[0081] The individualized neuromuscular conduction delay is obtained through a delay calibration module, which is configured to: emit test electrical stimulation pulses at fixed times, simultaneously acquire multi-channel electromyographic signals of the pelvic floor muscles, calculate the time interval from the emission of stimulation to the electromyographic amplitude exceeding a preset multiple of the resting baseline, and repeat the process multiple times to obtain the median.

[0082] The parameter optimization decision submodule also includes trigger effect verification and adaptive correction functions. The parameter optimization decision submodule also includes detecting the actual contraction peak time of the multi-channel electromyographic signal of the pelvic floor muscles after electrical stimulation triggering, and calculating the timing synchronization error between the measured dorsiflexion peak and the actual contraction peak.

[0083] If the absolute value of the synchronization error exceeds the preset threshold, the individualized neuromuscular transmission delay is adaptively corrected according to the preset learning rate, so that the corrected delay converges iteratively in subsequent cycles.

[0084] After stimulation, the system detects the actual peak contraction time of the multi-channel electromyographic signals of the pelvic floor muscles. Calculate the timing synchronization error: ,like Then, the individualized neuromuscular conduction delay is adjusted according to the preset learning rate:

[0085] ;

[0086] in, The learning rate is set to 0.1 by default. This correction takes effect in subsequent cycles, achieving adaptive convergence. The typical number of convergence cycles is 5 to 10 training cycles, and the synchronization error can be controlled within ±15ms.

[0087] The fatigue estimation submodule is used to assess the fatigue level of the pelvic floor muscles in real time to prevent overtraining. This module uses each electrical stimulation trigger as the time reference and extracts a 500ms window of multichannel electromyographic signals from the pelvic floor muscles for analysis. Three fatigue-sensitive features were extracted: median frequency. Root mean square amplitude and peak contraction time The comprehensive fatigue index is calculated by weighted fusion of multidimensional features extracted from the multichannel electromyography signals, including median frequency, root mean square amplitude, and peak contraction time.

[0088] Median frequency Through the The power spectral density is obtained by performing a short-time Fourier transform on the fragment. , Satisfies the condition of bisection of the power spectrum area:

[0089] ;

[0090] When muscles are fatigued, Typical decrease is 20% to 40%.

[0091] Root mean square amplitude The formula for calculation is:

[0092] ;

[0093] in, The number of sampling points. For the first Electromyography signal amplitude at each sampling point, in the early stage of fatigue It may increase compensatorily and decrease during deep fatigue.

[0094] Peak contraction time Defined as from stimulus trigger to The time interval of the envelope peak; neuromuscular conduction slows down during fatigue. extend.

[0095] Multidimensional feature vector construction: The average feature value of the first 3 contraction cycles after the start of training is taken as the multidimensional feature vector. The multidimensional feature vector is rebuilt at the start of each independent training session.

[0096] Comprehensive fatigue index The following is obtained by normalizing the deviation of each feature and fusing them according to a preset weight:

[0097] ;

[0098] ;

[0099] ;

[0100] ;

[0101] in, This represents the normalized decrease in median frequency relative to the reference. This is the normalized deviation of the root mean square amplitude from the reference. This is the normalized extension of the peak contraction time relative to the baseline; , , For the preset weighting coefficients, satisfy + + =1, default values ​​are respectively =0.5、 =0.3、 =0.2, the weight can be automatically optimized through individual historical data.

[0102] Similarly, the fatigue estimation submodule only participates in the calculation of the comprehensive fatigue index. The final decision is still made through the parameter optimization decision submodule, which outputs training adjustment instructions based on the comprehensive fatigue index. The method includes:

[0103] A preset fatigue grading threshold is set. Based on the comparison between the comprehensive fatigue index and the fatigue grading threshold, corresponding training adjustment instructions are output:

[0104] When the overall fatigue index is at a normal level, the output training adjustment instruction is to maintain or increase the intensity of rehabilitation training.

[0105] When the overall fatigue index is at the mild to moderate fatigue level, the output training adjustment instruction is to prolong the rest time between contraction and relaxation and / or reduce the intensity of stimulation.

[0106] When the overall fatigue index is at the severe fatigue level, the output training adjustment instruction is to suspend rehabilitation training and enter a forced rest period. During the rest period, the overall fatigue index is continuously monitored, and rehabilitation training is automatically resumed when it drops back to the normal level.

