Intelligent generation method and system for home pelvic floor muscle rehabilitation training program

Abstract

The present application relates to the technical field of medical device use, and discloses a method and system for intelligently generating a home pelvic floor muscle rehabilitation training program. In the rehabilitation training process, a selection operation of a pelvic floor muscle dynamics model and a local program generation model is introduced. On the one hand, the pelvic floor muscle dynamics model is used to simulate elastic response to obtain accurate and reliable local stage evaluation data of the target training stage. On the other hand, by selecting a local program generation model that matches the actual situation of the target training stage, a more matched targeted training program can be obtained according to the accurate and reliable local stage evaluation data and the stage test data of the target training stage through the matching local program generation model, thereby providing accurate and high-quality data dependence for adjusting the global rehabilitation training program, and being able to improve the intelligence level of training program generation, thereby effectively improving the home rehabilitation training effect of the pelvic floor muscles.

Description

Technical Field

[0001] The present application relates to the technical field of medical device use, and in particular to a method and system for intelligently generating a home pelvic floor muscle rehabilitation training program. Background Art

[0002] Pelvic floor muscle problems are a type of health problem that seriously troubles people's lives. Postpartum women often face pelvic floor muscle relaxation or damage, which affects their daily activities and quality of life; as middle-aged and elderly people age, their pelvic floor muscles gradually weaken, which can easily lead to problems such as urinary incontinence; and people who sit for a long time may also cause pelvic floor muscle damage due to long-term pressure.

[0003] Pelvic floor muscle rehabilitation training is a series of specific exercises and treatment methods to help enhance the strength and function of the pelvic floor muscles and improve various problems caused by pelvic floor muscle relaxation or injury. It usually requires going to the hospital for a pelvic floor muscle test. The doctor will evaluate the health of the pelvic floor muscles based on the test results and develop a personalized rehabilitation training plan. With the advancement of technology, the assistance of some intelligent equipment makes it possible to conduct pelvic floor muscle rehabilitation training at home. After the pelvic floor muscle test, users can obtain a customized training plan based on the test data and conduct effective rehabilitation training.

[0004] However, the intelligence level of generating pelvic floor muscle rehabilitation training programs in related technologies needs to be improved. Summary of the invention

[0005] The present application provides a method and system for intelligently generating a home pelvic floor muscle rehabilitation training program, which solves the technical problem in the related art that the degree of intelligence in generating pelvic floor muscle rehabilitation training programs needs to be improved, and achieves the technical effect of providing users with a more effective rehabilitation training program.

[0006] In order to achieve the above objectives, the main technical solutions adopted in this application include:

[0007] In a first aspect, an embodiment of the present application provides a method for intelligently generating a home pelvic floor muscle rehabilitation training program, the method comprising:

[0008] Acquire historical test data of the pelvic floor muscles of the historical subject generated during the full training phase; wherein the full training phase is divided into a plurality of local training phases;

[0009] Model training is performed based on the historical test data to obtain a global solution generation model and a local generation model set; wherein the global solution generation model is used to perform solution prediction for the entire training stage to obtain a global rehabilitation training plan, and the local generation model set includes a local solution generation model corresponding to each local training stage, and the local solution generation model is used to perform solution prediction for the corresponding local training stage prediction to obtain a targeted training plan;

[0010] Acquire electromyographic test data of the pelvic floor muscles of the current subject generated during the entire training phase; and determine a target training phase that needs to be improved in the multiple local training phases according to the electromyographic test data;

[0011] Determine the local solution generation model corresponding to the target training stage in the local generation model set, and determine the stage test data of the target training stage in the electromyography test data;

[0012] Performing elastic response simulation on the pelvic floor muscle dynamics model according to the stage test data of the target training stage to obtain local stage evaluation data of the target training stage; wherein the local stage evaluation data is determined based on the elastic response simulation result of the pelvic floor muscle dynamics model;

[0013] Input the electromyographic test data into the global solution generation model for global prediction to obtain the global rehabilitation training solution for the current subject; perform local prediction based on the stage test data and local stage evaluation data of the target training stage through the local solution generation model corresponding to the target training stage to obtain a targeted training solution suitable for the target training stage;

[0014] The global rehabilitation training program is adjusted based on the targeted training program suitable for the target training stage to obtain a target rehabilitation training program suitable for the pelvic floor muscles of the current subject.

[0015] In a second aspect, an embodiment of the present application provides a system for intelligently generating a home pelvic floor muscle rehabilitation training program, the system comprising:

[0016] A historical test data acquisition module, used to acquire historical test data generated by the pelvic floor muscles of a historical subject during a full training phase; wherein the full training phase is divided into a plurality of local training phases;

[0017] A global-local model training module, used for performing model training based on the historical test data to obtain a global solution generation model and a local generation model set; wherein the global solution generation model is used for performing solution prediction for the entire training stage to obtain a global rehabilitation training plan, and the local generation model set includes a local solution generation model corresponding to each local training stage, and the local solution generation model is used for performing solution prediction for the corresponding local training stage prediction to obtain a targeted training plan;

[0018] The electromyographic test data acquisition module is used to acquire the electromyographic test data generated by the pelvic floor muscles of the current subject during the whole training phase; and determine the target training phase that needs to be improved in the multiple local training phases according to the electromyographic test data;

[0019] A local model data selection module, used to determine the local solution generation model corresponding to the target training stage in the local generation model set, and to determine the stage test data of the target training stage in the electromyography test data;

[0020] A local evaluation data determination module, used to perform elastic response simulation on the pelvic floor muscle dynamics model according to the stage test data of the target training stage, to obtain local stage evaluation data of the target training stage; wherein the local stage evaluation data is determined based on the elastic response simulation result of the pelvic floor muscle dynamics model;

[0021] A global-local solution generation module is used to input the electromyography test data into the global solution generation model for global prediction, so as to obtain the global rehabilitation training solution of the current object; through the local solution generation model corresponding to the target training stage, local prediction is performed according to the stage test data and local stage evaluation data of the target training stage, so as to obtain a targeted training solution suitable for the target training stage;

[0022] The local global program adjustment module is used to adjust the global rehabilitation training program based on the targeted training program suitable for the target training stage to obtain a target rehabilitation training program for the pelvic floor muscles of the current object.

[0023] In an embodiment of the present application, a global solution generation model, a local generation model set and a pelvic floor muscle dynamics model are pre-constructed. After obtaining the electromyographic test data of the pelvic floor muscles of the current subject during the entire training stage, the target training stage that needs to be targeted for improvement and the stage test data of the target training stage are first determined; secondly, according to the stage test data of the target training stage, elastic response simulation is performed on the pelvic floor muscle dynamics model to obtain local stage evaluation data of the target training stage; then, the local solution generation model corresponding to the target training stage is determined in the local generation model set, and a targeted training solution suitable for the target training stage is obtained through the local solution generation model according to the local stage evaluation data and stage test data of the target training stage; finally, the global rehabilitation training solution predicted by the global solution generation model is adjusted using the targeted training solution to obtain the target rehabilitation training solution. During the rehabilitation training process, the selection operation of the pelvic floor muscle dynamics model and the local solution generation model is introduced. On the one hand, the pelvic floor muscle dynamics model is used to perform elastic response simulation to obtain accurate and reliable local stage evaluation data of the target training stage. On the other hand, by selecting the local solution generation model that matches the actual situation of the target training stage, a more matched targeted training plan can be obtained according to the accurate and reliable local stage evaluation data and the stage test data of the target training stage through the matching local solution generation model, which provides accurate and high-quality data dependence for adjusting the global rehabilitation training plan, can improve the intelligence level of training plan generation, and thus effectively improve the home rehabilitation training effect of the pelvic floor muscles. BRIEF DESCRIPTION OF THE DRAWINGS