[0107] The collaborative control unit also includes a compensation identification submodule, which is used to acquire multi-channel electromyographic signals, including the reference electromyographic signals of the pelvic floor muscles, the reference electromyographic signals of the abdomen, and the reference electromyographic signals of the adductor muscles of the thigh. The activation timing and amplitude ratio characteristics of the multi-channel electromyographic signals are used as comparison parameters to identify the compensation force exertion mode by comparing the similarity with a preset template.

[0108] The preset templates include normal activation templates and compensation mode templates. The compensation mode templates include at least abdominal muscle priority activation compensation, thigh muscle synchronous activation compensation, and pelvic floor muscle ineffective activation compensation.

[0109] Among them, similarity comparison and recognition calculates the similarity distance between real-time multi-channel electromyography signals and normal activation templates and various compensation mode templates through dynamic time warping algorithm, and determines the activation mode category of the current cycle based on the principle of minimum distance.

[0110] The identification results are output to the parameter optimization decision submodule.

[0111] The compensation identification submodule is used to identify whether patients have compensatory force patterns in muscles such as abdominal muscles and adductor muscles during rehabilitation training, ensuring that the training targets the pelvic floor muscles.

[0112] The normal pelvic floor muscle activation characteristic template is used as template A, and the compensatory mode template is used as template B. The characteristics of template A include: exceeding the activation threshold within 50-200ms after electrical stimulation triggering; pelvic floor muscle / abdominal muscle amplitude ratio ≥2.0; abdominal muscle reference activation time is at least 50ms later than pelvic floor muscle reference activation time, or abdominal muscle amplitude is less than 1.5 times its baseline.

[0113] Compensation pattern template B includes at least three sub-patterns:

[0114] Compensation mode B1 is abdominal muscle compensation: the abdominal muscle reference position is activated earlier than or almost simultaneously with the pelvic floor muscle reference position, and the abdominal muscle / pelvic floor muscle amplitude ratio is greater than 1.5.

[0115] Compensation mode B2 is thigh compensation: the amplitude of the electromyographic signal at the thigh adductor reference position exceeds twice the baseline and is activated before the pelvic floor muscle reference position;

[0116] Compensation mode B3 is a whole-body compensation: the abdominal muscle reference site and the thigh adductor muscle reference site are activated simultaneously, while the amplitude of the electromyographic signal of the pelvic floor muscle reference site is 1.5 times lower than the baseline, indicating that the pelvic floor muscles are not actually effectively activated.

[0117] Similarity calculation is based on dynamic time warping The algorithm calculates the similarity distance between real-time multi-channel electromyography signals and each template:

[0118] Let the real-time signal characteristic sequence be... The template sequence is Similarity distance The calculation formula is as follows:

[0119] ;

[0120] in, To standardize the path, Indicates the real-time sequence number Point and template sequence number The correspondence between points This represents the Euclidean distance between corresponding points.

[0121] Pattern determination is based on the minimum distance principle to determine the activation pattern category of the current contraction cycle: calculating the similarity distance between real-time features and each sub-pattern of template A and template B. ,like If the value is the minimum, it is determined to be normal activation; otherwise, it is determined to be the corresponding compensatory mode.

[0122] The compensation identification submodule further includes a correction feedback module, which is configured as follows:

[0123] A compensation detection module is set up to detect the number of times the compensation mode is triggered.

[0124] When a single compensatory exertion is detected, the first corrective instruction is output, including generating a corresponding voice prompt or tactile feedback, and the current contraction cycle is not counted as an effective rehabilitation training session.

[0125] When the number of consecutive compensations exceeds the preset first threshold, a second correction instruction is output, including reducing the ankle pump movement speed and increasing the pre-trigger amount, in order to prolong the time window for the patient to perceive pelvic floor muscle contraction.

[0126] When the number of consecutive compensations exceeds the preset second threshold, a third correction instruction is output, including pausing ankle pump movement and outputting low-intensity electrical stimulation to guide the patient to reconstruct the correct force exertion pattern.

[0127] The second threshold is greater than the first threshold.

[0128] In addition, the system calculates the compensation rate after each training session: compensation rate = number of compensations / total number of contractions. When the compensation rate stabilizes, the system determines that the patient has established a correct exertion pattern and deactivates the automatic deceleration protection.

[0129] The method by which the parameter optimization decision submodule outputs the optimal combination of stimulus parameters is to set up an individual modeling submodule to achieve individualized adaptive adaptation of stimulus parameters. The individual modeling submodule includes:

[0130] The sparse sampling module is used to acquire parameter combinations consisting of stimulation frequency, stimulation intensity, and pulse width during rehabilitation training, and to record the amplitude of the electromyographic response on the pelvic floor muscles under different parameter combinations; defining the parameter space: stimulation frequency Here, four representative frequency points are defined, and the stimulus intensity is... Here, 6 intensity levels are defined, and the pulse width is... Here, three width levels are defined, resulting in a total of 4×6×3=72 parameter combinations.

[0131] To reduce calibration time, a Latin hypersquare sampling strategy was used to select 16 representative combinations from 72 combinations. Three pulse stimulations were applied to each selected combination, and the mean peak response amplitude of the electromyographic signal at the pelvic floor muscle reference position was recorded. A 10-second interval is maintained between adjacent combinations to prevent residual effects from interfering.

[0132] The response surface fitting module is used to fit an individualized stimulus response function using polynomial response surface regression, which includes independent variable cross terms and quadratic terms. The function uses stimulation frequency, stimulation intensity, and pulse width as independent variables and the amplitude of the electromyographic response on the pelvic floor muscle surface as the dependent variable. The expression of the individualized stimulus response function is as follows:

[0133] ;

[0134] in, To predict the response amplitude, For the combination of stimulus parameters, ~ The regression coefficient vector is obtained by estimation using the least squares method;

[0135] The parameter optimization decision submodule targets the activation amplitude. (Default setting is 2.5 times the baseline amplitude) to constrain and minimize stimulus intensity. To achieve the objective, constrained nonlinear programming is performed on the response surface to solve for the optimal parameter combination in the current training iteration. The optimization problem can be formally represented as:

[0136] ; , , .

[0137] The individual modeling submodule also includes an online update module, which is configured as follows:

[0138] During each rehabilitation training session, additional sampling was performed on parameter combinations with high prediction uncertainty of the individualized stimulus response function, and the measured data were added to the historical dataset.

[0139] The historical dataset was re-regressed using time decay weights to update the individualized stimulus-response function. The time decay weights were set such that the weight of recent measured data was greater than that of distant data.

[0140] The function expression for re-regressing the historical dataset using time decay weights is as follows:

[0141]

[0142] in, The time decay factor, This is the current training cycle number. For the first The data points are collected from training periods, with recent measured data having a greater weight than older data, and the maximum change before and after a model update. When the value is below the convergence threshold, the model is marked as stable, and the active sampling frequency can be reduced to save computational resources.

[0143] The rehabilitation process also needs to be tracked. Every 5 training sessions, the system calculates the average effective activation amplitude of the most recent 5 training sessions. If the average amplitude consistently exceeds the current... 120% automatically Increase by 10% to achieve progressive rehabilitation intensity; if the average amplitude remains below [a certain value], [further action will be taken]. If it reaches 80%, then maintain or reduce it. To prevent patient frustration, the system generates individualized rehabilitation progress reports and plots the optimal stimulus intensity. Curve showing change with the number of training sessions: A decrease indicates improved neuromuscular response of the pelvic floor muscles, suggesting effective rehabilitation. If the results remain unchanged or increase over a long period, it indicates that the training program needs to be adjusted.

[0144] The parameter optimization decision submodule also includes a priority arbitration module, which is used to set the priority of training adjustment instructions to be higher than that of electrical stimulation instructions, so as to ensure the safety of training.

[0145] The final parameter optimization decision submodule makes decisions in the following four aspects:

[0146] In terms of electrical stimulation trigger control, the parameter optimization decision submodule issues an electrical stimulation trigger command to align the peak contraction time of the pelvic floor muscles with the peak time of ankle pump dorsiflexion.

[0147] In terms of training rhythm regulation, the parameter optimization decision submodule outputs corresponding training adjustment instructions, including the proportion of rest phase extension, the proportion of stimulus intensity reduction, and forced rest trigger.

[0148] In terms of compensatory correction, the parameter optimization decision submodule triggers hierarchical correction feedback instructions, including voice prompts, ankle pump deceleration, pause drive, and entry into individual perception guidance mode.

[0149] Regarding stimulus parameter optimization, the parameter optimization decision submodule outputs the optimal combination of stimulus parameters for the current training iteration. It is also equipped with an electromyography (EMG) stimulation device.