[0024] In order to more clearly illustrate the specific implementation methods of the present application or the technical solutions in the prior art, the drawings required for use in the specific implementation methods or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are some implementation methods of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0025] Figure 1 A schematic diagram of an intelligent generation system for a home pelvic floor muscle rehabilitation training program in an example scenario of this application;

[0026] Figure 2 A flow chart of a method for intelligently generating a home pelvic floor muscle rehabilitation training program provided in an embodiment of the present application;

[0027] Figure 3 A flow chart of a method for intelligently generating a home pelvic floor muscle rehabilitation training program provided in an embodiment of the present application;

[0028] Figure 4 A flow chart of a method for intelligently generating a home pelvic floor muscle rehabilitation training program provided in an embodiment of the present application;

[0029] Figure 5 A flow chart of a method for intelligently generating a home pelvic floor muscle rehabilitation training program provided in an embodiment of the present application;

[0030] Figure 6 A flow chart of a method for intelligently generating a home pelvic floor muscle rehabilitation training program provided in an embodiment of the present application;

[0031] Figure 7 A flow chart of a method for intelligently generating a home pelvic floor muscle rehabilitation training program provided in an embodiment of the present application;

[0032] Figure 8 A flow chart of a method for intelligently generating a home pelvic floor muscle rehabilitation training program provided in an embodiment of the present application;

[0033] Fig. 9 A schematic diagram of a system for intelligently generating a home pelvic floor muscle rehabilitation training program provided in an embodiment of the present application;

[0034] Fig.10 A schematic diagram of the structure of a computer device provided in an embodiment of this specification. DETAILED DESCRIPTION

[0035] In order to make the purpose, technical solution and advantages of the embodiments of the present application clearer, the technical solution in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative work are within the scope of protection of the present application.

[0036] Pelvic floor muscle problems usually require professional pelvic floor muscle electromyography testing in the hospital. Doctors evaluate the health of the patient's muscles based on the results of the electromyography test and develop a personalized rehabilitation training plan. With the advancement of technology, the emergence of some intelligent devices has made it possible to perform pelvic floor muscle rehabilitation training at home.

[0037] Pelvic floor muscle rehabilitation training at home is convenient for patients. There is no need to travel back and forth to the hospital frequently, which saves time and transportation costs. Training can also be flexibly performed according to personal schedules. Secondly, it is private, avoiding the embarrassment of receiving pelvic floor muscle testing or training in the hospital, and reducing the psychological pressure on patients.

[0038] However, the intelligence level of relevant technologies in generating pelvic floor muscle rehabilitation training programs still needs to be improved, especially how to achieve personalized and precise treatment effects to achieve similar results as receiving treatment under the guidance of a doctor in a hospital.

[0039] Therefore, this application scenario example proposes a method for intelligently generating a home pelvic floor muscle rehabilitation training program, please refer to Figure 1 The method is applied to a home-based pelvic floor muscle rehabilitation training program intelligent generation system. The pelvic floor muscle stimulation feedback device 101 can be used to collect the surface electromyographic signals of the patient's pelvic floor muscles, conduct biofeedback training, and treat muscle dysfunction through electrical stimulation and electromyographic triggered electrical stimulation. The computer device 102 has a home-based pelvic floor muscle rehabilitation training program intelligent generation system, which can be used to generate a pelvic floor muscle rehabilitation training program based on the pelvic floor muscle electromyographic test data.

[0040] Among them, the pelvic floor muscle stimulation feedback instrument 101 collects the surface electromyographic signals of the patient's pelvic floor muscles, which usually include the pre-resting stage, the fast muscle stage, the slow muscle stage, the endurance test stage and the post-resting stage. In the pre-resting stage, the baseline activity of the electromyographic signal is first recorded to evaluate the electrical activity of the pelvic floor muscles in a relaxed state; then enter the fast muscle stage, at this time, through stimulation or spontaneous contraction, observe the responsiveness of the pelvic floor muscles to rapid reactions, and mainly evaluate the activation of fast muscle fibers; in the slow muscle stage, detect the low-frequency activity of the pelvic floor muscles over a long period of time, evaluate its adaptability to continuous loads, and reflect the endurance level of slow muscle fibers; the endurance test stage focuses on evaluating the sustained contraction ability of the pelvic floor muscles, and observes the time and intensity of contraction through continuous activation; finally enter the post-resting stage, record the recovery of the pelvic floor muscles after relaxation, and evaluate the recovery ability and function of the muscles. After the electromyographic signal collection of these stages is completed, electromyographic test data is obtained through filtering, analog-to-digital conversion and other processing. The data is time series data and can be input into the computer device 102 for analysis to evaluate the functional status of the pelvic floor muscles and generate a pelvic floor muscle evaluation report (see Table 1), and generate a personalized rehabilitation training plan for the patient.

[0041] Table 1 Sample pelvic floor muscle assessment report

[0042]

[0043] Exemplarily, the home pelvic floor muscle rehabilitation training program intelligent generation system generates a rehabilitation training program for the patient through a pre-trained long short-term memory (LSTM) large model. The model is trained using patient background data (gender, age, etc.), electromyography test data, initial assessment results of pelvic floor muscle function, and treatment plan data as training data sets. During training, labels such as "high intensity" and "low intensity" are used for the treatment plan. When the system calls the model to generate a rehabilitation training program, the model input layer first receives the electromyography test data, user gender, and age. The patient's gender and age can be obtained when the system registers the user, and the electromyography test data can be input by the pelvic floor muscle stimulation feedback instrument 101. Then the LSTM layer of the model can infer the future pelvic floor muscle response based on the pelvic floor muscle electromyography test data. Then the fully connected layer generates a personalized treatment plan for each patient based on the electromyography test data, patient background information and other features, combined with the time series pattern learned by the LSTM model. Finally, the output layer of the model outputs the rehabilitation training plan, including suggestions for the treatment plan, such as training items, training intensity, frequency, duration, etc.

[0044] Typically, the rehabilitation training program includes one or more of biofeedback training, electrical stimulation, electromyographically triggered electrical stimulation, Kegel training, and breathing relaxation training. Among them, biofeedback training, electrical stimulation, and electromyographically triggered electrical stimulation can be performed with the assistance of the pelvic floor muscle stimulation feedback device 101, and Kegel training and breathing relaxation training can be performed with the assistance of the computer device 102.

[0045] Biofeedback training is a training method that helps patients enhance their control over their pelvic floor muscles by monitoring pelvic floor muscle activity in real time through the pelvic floor muscle stimulation feedback instrument 101. During the training process, patients obtain real-time activity data of their pelvic floor muscles, such as the intensity and frequency of muscle contraction, through sensors connected to the pelvic floor muscle stimulation feedback instrument 101. These data are fed back to patients in a timely manner through screen display or voice broadcast, helping them to be aware of the state of their muscles and thus guiding them to correctly contract and relax their muscles.

[0046] Electrical stimulation utilizes the low-frequency current generated by the pelvic floor muscle stimulation feedback device 101 to stimulate the pelvic floor muscles, causing them to contract and relax, simulating normal muscle activity to promote blood circulation and improve muscle tension.

[0047] EMG-triggered electrical stimulation is a treatment method that combines pelvic floor muscle EMG signal monitoring with electrical stimulation. It uses the natural EMG activity of the patient's pelvic floor muscles to trigger electrical stimulation, thereby enhancing the function of the pelvic floor muscles. The pelvic floor muscle stimulation feedback device 101 monitors the EMG signals of the pelvic floor muscles in real time through electrodes. When the EMG signal reaches the preset threshold, the pelvic floor muscle stimulation feedback device 101 will automatically output electrical stimulation to cause the pelvic floor muscles to contract. This method can not only promote the strength and endurance of the pelvic floor muscles, but also improve the coordination of nerves and muscles.

[0048] Kegel training is mainly carried out through self-awareness and control. By repeatedly practicing contracting (tightening) and relaxing the pelvic floor muscles, the control of these muscles is enhanced, thereby improving the overall health of the pelvic floor muscles. Breathing relaxation training focuses on improving the function of the pelvic floor muscles through deep breathing and relaxation techniques. Through deep abdominal breathing (such as diaphragmatic breathing), patients can learn how to relax the pelvic floor muscles when inhaling and slightly contract the pelvic floor muscles when exhaling. The duration and movement methods of these two projects can be formulated in the rehabilitation training program.

[0049] After the patient has been trained using a certain rehabilitation training program for a period of time, the patient can undergo electromyography testing again using the pelvic floor muscle stimulation feedback device 101, and input the new data into the computer device 102 to evaluate the condition of the pelvic floor muscles and generate a new rehabilitation training program.

[0050] According to an embodiment of the present application, an embodiment of a method for intelligently generating a home pelvic floor muscle rehabilitation training program is provided. It should be noted that the steps shown in the flowchart of the accompanying drawings can be executed in a computer system such as a set of computer executable instructions, and although a logical order is shown in the flowchart, in some cases, the steps shown or described can be executed in an order different from that shown here.

[0051] See also Figure 2 The method for intelligently generating a home pelvic floor muscle rehabilitation training program provided in the embodiment of the present application includes:

[0052] S101. Obtain historical test data of the pelvic floor muscles of a historical subject generated during the entire training phase.

[0053] Among them, the full training stage is divided into multiple local training stages.

[0054] S103: Perform model training based on historical test data to obtain a global solution generation model and a set of local generation models.

[0055] Among them, the global solution generation model is used to predict the solution for the entire training stage to obtain a global rehabilitation training plan. The local generation model set includes a local solution generation model corresponding to each local training stage. The local solution generation model is used to predict the solution for the corresponding local training stage to obtain a targeted training plan.

[0056] S105, obtaining electromyographic test data of the pelvic floor muscles of the current subject during the entire training phase; and determining a target training phase that needs to be improved in a plurality of local training phases according to the electromyographic test data.

[0057] S107, determining a local solution generation model corresponding to the target training stage in the local generation model set, and determining stage test data of the target training stage in the electromyography test data.

[0058] S109. Perform elastic response simulation on the pelvic floor muscle dynamics model according to the stage test data of the target training stage to obtain local stage evaluation data of the target training stage.

[0059] Among them, the local stage assessment data is determined based on the elastic response simulation results of the pelvic floor muscle dynamics model.

[0060] S111. Input the electromyography test data into the global program generation model for global prediction to obtain the global rehabilitation training program for the current object; through the local program generation model corresponding to the target training stage, perform local prediction based on the stage test data of the target training stage and the local stage evaluation data to obtain a targeted training program suitable for the target training stage.

[0061] S113: adjusting the global rehabilitation training program based on the targeted training program suitable for the target training stage to obtain a target rehabilitation training program for the pelvic floor muscles suitable for the current subject.

[0062] In this embodiment, a first training data set corresponding to the full training stage and a second training data set corresponding to each local training stage are constructed based on historical test data. The first training data set is used to perform model training to obtain a global solution generation model. For each local training stage, the second training data set corresponding to each local training stage is used to perform model training to obtain a local solution generation model corresponding to each local training stage, thereby forming a local generation model set.

[0063] In this embodiment, the pelvic floor muscle dynamics model can be understood as a biomechanical model used to understand and simulate the movement and mechanical response of the pelvic floor muscles and their interaction with surrounding tissues. The model can be used to quantitatively analyze and predict the function of the pelvic floor muscles, thereby obtaining local stage evaluation data for the target training stage. Specifically, after obtaining the stage test data for the target training stage, the stage test data for the target training stage is applied to the pelvic floor muscle dynamics model to perform elastic response simulation to obtain elastic response simulation results. Since the elastic response simulation results quantitatively reflect the strength and endurance of the pelvic floor muscles, the local stage evaluation data for the target training stage is determined based on the elastic response simulation results, such as using the elastic response simulation results as local stage evaluation data for the target training stage.

[0064] In the above embodiment, a global solution generation model, a local generation model set and a pelvic floor muscle dynamics model are pre-constructed. After obtaining the electromyographic test data generated by the pelvic floor muscles of the current subject during the entire training stage, the target training stage that needs to be targeted for improvement and the stage test data of the target training stage are first determined; secondly, according to the stage test data of the target training stage, elastic response simulation is performed on the pelvic floor muscle dynamics model to obtain local stage evaluation data of the target training stage; then, the local solution generation model corresponding to the target training stage is determined in the local generation model set, and a targeted training solution suitable for the target training stage is obtained through the local solution generation model according to the local stage evaluation data and stage test data of the target training stage; finally, the global rehabilitation training solution predicted by the global solution generation model is adjusted using the targeted training solution to obtain the target rehabilitation training solution. During the rehabilitation training process, the selection operation of the pelvic floor muscle dynamics model and the local solution generation model is introduced. On the one hand, the pelvic floor muscle dynamics model is used to perform elastic response simulation to obtain accurate and reliable local stage evaluation data of the target training stage. On the other hand, by selecting the local solution generation model that matches the actual situation of the target training stage, a more matched targeted training plan can be obtained according to the accurate and reliable local stage evaluation data and the stage test data of the target training stage through the matching local solution generation model, which provides accurate and high-quality data dependence for adjusting the global rehabilitation training plan, can improve the intelligence level of training plan generation, and thus effectively improve the home rehabilitation training effect of the pelvic floor muscles.

[0065] The present application also provides a method for intelligently generating a home pelvic floor muscle rehabilitation training program, comprising the following steps:

[0066] S110, obtaining electromyographic test data of the pelvic floor muscles of the current subject generated during the entire training phase.

[0067] The full training phase includes a target training phase in which rehabilitation training parameters need to be adjusted in a targeted manner, and the electromyography test data includes stage test data of the target training phase.

[0068] The full training phase can be a comprehensive phase of pelvic floor muscle training, including the pre-resting training phase, the fast muscle training phase, the slow muscle training phase, the endurance training phase, and the post-resting training phase. Collecting electromyographic test data during the full training phase can help comprehensively evaluate the functional status of the pelvic floor muscles.

[0069] The electromyographic test data can be obtained by processing the continuous electromyographic signals collected during the entire training phase. For example, the electromyographic test can be performed on the current subject through a pelvic floor muscle stimulation feedback instrument to obtain the electromyographic signals of the pelvic floor muscle activity during the entire training phase, and the electromyographic test data can be obtained through filtering, analog-to-digital conversion, and other processing. The electromyographic test data is time series data.

[0070] Rehabilitation training parameters can be specific values ​​or standards used to guide and adjust the intensity, duration, etc. of the training in the rehabilitation training program. These parameters help ensure the effectiveness and safety of the training and can be personalized according to the specific situation of the current user.

[0071] The stage test data may be electromyographic test data of a certain stage of the pre-rest training stage, the fast muscle training stage, the slow muscle training stage, the endurance training stage, and the post-rest training stage.

[0072] Specifically, the current user can use the pelvic floor muscle stimulation feedback instrument to train the pelvic floor muscles according to the voice and screen prompts of the instrument, and complete the five stages of training in sequence, namely the pre-resting training stage, the fast muscle training stage, the slow muscle training stage, the endurance training stage, and the post-resting training stage. The pelvic floor muscle stimulation feedback instrument first collects electromyographic signals through sensors, and then processes the electromyographic signals and converts them into digital signals to obtain the electromyographic test data of the current subject in the entire training stage, including the pre-resting training stage, the fast muscle training stage, the slow muscle training stage, the endurance training stage, and the post-resting training stage. After obtaining the electromyographic test data of the entire training stage, the pelvic floor muscles have different performances at different stages. According to the performance of the pelvic floor muscles, the target training stage where the rehabilitation training parameters are inappropriate can be clearly identified, and the rehabilitation training parameters of this stage need to be adjusted in a targeted manner.