[0150] The following example illustrates the entire system's operation flow in detail:

[0151] Patient scenario: Female, 28 years old, 6 weeks postpartum, experiencing mild pelvic floor muscle relaxation, poor lower limb circulation, and inability to actively engage her pelvic floor muscles. The procedure for using this system is as follows:

[0152] The initial calibration phase takes approximately 3-5 minutes.

[0153] After the system is powered on, each sensor channel performs a self-test, measures the baseline noise level, and the dorsolateral and plantar flexible pressure sensors perform zero-point calibration while the patient is stationary wearing foot covers. The system reads historical individualized parameters from non-volatile memory and enters the automatic calibration process.

[0154] Delay Calibration: The system delivered five single-pulse test stimuli via an electromyography (EMG) device. The stimulation signal frequency was 40 Hz, intensity was 5 mA, and pulse width was 200 μs. Simultaneously, EMG signals from the pelvic floor muscles at a reference position were recorded. The time interval from the emission of the stimulation pulse to the amplitude of the EMG signal at the pelvic floor muscles at the reference position exceeding three standard deviations from the resting baseline was measured. The five measurements were 102 ms, 88 ms, 95 ms, 108 ms, and 91 ms, respectively. The median was used to obtain the individualized neuromuscular conduction delay. = 95ms.

[0155] Parameter response calibration: The sparse sampling module performs Latin hypersquare sampling, selecting 16 representative combinations from 72 parameter combinations. Each combination is stimulated 3 times, and the peak response amplitude of the electromyographic signal at the reference position of the pelvic floor muscles is recorded. After fitting with second-order polynomial response surface regression, the individualized stimulation response function of the patient is obtained. The response surface shows that the patient has the best response amplitude to the parameter combination of stimulation frequency 40Hz, stimulation intensity 15mA, and pulse width 300μs. This combination is used as the initial optimal parameter combination.

[0156] Compensation baseline establishment: The patient was guided by voice to perform 5 standard pelvic floor muscle contraction attempts. The baseline vector of normal activation feature template A was collected. Analysis revealed that the patient had a significant tendency for abdominal muscle compensation. The activation time of the electromyographic signal of the abdominal muscle reference position was about 80ms earlier than that of the electromyographic signal of the pelvic floor muscle reference position. The amplitude ratio of the two was 2:3, which was determined to be compensation mode B1 (abdominal muscle compensation). This baseline information was recorded for compensation recognition comparison in subsequent training.

[0157] Fatigue baseline establishment: After the training officially begins, the electromyographic signals of the pelvic floor muscles at the reference position are automatically collected during the first 3 contraction cycles to establish a median frequency baseline of 78Hz, a root mean square amplitude baseline of 28μV, and a peak contraction time baseline of 142ms.

[0158] Formal training phase:

[0159] Training sessions 1-3 (approximately 10 minutes each): The system initiates electromyography (EMG) with initial optimal parameters. The pneumatic ankle pump drive operates at a frequency of 0.5 Hz. The phase prediction submodule initializes after acquiring three complete cycles and switches to advanced trigger mode. The trigger timing was calculated. The compensation recognition submodule frequently detected compensation mode B1, and the parameter optimization decision submodule continuously triggered the voice prompt "Pay attention to relaxing the abdomen," and automatically reduced the ankle pump drive frequency to 0.25Hz, while increasing the premature triggering amount by 20%. At the same time, the fatigue estimation submodule detected that a mild to moderate fatigue level was reached after 8-9 minutes of training, and the system automatically extended the rest time by 50%. During this stage, the compensation rate decreased from 62% in the first session to 35% in the third session.

[0160] Training sessions 4-8: The compensation rate stabilized below 20%, the system decelerated, the ankle pump frequency recovered to 0.5Hz, and the fatigue curve showed an average duration of 8 minutes. Based on this, the parameter optimization decision submodule automatically set the recommended training duration per session to 6.4 minutes, calculated using an 80% safety factor: 8 × 0.8 = 6.4. The individual modeling submodule updated and displayed the optimal stimulus intensity online. The decrease from an initial 15mA to 12mA indicates an improvement in the neuromuscular response of the pelvic floor muscles, suggesting effective rehabilitation.