[0073] S120, inputting the electromyography test data into a global program generation model for global prediction, and obtaining a global rehabilitation training program for the current subject.

[0074] Among them, the global solution generation model can be a pre-trained time series model using LSTM (Long Short-Term Memory Network). LSTM is a recursive neural network (RNN) specially designed for processing and predicting time series data. It can capture long-term dependencies and avoid the gradient vanishing problem in traditional RNN. The global rehabilitation training program is a global rehabilitation training program obtained after a global prediction based on the electromyography test data of the full training phase and a comprehensive assessment of the pelvic floor muscle function status of the current subject.

[0075] Specifically, the EMG test data is first feature extracted and input into the input layer of the timing model. Secondly, it passes through multiple LSTM layers to capture the short-term and long-term dependencies in the EMG test timing data, helping the model learn the dynamic changes of pelvic floor muscle activity. Then, the output of the LSTM layer is connected to a fully connected layer to generate a personalized treatment plan for each current user based on the EMG test data and the timing pattern learned by the LSTM model. Finally, the output layer of the model generates a global rehabilitation training plan, which may include training plans, training items, intensity, duration, etc.

[0076] S130. Perform a local stage function assessment on the pelvic floor muscles of the current subject according to the stage test data of the target training stage to obtain local stage assessment data of the target training stage.

[0077] The target training phase may be a partial phase among the pre-rest training phase, the fast muscle training phase, the slow muscle training phase, the endurance training phase and the post-rest training phase.

[0078] According to the start and end time points of each local stage, the stage test data of the local stage can be obtained, and the test value of each local stage can be obtained by statistics. Then, the test value and the reference value are compared and analyzed to obtain the evaluation data of each local stage. For example, please refer to Table 1. Multiple test values ​​of each local stage can be compared with the corresponding reference values, scored, and then each score is added according to the weight to obtain the local stage evaluation data. For example, the local stage evaluation data can also be in the form of text descriptions such as "high", "low", and "normal". The target training stage can be the training stage with the worst local stage evaluation data. If the performance of a certain stage is found to be the worst in the evaluation of these training stages, it will be used as the target training stage for key intervention. Through in-depth training and adjustment of this stage, the current user can be helped to improve the functions of this stage.

[0079] Exemplarily, the target training stage may also be a training stage selected by the current subject according to his / her own will if there are multiple training stages with poor evaluation results.

[0080] S140, predicting a local training plan based on the stage test data of the target training stage and the local stage evaluation data, and obtaining a targeted training plan suitable for the target training stage.

[0081] Among them, the targeted training program is an adjustment program specially customized according to the needs of the target training stage. Compared with the global rehabilitation training program, it is more refined and specific. Targeted training programs usually include detailed adjustments to training intensity, frequency, duration, and movement methods. For example, in the pre-resting training stage, it may be necessary to strengthen muscle relaxation and recovery to reduce tension; in the fast-twitch training stage, the training program may focus on improving the rapid response ability of muscles and enhancing muscle explosiveness through high-intensity, short-term contraction training; in the slow-twitch training stage, the focus may be on improving muscle endurance and sustained contraction ability. The training intensity is relatively low, but the training time is long, aiming to increase muscle tolerance. Through such personalized adjustments, the training effect can be effectively improved.

[0082] Specifically, a knowledge base between stage test data, local stage evaluation data and rehabilitation training parameter value ranges is pre-built on the computer system. The method for predicting the local training plan can be to first extract features from the stage test data and local stage evaluation data of the current target training stage, then perform feature fusion to obtain fusion features of the stage test data and local stage evaluation data, and finally perform a similarity comparison between the fusion feature and each fusion feature in the knowledge base, and select the training plan corresponding to the fusion feature with the highest similarity as the targeted training plan for the target training stage.

[0083] S150: Adjust the global rehabilitation training program based on the targeted training program to obtain a target rehabilitation training program for the pelvic floor muscles suitable for the current subject.

[0084] Among them, the target rehabilitation training program is obtained by adjusting the specific performance of the current subject in the target training stage based on the formulation of the global rehabilitation training program. It can be personalized according to the needs and wishes of the current subject in a certain local stage, or it can be more targeted according to the local stage with the worst evaluation data obtained by system analysis.

[0085] Specifically, as a macro and comprehensive training program, the global rehabilitation training program can train the pelvic floor muscles of the current subject from a more comprehensive global perspective. The global training program can be regarded as a framework for providing general training guidance, while the targeted training program is aimed at the needs of the pelvic floor muscles of the current subject in the target training stage. The targeted training program is an adjustment program specifically for the target training stage. The targeted training program is more refined and specific than the global rehabilitation training program. Therefore, after determining the global rehabilitation training program of the current subject and the targeted training program suitable for the target training stage, the global rehabilitation training program is accurately adjusted using the targeted training program to obtain a personalized target rehabilitation training program that can be more suitable for the actual situation of the current subject. Furthermore, during the rehabilitation training process of the current subject, the program can be accurately adjusted according to the rehabilitation training effect, which can accurately improve the rehabilitation training effect.

[0086] Exemplarily, the global rehabilitation training program is to conduct electrical stimulation training every week for 10 minutes, with medium current intensity, frequency and pulse width. The target training stage is the pre-resting training stage, and the targeted training program is to conduct electrical stimulation training every week for 10 minutes, with medium current intensity and pulse width, and low frequency; and perform breathing relaxation training for 15 minutes every week. Then, based on the targeted training program, the global rehabilitation training program is adjusted to obtain a target rehabilitation training program for the pelvic floor muscles suitable for the current object, which can be electrical stimulation training every week for 10 minutes, with medium current intensity and pulse width, and low frequency; and perform breathing relaxation training for 15 minutes every week.

[0087] In the above embodiment, the electromyographic test data generated by the pelvic floor muscles of the current object during the whole training stage is first obtained; secondly, the electromyographic test data is input into the global scheme generation model for global prediction to obtain the global rehabilitation training scheme of the current object; then the local stage function evaluation of the pelvic floor muscles of the current object is performed according to the stage test data of the target training stage to obtain the local stage evaluation data of the target training stage; then the local training scheme is predicted based on the stage test data and the local stage evaluation data of the target training stage to obtain a targeted training scheme suitable for the target training stage; finally, the global rehabilitation training scheme is adjusted based on the targeted training scheme to obtain the target rehabilitation training scheme suitable for the pelvic floor muscles of the current object. In the rehabilitation training process, based on the premise of global considerations, the global scheme is finely adjusted locally according to the actual situation of the current object in each local stage, so as to realize an intelligent rehabilitation training method combining global considerations with local fine adjustments. This method automatically generates and adjusts the rehabilitation training scheme by analyzing the electromyographic test data and local stage function evaluation of the current object, thereby improving the intelligent level of training scheme generation, thereby effectively improving the home rehabilitation training effect of the pelvic floor muscles.

[0088] In some embodiments, see Figure 3 , through the local scheme generation model corresponding to the target training stage, local prediction is performed according to the stage test data of the target training stage and the local stage evaluation data, and a targeted training scheme suitable for the target training stage is obtained, including:

[0089] S210 , generating a surface electromyogram of the target training phase based on the phase test data of the target training phase.

[0090] Specifically, the corresponding stage test data is obtained from the electromyography test data according to the start and end time of the target training stage, so the stage test data is also time series data. The surface electromyogram, that is, the EMG graph, can be generated by inputting the stage training data into a pre-trained regression model (such as a simple neural network or a deep learning model). Because the regression model learns the relationship between the time series data and the EMG signal during the training phase, when the new time series data is input, the corresponding EMG graph can be generated.