[0161] 12th training session: Compensation rate drops to 8%, fatigue model stabilizes. The amplitude further decreased to 10mA, and the system generated a rehabilitation progress report: "Pelvic floor muscle neuromuscular function has significantly improved, and it is recommended to enter the intensity increase stage," automatically adjusting the target activation amplitude. Improve by 10%, and enter the next stage of rehabilitation.

[0162] 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 invention can be implemented in other specific forms without departing from its spirit or essential characteristics. Therefore, the embodiments should be considered in all respects as exemplary and non-limiting, and the scope of the invention is defined by the appended claims rather than the foregoing description. Thus, all variations falling within the meaning and scope of equivalents of the claims are intended to be included within the present invention. No reference numerals in the claims should be construed as limiting the scope of the claims.

[0163] Furthermore, it should be understood that although this specification describes embodiments, not every embodiment contains only one independent technical solution. This narrative style is merely for clarity. Those skilled in the art should consider the specification as a whole, and the technical solutions in each embodiment can also be appropriately combined to form other embodiments that can be understood by those skilled in the art.

Claims

1. A pressure feedback closed-loop-based ankle pump-pelvic floor muscle synergistic activation rehabilitation system, characterized in that, include: The signal acquisition unit is used to simultaneously acquire the foot pressure timing signal during ankle pump movement and the multi-channel electromyographic signal during electromyographic stimulation. The collaborative control unit, which is communicatively connected to the signal acquisition unit, includes: The phase prediction submodule is used to estimate the ankle pump phase based on the temporal characteristics of the foot pressure signal and calculate the predicted peak time to reach the next dorsiflexion peak. The fatigue estimation submodule is used to extract multidimensional features of median frequency, root mean square amplitude and peak contraction time from the multi-channel electromyography signal, and to calculate the comprehensive fatigue index by weighted fusion. The parameter optimization decision submodule is used to output electrical stimulation instructions, optimal stimulation parameter combinations, and training adjustment instructions based on the predicted peak time and the comprehensive fatigue index. The rehabilitation execution unit, which is communicatively connected to the collaborative control unit, includes a pneumatic ankle pump drive device for driving ankle pump movement and an electromyographic stimulation device for applying electrical stimulation to the pelvic floor muscles.

2. The ankle pump-pelvic floor muscle synergistic activation rehabilitation system based on pressure feedback closed loop according to claim 1, characterized in that, The collaborative control unit also includes a compensation identification submodule, which is used to acquire multi-channel electromyographic signals, including the reference electromyographic signals of the pelvic floor muscles, the reference electromyographic signals of the abdomen, and the reference electromyographic signals of the adductor muscles of the thigh. The activation timing and amplitude ratio characteristics of the multi-channel electromyographic signals are used as comparison parameters to identify the compensation force exertion mode by comparing the similarity with a preset template. The preset templates include normal activation templates and compensation mode templates. The compensation mode templates include at least abdominal muscle priority activation compensation, thigh muscle synchronous activation compensation, and pelvic floor muscle ineffective activation compensation. Among them, similarity comparison and recognition calculates the similarity distance between real-time multi-channel electromyography signals and normal activation templates and various compensation mode templates through dynamic time warping algorithm, and determines the activation mode category of the current cycle based on the principle of minimum distance. The identification results are output to the parameter optimization decision submodule.

3. The ankle pump-pelvic floor muscle synergistic activation rehabilitation system based on pressure feedback closed loop according to claim 2, characterized in that, The compensation identification submodule further includes a correction feedback module, which is configured as follows: A compensation detection module is set up to detect the number of times the compensation mode is triggered. When a single compensatory exertion is detected, the first corrective instruction is output, including generating a corresponding voice prompt or tactile feedback, and the current contraction cycle is not counted as an effective rehabilitation training session. When the number of consecutive compensations exceeds the preset first threshold, a second correction instruction is output, including reducing the ankle pump movement speed and increasing the pre-trigger amount, in order to prolong the time window for the patient to perceive pelvic floor muscle contraction. When the number of consecutive compensations exceeds the preset second threshold, a third correction instruction is output, including pausing ankle pump movement and outputting low-intensity electrical stimulation to guide the patient to reconstruct the correct force exertion pattern. The second threshold is greater than the first threshold.

4. The ankle pump-pelvic floor muscle synergistic activation rehabilitation system based on pressure feedback closed loop according to claim 1, characterized in that, The phase prediction submodule includes: The sliding window module is used to collect the sliding time window of the most recent N complete ankle pump exercise cycles and extract the dorsiflexion peak time sequence and cycle duration sequence of each cycle; The weighted estimation module is used to calculate the weighted average period duration of the period duration series with time-increasing weights. The phase estimation module is used to estimate the current motion phase in real time and predict the peak time of the next backbend based on the weighted average period duration and the previous backbend peak time.