[0091] S220, inputting the surface electromyogram, stage test data and local stage evaluation data of the target training stage into the local plan generation model for local prediction to obtain a targeted training plan.

[0092] Among them, the local solution generation model can be a pre-trained parallel sequence-to-sequence model (Seq2Seq) model and an image processing branch.

[0093] Specifically, the input surface electromyogram is subjected to feature extraction through the image processing branch to obtain image features of the surface electromyogram, which can help the model identify information such as the area, intensity, and frequency of pelvic floor muscle activity.

[0094] The encoder of the sequence-to-sequence model extracts features from the stage test data and the local stage evaluation data to obtain data features of the stage test data and the local stage evaluation data. The model can thereby identify the time dependency of the stage test data, such as the time pattern of pelvic floor muscle activity, dynamic changes of signals, etc.

[0095] The obtained image features and data features are concatenated and then decoded by the decoder of the sequence-to-sequence model. By concatenating multimodal features, that is, combining information from two different sources, the sequence-to-sequence model can simultaneously consider the features of the spatial (surface electromyography) and temporal (time series data) dimensions, thereby making more accurate and targeted training plans.

[0096] In the above embodiment, the surface electromyogram of the target training phase is first generated based on the stage test data of the target training phase, and then the surface electromyogram of the target training phase, the stage test data and the local stage evaluation data are input into the local solution generation model for local prediction to obtain a targeted training solution. A more accurate targeted training solution is obtained by multimodal data fusion, which enhances the personalization and accuracy of the training solution.

[0097] In some embodiments, see Figure 4 The local solution generation model includes a first encoder and a first decoder; the surface electromyogram of the target training stage, the stage test data and the local stage evaluation data are input into the local solution generation model for local prediction to obtain a targeted training solution, including:

[0098] S310, inputting the surface electromyogram, stage test data and local stage evaluation data of the target training stage into the first encoder for feature extraction, and obtaining the associated feature data and time series feature data of the target training stage;

[0099] S320, splicing the associated feature data and the time series feature data of the target training stage to obtain spliced ​​feature data of the target training stage;

[0100] S330: Input the spliced ​​feature data into a first decoder for decoding to obtain a targeted training plan.

[0101] The time series feature data may be feature data extracted by the first encoder, representing the change pattern of the phase test data and the local phase evaluation data in the time dimension. These time series feature data not only reflect the state at each time point, but also reflect the correlation and change trend between different time points, which is the basis for the subsequent generation of targeted training programs.

[0102] The associated feature data is the feature data extracted by the first encoder from the surface electromyography (EMG) of the target training stage, which characterizes the association between multiple feature dimensions. These feature dimensions include electromyography test data features, training action features (action names), training duration features, training intensity features, and training posture features (such as standing, sitting, or lying down). By identifying these associated features, the first encoder can identify the association between these features, thereby more accurately understanding the dynamic changes of the target training stage, and improving the evaluation of the functional status of the pelvic floor muscles and the prediction of the training effect.

[0103] Specifically, the local solution generation model can be a pre-trained sequence-to-sequence (Seq2Seq) model, in which there is a first encoder and a first decoder based on a neural network.

[0104] The first encoder can be used to perform feature extraction on the stage test data and the local stage evaluation data to obtain timing feature data of the stage test data and the local stage evaluation data, so as to identify the time dependency of the stage test data, for example, the time pattern of pelvic floor muscle activity, the dynamic change of the signal, etc.

[0105] The first encoder can also be used to extract features of the surface electromyogram in the target training phase to obtain associated feature data. The first encoder can use this feature to correlate the relationship between electromyography test data features, training action features (action names), training duration features, training intensity features, and training posture features.

[0106] The obtained correlation feature data and time series feature data are spliced ​​and input into the first decoder for decoding. By splicing multimodal features together and combining information from two different sources, the first decoder can simultaneously consider the characteristics of the pelvic floor muscles in space and time dimensions and the association with the training parameters, thereby making a more accurate and targeted training plan.

[0107] It should be noted that the neural network-based encoder and decoder can also adopt a long short-term memory model (LSTM).

[0108] In some embodiments, see Figure 5 , the first encoder includes a convolutional neural network branch and a recurrent neural network branch connected in parallel; the surface electromyogram, stage test data and local stage evaluation data of the target training stage are input into the first encoder for feature extraction to obtain the associated feature data and time series feature data of the target training stage, including:

[0109] S410, extracting features from the surface electromyogram through a convolutional neural network branch to obtain associated feature data of a target training phase;

[0110] S420, concatenating the stage test data of the target training stage and the local stage evaluation data to obtain the test evaluation joint data of the target training stage;

[0111] S430, extracting features from the test evaluation joint data of the target training phase through a recurrent neural network branch to obtain time series feature data of the target training phase.

[0112] Specifically, the local solution generation model can be a Transformer architecture, including a first encoder and a first decoder, wherein the first encoder includes a convolutional neural network branch and a recurrent neural network branch connected in parallel.

[0113] Extracting features from the surface electromyogram through a convolutional neural network branch to obtain associated feature data for a target training stage may include: first, performing convolution processing on the surface electromyogram to obtain test characterization features of the stage test data; then, performing pooling processing on the test characterization features based on the time interval corresponding to the stage test data to obtain associated feature data. Exemplarily, an attention allocation module may also be provided for the convolutional data network to perform weighted processing on the associated features, because different feature data (such as training action features and training duration features) have different degrees of influence on the stage test data.

[0114] Furthermore, in order to improve the accuracy of the local solution, not only the stage test data of the target training stage but also the local stage evaluation data of the target training stage are considered in the prediction process of the local solution. Therefore, the stage test data of the target training stage and the local stage evaluation data are spliced ​​to obtain the test evaluation joint data of the target training stage.

[0115] The test and evaluation joint data of the target training stage are subjected to feature extraction through a recurrent neural network branch to obtain the time series feature data of the target training stage, which may include: firstly, using a long short-term memory network to extract features from the stage training data to obtain a hidden layer feature matrix; secondly, using a weight parameter matrix to perform weighted processing on the hidden layer feature matrix in the time dimension and the feature dimension to obtain weighted time series features; finally, using the weight parameter on the feature dimension to perform weighted processing on the weighted time series features to obtain time series feature data.

[0116] In some embodiments, the multiple local training stages are respectively a pre-resting stage, a fast-twitch stage, a slow-twitch stage, an endurance test stage and a post-resting stage; the pre-resting stage, the fast-twitch stage, the slow-twitch stage, the endurance test stage and the post-resting stage correspond to their own local plan generation models.

[0117] Specifically, the pelvic floor muscle surface electromyographic signal of the current subject can be collected through the pelvic floor muscle stimulation feedback instrument, and the current subject can perform full-stage training according to the voice prompts of the pelvic floor muscle stimulation feedback instrument and the action prompts displayed on the screen, including the pre-resting stage, the fast muscle stage, the slow muscle stage, the endurance test stage and the post-resting stage. In the pre-resting stage, the baseline activity of the electromyographic signal is first recorded to evaluate the electrical activity of the pelvic floor muscle in a relaxed state; then enter the fast muscle stage, at this time, through stimulation or spontaneous contraction, observe the responsiveness of the pelvic floor muscle to rapid response, mainly evaluate the activation of fast muscle fibers; in the slow muscle stage, detect the low-frequency activity of the pelvic floor muscle for a long time, evaluate its adaptability to continuous load, and reflect the endurance level of slow muscle fibers; the endurance test stage focuses on evaluating the sustained contraction ability of the pelvic floor muscle, and observes the time and intensity of contraction through continuous activation; finally enter the post-resting stage, record the recovery of the pelvic floor muscle after relaxation, in order to evaluate the recovery ability and function of the muscle. The pelvic floor muscle stimulation feedback instrument collects electromyographic signals through sensors, and then converts the electromyographic signals into digital signals through signal processing to obtain electromyographic test data.

[0118] Exemplarily, the pre-resting stage, fast-twitch stage, slow-twitch stage, endurance test stage and post-resting stage can respectively have corresponding seq2seq local solution generation models. When any stage is selected as the target training stage, the corresponding seq2seq local solution generation model can predict the local training solution based on the input stage test data and local stage evaluation data to obtain a targeted training solution suitable for the target training stage.

[0119] In some embodiments, see Figure 6 The full training stage includes multiple local training stages, and the electromyography test data includes the stage test data of each of the multiple local training stages; the target training stage is determined by the following method:

[0120] S610, performing local performance evaluation on the pelvic floor muscles of the current subject according to the respective stage test data of the multiple local training stages, and obtaining the respective local performance data of the multiple local training stages.

[0121] S620: Determine, among the multiple local training stages, a target training stage in which the local performance data satisfies a preset parameter adjustment condition.

[0122] Among them, the preset parameters are the relevant training parameters in the global rehabilitation training program. When the local performance data meets specific conditions, the local stage can be determined as the target training stage, and then the local training program can be predicted based on the stage test data and local performance data of the target training stage to obtain a targeted training program suitable for the target training stage. Finally, the relevant training parameters in the global rehabilitation training program can be adjusted according to the targeted training program. The aforementioned conditions are the preset parameter adjustment conditions.

[0123] Specifically, please refer to Table 1. For a certain local training stage, the stage test data of the stage can be analyzed and statistically obtained to obtain the test values ​​of the required parameters (such as maximum value, average value, rise time, etc.), and then each test value is compared with the corresponding reference value, scored, and then each score is added according to the weight to obtain the local performance data. A percentage system can be used. When a certain local performance data is lower than a certain threshold (such as lower than 80 points), it is determined that the local stage is a local stage that needs to be trained.

[0124] Exemplarily, if the evaluation data obtained in the pre-resting stage, fast muscle stage, slow muscle stage, endurance test stage and post-resting stage are 77 points, 80 points, 90 points, 85 points and 95 points respectively, then the target training stage is the pre-resting stage. If there are two or more local performance data below 80, the one with the lower score can be determined as the target training stage according to the score. After a period of rehabilitation training, the home pelvic floor muscle rehabilitation training program intelligent generation system replaces another local stage that needs to be trained as the target training stage, and generates a new targeted training program again. According to the program, the preset parameters of the global rehabilitation training program obtained last time are adjusted to obtain a new target rehabilitation training program for the current object.

[0125] In the above embodiment, firstly, the local performance evaluation of the pelvic floor muscles of the current subject is performed according to the stage test data of each of the multiple local training stages, and the local performance data of each of the multiple local training stages is obtained; secondly, in the multiple local training stages, the target training stage whose local performance data meets the preset parameter adjustment conditions is determined. Clear target training stage information is intelligently provided for the preset parameter adjustment.

[0126] In some embodiments, see Figure 7 The global solution generation model includes an input embedding layer, a second encoder and a second decoder; the electromyography test data is input into the global solution generation model for global prediction to obtain a global rehabilitation training plan for the current object, including:

[0127] S710, performing vector representation on the electromyographic test data through an input embedding layer to obtain an electromyographic embedding vector corresponding to the electromyographic test data.

[0128] Among them, the global solution generation model can be a time series model of a pre-trained transformer architecture. The Transformer architecture has powerful sequence modeling capabilities and can capture long-term dependencies in time series data. Therefore, it is suitable for processing electromyography test data sequences at various stages of rehabilitation training.

[0129] Specifically, the continuous EMG test data is first divided into segments of fixed length according to the time step (for example, using the sliding window technology), and each segment is standardized or normalized. Then, each time series segment is input into the embedding layer and converted into a high-dimensional vector representation, that is, the EMG embedding vector corresponding to the EMG test data. The embedding vector can effectively capture the time series characteristics of the EMG test data.

[0130] S720. Position-encode the electromyographic embedding vector through a second encoder to obtain electromyographic feature data.

[0131] Since the Transformer model itself does not have any prior information about the sequence order, the addition of position encoding enables the model to understand the relative and absolute positions of elements in the sequence. The position encoding is added to the input EMG embedding vector to provide sequence order information. The position encoding can be generated using sine and cosine functions, using sine waves with different frequencies, and the dimension of each position encoding is the same as the dimension of the input EMG embedding vector. The calculation formula can be:

[0132]

[0133]

[0134] Where pos is the index of the position in the sequence, i is the index of the embedding dimension, is the dimension of the embedding vector.

[0135] The EMG feature data obtained through position encoding is used to characterize the long-term dependency information (such as the duration and periodicity of pelvic floor muscle activity) and global information (such as the overall pattern and trend of EMG data) in the EMG test data. The second encoder can establish long-distance dependencies between different time steps through the self-attention mechanism, thereby helping the model understand the overall EMG pattern.

[0136] S730 , decoding the electromyographic characteristic data by a second decoder to obtain a global rehabilitation training program for the current subject.

[0137] Specifically, first, the decoder receives the pelvic floor muscle EMG feature data from the second encoder and uses the self-attention mechanism to identify the association between different muscle activities. Then, the decoder maps these features to training goals, such as training intensity, duration, frequency, etc. Next, based on the long-range dependency of the EMG test data, the decoder recommends a suitable training cycle. Finally, a global rehabilitation training plan for the current subject is obtained.

[0138] In the above embodiment, the electromyographic test data is firstly represented by a vector through the input embedding layer to obtain the electromyographic embedding vector corresponding to the electromyographic test data; secondly, the electromyographic embedding vector is positionally encoded through the second encoder to obtain electromyographic feature data; finally, the electromyographic feature data is decoded through the second decoder to obtain the global rehabilitation training program of the current object. The global rehabilitation training program is intelligently generated according to the electromyographic test data through the deep learning model.

[0139] In some embodiments, see Figure 8 Before inputting the electromyographic test data into the global program generation model for global prediction to obtain the global rehabilitation training program for the current object, the method may further include:

[0140] S810: Evaluate the functional status of the pelvic floor muscles of the current subject according to the electromyography test data to obtain pelvic floor muscle function evaluation data.

[0141] The electromyography test data is input into the global program generation model for global prediction to obtain the global rehabilitation training program for the current subject, including:

[0142] S820: If the pelvic floor muscle function assessment data indicates that the pelvic floor muscles of the current subject need rehabilitation training, the electromyography test data is input into the global program generation model for global prediction to obtain a global rehabilitation training program for the current subject.

[0143] Among them, the pelvic floor muscle function assessment data is obtained through a comprehensive evaluation of the training test data of the entire training phase. It is used to evaluate the overall functional level of the pelvic floor muscles, identify whether there are problems such as fatigue, excessive tension, weakness or incoordination, and provide a basis for the formulation of global rehabilitation training plans.

[0144] Specifically, feature extraction can be performed on the electromyographic test data of the current object, and the similarity of the features of the electromyographic test data can be calculated and referenced, which can be the calculation of cosine similarity or Euclidean distance. When the similarity value is within a certain range, the pelvic floor muscle function assessment data is determined to be not requiring rehabilitation training. If it exceeds a certain range, the pelvic floor muscle function assessment data is determined to be requiring rehabilitation training. When the pelvic floor muscle function assessment data indicates that the pelvic floor muscles of the current object need rehabilitation training, the electromyographic test data is input into the pre-trained LSTM global solution generation model for global prediction to obtain the global rehabilitation training solution for the current object.