5. The ankle pump-pelvic floor muscle synergistic activation rehabilitation system based on pressure feedback closed loop according to claim 1, characterized in that, The parameter optimization decision submodule calculates the advanced triggering time based on the predicted peak time and the individualized neuromuscular conduction delay, and sends an electrical stimulation triggering command to the electromyography stimulation device to align the peak contraction time of the pelvic floor muscles with the peak time of the ankle pump dorsiflexion. The advanced triggering time is calculated using the following formula: ； This is the predicted peak time for the next dorsiflexion peak. For individualized neuromuscular conduction delay, Fixed system processing delay; The individualized neuromuscular conduction delay is obtained through a delay calibration module, which is configured to: emit test electrical stimulation pulses at fixed times, simultaneously acquire multi-channel electromyographic signals of the pelvic floor muscles, calculate the time interval from the emission of stimulation to the electromyographic amplitude exceeding a preset multiple of the resting baseline, and repeat the process multiple times to obtain the median.

6. The ankle pump-pelvic floor muscle synergistic activation rehabilitation system based on pressure feedback closed loop according to claim 5, characterized in that, The parameter optimization decision submodule also includes detecting the actual peak contraction time of the multi-channel electromyographic signal of the pelvic floor muscles after electrical stimulation triggering, and calculating the timing synchronization error between the measured dorsiflexion peak and the actual peak. If the absolute value of the synchronization error exceeds the preset threshold, the individualized neuromuscular transmission delay is adaptively corrected according to the preset learning rate, so that the corrected delay converges iteratively in subsequent cycles.

7. The ankle pump-pelvic floor muscle synergistic activation rehabilitation system based on pressure feedback closed loop according to claim 5, characterized in that, The parameter optimization decision submodule outputs training adjustment instructions based on the comprehensive fatigue index, and the method includes: A preset fatigue grading threshold is set. Based on the comparison between the comprehensive fatigue index and the fatigue grading threshold, corresponding training adjustment instructions are output: When the overall fatigue index is at a normal level, the output training adjustment instruction is to maintain or increase the intensity of rehabilitation training. When the overall fatigue index is at the mild to moderate fatigue level, the output training adjustment instruction is to prolong the rest time between contraction and relaxation and / or reduce the intensity of stimulation. When the overall fatigue index is at the severe fatigue level, the output training adjustment instruction is to suspend rehabilitation training and enter a forced rest period. During the rest period, the overall fatigue index is continuously monitored, and rehabilitation training is automatically resumed when it drops back to the normal level.

8. The ankle pump-pelvic floor muscle synergistic activation rehabilitation system based on pressure feedback closed loop according to claim 1, characterized in that, The method by which the parameter optimization decision submodule outputs the optimal combination of stimulus parameters is to set up an individual modeling submodule, which includes: The sparse sampling module is used to acquire the combination of parameters consisting of stimulation frequency, stimulation intensity and pulse width during the rehabilitation training process, and to record the amplitude of the electromyographic response on the pelvic floor muscle surface under different parameter combinations. The response surface fitting module is used to fit an individualized stimulus response function with stimulus frequency, stimulus intensity, and pulse width as independent variables and the amplitude of electromyographic response on the pelvic floor muscle surface as the dependent variable by using polynomial response surface regression that includes independent variable cross terms and quadratic terms.

9. The ankle pump-pelvic floor muscle synergistic activation rehabilitation system based on pressure feedback closed loop according to claim 8, characterized in that, The individual modeling submodule also includes an online update module, which is configured as follows: During each rehabilitation training session, additional sampling was performed on parameter combinations with high prediction uncertainty of the individualized stimulus response function, and the measured data were added to the historical dataset. The historical dataset was re-regressed using time decay weights to update the individualized stimulus-response function. The time decay weights were set such that the weight of recent measured data was greater than that of distant data.

10. The ankle pump-pelvic floor muscle synergistic activation rehabilitation system based on pressure feedback closed loop according to claim 1, characterized in that, The parameter optimization decision submodule also includes a priority arbitration module, which is used to set the priority of training adjustment instructions to be higher than that of electrical stimulation instructions.

Images