[0145] In the above embodiment, the functional state of the pelvic floor muscles of the current subject is first evaluated according to the electromyographic test data to obtain the pelvic floor muscle function evaluation data; secondly, if the pelvic floor muscle function evaluation data indicates that the pelvic floor muscles of the current subject need rehabilitation training, the electromyographic test data is input into the global program generation model for global prediction to obtain the global rehabilitation training program of the current subject. By evaluating the functional state of the pelvic floor muscles, a decision basis is intelligently provided for whether a global rehabilitation training program needs to be predicted.

[0146] See also Fig. 9 The embodiment of the present application further provides a home pelvic floor muscle rehabilitation training program intelligent generation system 900, and the home pelvic floor muscle rehabilitation training program intelligent generation system 900 includes:

[0147] The historical test data acquisition module 910 is used to acquire the historical test data of the pelvic floor muscles of the historical subject generated during the whole training phase; wherein the whole training phase is divided into a plurality of local training phases;

[0148] A global-local model training module 920 is used to perform model training based on the historical test data to obtain a global solution generation model and a local generation model set; wherein the global solution generation model is used to perform solution prediction for the entire training stage to obtain a global rehabilitation training plan, and the local generation model set includes a local solution generation model corresponding to each local training stage, and the local solution generation model is used to perform solution prediction for the corresponding local training stage prediction to obtain a targeted training plan;

[0149] The electromyographic test data acquisition module 930 is used to acquire the electromyographic test data generated by the pelvic floor muscles of the current subject during the whole training phase; and determine the target training phase that needs to be improved in the multiple local training phases according to the electromyographic test data;

[0150] A local model data selection module 940 is used to determine the local solution generation model corresponding to the target training stage in the local generation model set, and to determine the stage test data of the target training stage in the electromyography test data;

[0151] A local evaluation data determination module 950 is used to perform elastic response simulation on the pelvic floor muscle dynamics model according to the stage test data of the target training stage to obtain local stage evaluation data of the target training stage; wherein the local stage evaluation data is determined based on the elastic response simulation result of the pelvic floor muscle dynamics model;

[0152] The global-local solution generation module 960 is used to input the electromyography test data into the global solution generation model for global prediction to obtain the global rehabilitation training solution of the current subject; through the local solution generation model corresponding to the target training stage, local prediction is performed according to the stage test data and local stage evaluation data of the target training stage to obtain a targeted training solution suitable for the target training stage;

[0153] The local global program adjustment module 970 is used to adjust the global rehabilitation training program based on the targeted training program suitable for the target training stage to obtain a target rehabilitation training program suitable for the pelvic floor muscles of the current subject.

[0154] The further functional description of each of the above modules and units is the same as that of the above corresponding embodiments and will not be repeated here.

[0155] The intelligent generation system of the home pelvic floor muscle rehabilitation training program in this embodiment is presented in the form of functional units, where the units refer to ASIC (Application Specific Integrated Circuit) circuits, processors and memories that execute one or more software or fixed programs, and / or other devices that can provide the above functions.

[0156] The embodiment of the present application also provides a system for intelligently generating a home pelvic floor muscle rehabilitation training program, the system comprising a pelvic floor muscle stimulation feedback instrument, and a computer device that is communicatively connected to the pelvic floor muscle stimulation feedback instrument;

[0157] See also Fig.10 , Fig.10 It is a schematic diagram of the structure of the computer equipment, such as Fig.10As shown, the computer device includes: one or more processors 10, a memory 20, and interfaces for connecting various components, including high-speed interfaces and low-speed interfaces. Various components are connected to each other using different buses for communication, and can be installed on a common mainboard or installed in other ways as needed. The processor can process the instructions executed in the computer device, including instructions stored in or on the memory to display the graphical information of the GUI on an external input / output device (such as, a display device coupled to the interface). In some optional embodiments, if necessary, multiple processors and / or multiple buses can be used together with multiple memories and multiple memories. Similarly, multiple computer devices can be connected, and each device provides some necessary operations (for example, as a server array, a group of blade servers, or a multi-processor system). Fig.10 A processor 10 is taken as an example.

[0158] The processor 10 may be a central processing unit, a network processor or a combination thereof. The processor 10 may further include a hardware chip. The hardware chip may be a dedicated integrated circuit, a programmable logic device or a combination thereof. The programmable logic device may be a complex programmable logic device, a field programmable gate array, a general purpose array logic or any combination thereof.

[0159] The memory 20 stores instructions executable by at least one processor 10, so that the at least one processor 10 executes the method shown in the above embodiment.

[0160] The memory 20 may include a program storage area and a data storage area, wherein the program storage area may store an operating system, an application required for at least one function; the data storage area may store data created according to the use of the computer device, etc. In addition, the memory 20 may include a high-speed random access memory, and may also include a non-transient memory, such as at least one disk storage device, a flash memory device, or other non-transient solid-state storage device. In some optional embodiments, the memory 20 may optionally include a memory remotely arranged relative to the processor 10, and these remote memories may be connected to the computer device via a network. Examples of the above-mentioned network include, but are not limited to, the Internet, an intranet, a local area network, a mobile communication network, and combinations thereof.

[0161] The memory 20 may include a volatile memory, such as a random access memory; the memory may also include a non-volatile memory, such as a flash memory, a hard disk or a solid state drive; the memory 20 may also include a combination of the above types of memory.

[0162] The computer device further comprises a communication interface 30 for the computer device to communicate with other devices or a communication network.

[0163] The embodiment of the present application also provides a computer-readable storage medium. The above method according to the embodiment of the present application can be implemented in hardware, firmware, or can be implemented as a computer code that can be recorded in a storage medium, or can be implemented as a computer code that is originally stored in a remote storage medium or a non-temporary machine-readable storage medium and will be stored in a local storage medium through a network download, so that the method described herein can be stored in such software processing on a storage medium using a general-purpose computer, a dedicated processor, or programmable or dedicated hardware. Among them, the storage medium can be a magnetic disk, an optical disk, a read-only storage memory, a random access memory, a flash memory, a hard disk or a solid-state hard disk, etc.; further, the storage medium can also include a combination of the above types of memories. It can be understood that a computer, a processor, a microprocessor controller, or programmable hardware includes a storage component that can store or receive software or computer code. When the software or computer code is accessed and executed by a computer, a processor, or hardware, the method shown in the above embodiment is implemented.

[0164] The embodiment of the present application provides a computer program product, which includes computer instructions, which are stored in a computer-readable storage medium. A processor of a computer device reads the computer instructions from the computer-readable storage medium, and the processor executes the computer instructions, so that the computer device executes the method of any embodiment of the present application.

[0165] The systems, devices, modules or units described in the above embodiments may be implemented by computer chips or entities, or by products with certain functions. For the convenience of description, the above devices are described separately by functions in various units. Of course, when implementing the present application, the functions of each unit may be implemented in the same or multiple software and / or hardware.

[0166] Those skilled in the art will appreciate that the embodiments of the present application may be provided as methods, systems, or computer program products. Therefore, the present application may adopt the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware. Moreover, the present application may adopt the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program codes.

[0167] The present application is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems), and computer program products according to the embodiments of the present application. It should be understood that each process and / or box in the flowchart and / or block diagram, as well as the combination of the processes and / or boxes in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0168] It should also be noted that the terms "include", "comprises" or any other variations thereof are intended to cover non-exclusive inclusion, so that a process, method, commodity or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, commodity or device. In the absence of more restrictions, the elements defined by the sentence "comprises a ..." do not exclude the existence of other identical elements in the process, method, commodity or device including the elements.

[0169] The above is only an embodiment of the present application and is not intended to limit the present application. For those skilled in the art, the present application may have various changes and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application should be included in the scope of the claims of the present application.

[0170] Although the embodiments of the present application have been described in conjunction with the accompanying drawings, those skilled in the art may make various modifications and variations without departing from the spirit and scope of the present application, and such modifications and variations are all within the scope defined by the appended claims.

Claims

1. A method for intelligently generating a home pelvic floor muscle rehabilitation training program, characterized in that: The method comprises: Acquire historical test data of the pelvic floor muscles of the historical subject generated during the full training phase; wherein the full training phase is divided into a plurality of local training phases; Model training is performed based on the historical test data to obtain a global solution generation model and a local generation model set; wherein the global solution generation model is used to perform solution prediction for the entire training stage to obtain a global rehabilitation training plan, and the local generation model set includes a local solution generation model corresponding to each local training stage, and the local solution generation model is used to perform solution prediction for the corresponding local training stage prediction to obtain a targeted training plan; Acquire electromyographic test data of the pelvic floor muscles of the current subject generated during the entire training phase; and determine a target training phase that needs to be improved in the multiple local training phases according to the electromyographic test data; Determine the local solution generation model corresponding to the target training stage in the local generation model set, and determine the stage test data of the target training stage in the electromyography test data; Performing elastic response simulation on the pelvic floor muscle dynamics model according to the stage test data of the target training stage to obtain local stage evaluation data of the target training stage; wherein the local stage evaluation data is determined based on the elastic response simulation result of the pelvic floor muscle dynamics model; Input the electromyographic test data into the global solution generation model for global prediction to obtain the global rehabilitation training solution for the current subject; perform local prediction based on the stage test data and local stage evaluation data of the target training stage through the local solution generation model corresponding to the target training stage to obtain a targeted training solution suitable for the target training stage; The global rehabilitation training program is adjusted based on the targeted training program suitable for the target training stage to obtain a target rehabilitation training program suitable for the pelvic floor muscles of the current subject.

2. The method according to claim 1, characterized in that The local scheme generation model corresponding to the target training stage is used to perform local prediction according to the stage test data and the local stage evaluation data of the target training stage to obtain a targeted training scheme suitable for the target training stage, including: Generate a surface electromyogram of the target training phase based on the phase test data of the target training phase; The surface electromyogram, stage test data and local stage evaluation data of the target training stage are input into the local plan generation model for local prediction to obtain the targeted training plan.

3. The method according to claim 2, characterized in that The local solution generation model includes a first encoder and a first decoder; the surface electromyogram, stage test data and local stage evaluation data of the target training stage are input into the local solution generation model for local prediction to obtain the targeted training plan, including: Inputting the surface electromyogram, stage test data and local stage evaluation data of the target training stage into the first encoder for feature extraction to obtain associated feature data and time series feature data of the target training stage; splicing the associated feature data and the time series feature data of the target training stage to obtain spliced ​​feature data of the target training stage; The splicing feature data is input into the first decoder for decoding to obtain the targeted training plan.

4. The method according to claim 3, characterized in that The first encoder includes a convolutional neural network branch and a recurrent neural network branch connected in parallel; the surface electromyogram, stage test data and local stage evaluation data of the target training stage are input into the first encoder for feature extraction to obtain associated feature data and time series feature data of the target training stage, including: Extracting features from the surface electromyogram through the convolutional neural network branch to obtain associated feature data of the target training stage; Concatenating the stage test data and the local stage evaluation data of the target training stage to obtain the test evaluation joint data of the target training stage; The test evaluation joint data of the target training phase is subjected to feature extraction through the recurrent neural network branch to obtain the time series feature data of the target training phase.

5. The method according to claim 1, characterized in that: The multiple local training stages are respectively a pre-resting stage, a fast-twitch stage, a slow-twitch stage, an endurance test stage and a post-resting stage; the pre-resting stage, the fast-twitch stage, the slow-twitch stage, the endurance test stage and the post-resting stage correspond to respective local scheme generation models.

6. The method according to any one of claims 1 to 5, characterized in that: The full training stage includes multiple local training stages, and the electromyography test data includes the stage test data of each of the multiple local training stages; The target training phase is determined by: Performing a local performance evaluation on the pelvic floor muscles of the current subject according to the respective stage test data of the multiple local training stages, to obtain the respective local stage performance data of the multiple local training stages; Among the multiple local training stages, the target training stage whose local stage performance data satisfies a preset parameter adjustment condition is determined.

7. The method according to any one of claims 1 to 5, characterized in that: The global solution generation model includes an input embedding layer, a second encoder and a second decoder; the electromyographic test data is input into the global solution generation model for global prediction to obtain a global rehabilitation training solution for the current object, including: Performing vector representation on the electromyographic test data through the input embedding layer to obtain an electromyographic embedding vector corresponding to the electromyographic test data; Position encoding the electromyographic embedding vector is performed by the second encoder to obtain electromyographic feature data; wherein the electromyographic feature data is used to characterize long-term dependency information and global information in the electromyographic test data; The electromyographic characteristic data is decoded by the second decoder to obtain a global rehabilitation training program for the current object.

8. The method according to any one of claims 1 to 5, characterized in that: Before inputting the electromyographic test data into the global program generation model for global prediction to obtain the global rehabilitation training program for the current object, the method further includes: Performing a functional status assessment on the pelvic floor muscles of the current subject according to the electromyography test data to obtain pelvic floor muscle function assessment data; Inputting the electromyographic test data into the global program generation model for global prediction to obtain the global rehabilitation training program of the current subject includes: If the pelvic floor muscle function assessment data indicates that the pelvic floor muscles of the current subject need rehabilitation training, the electromyography test data is input into the global program generation model for global prediction to obtain a global rehabilitation training program for the current subject.

9. A home pelvic floor muscle rehabilitation training program intelligent generation system, characterized in that: The system comprises: A historical test data acquisition module, used to acquire historical test data generated by the pelvic floor muscles of a historical subject during a full training phase; wherein the full training phase is divided into a plurality of local training phases; A global-local model training module, used for performing model training based on the historical test data to obtain a global solution generation model and a local generation model set; wherein the global solution generation model is used for performing solution prediction for the entire training stage to obtain a global rehabilitation training plan, and the local generation model set includes a local solution generation model corresponding to each local training stage, and the local solution generation model is used for performing solution prediction for the corresponding local training stage prediction to obtain a targeted training plan; The electromyographic test data acquisition module is used to acquire the electromyographic test data generated by the pelvic floor muscles of the current subject during the whole training phase; and determine the target training phase that needs to be improved in the multiple local training phases according to the electromyographic test data; A local model data selection module, used to determine the local solution generation model corresponding to the target training stage in the local generation model set, and to determine the stage test data of the target training stage in the electromyography test data; A local evaluation data determination module, used to perform elastic response simulation on the pelvic floor muscle dynamics model according to the stage test data of the target training stage, to obtain local stage evaluation data of the target training stage; wherein the local stage evaluation data is determined based on the elastic response simulation result of the pelvic floor muscle dynamics model; A global-local solution generation module is used to input the electromyography test data into the global solution generation model for global prediction, so as to obtain the global rehabilitation training solution of the current object; through the local solution generation model corresponding to the target training stage, local prediction is performed according to the stage test data and local stage evaluation data of the target training stage, so as to obtain a targeted training solution suitable for the target training stage; The local global program adjustment module is used to adjust the global rehabilitation training program based on the targeted training program suitable for the target training stage to obtain a target rehabilitation training program for the pelvic floor muscles of the current object.

10. A home pelvic floor muscle rehabilitation training program intelligent generation system, characterized in that: The system includes a pelvic floor muscle stimulation feedback instrument, and a computer device that is communicatively connected to the pelvic floor muscle stimulation feedback instrument; The computer device includes a memory and a processor, the memory and the processor are communicatively connected to each other, the memory stores computer instructions, and the processor executes the method according to any one of claims 1 to 8 by executing the computer instructions.

Images