Intelligent course recommendation method and system for rehabilitation training

Abstract

The application relates to the field of rehabilitation training course recommendation, in particular to an intelligent course recommendation method and system for rehabilitation training. The method comprises the following steps: receiving user portrait information and original electromyography data; evaluating a pelvic floor muscle evaluation score, a force condition score and a muscle dysfunction severity of a user by using a preset algorithm; matching courses in a course database with the information and the user portrait information to obtain a matching score in each dimension; calculating a total matching degree value by weighting and generating a course recommendation list output; and determining information such as a user advantage item, a disadvantage item, a stage priority order and an improvement urgency, etc. The application achieves the technical effects of accurately evaluating the pelvic floor muscle condition and the training quality of a user and providing personalized rehabilitation training course recommendation for the user.

Description

Technical Field

[0001] This application relates to the field of rehabilitation training course recommendation, and in particular to an intelligent course recommendation method and system for rehabilitation training. Background Technology

[0002] Currently, the development and recommendation of rehabilitation training programs for pelvic floor muscles mainly rely on the clinical experience of therapists or traditional methods such as simple questionnaires. However, these methods have significant technical limitations and cannot meet the needs of precise and personalized rehabilitation training. First, rehabilitation assessments heavily rely on subjective human judgment, failing to quantify the user's pelvic floor muscle function based on objective physiological signals. This results in low reliability of assessment results, unable to provide dependable data support for rehabilitation course recommendations. Second, the electrophysiological signals of the user's pelvic floor muscles are continuous, high-dimensional, and temporal, differing in source and format from other basic personal information. Existing technologies struggle to effectively correlate and deeply integrate multi-source heterogeneous data, leading to a single-dimensional rehabilitation assessment and insufficient information utilization. Furthermore, existing recommendation schemes often employ simple rule matching, classifying rehabilitation courses into beginner, intermediate, and advanced levels solely based on total questionnaire scores or single indicators. This fails to provide refined matching based on multi-dimensional assessment results, easily leading to a mismatch between course difficulty and the user's actual abilities. This not only affects rehabilitation training effectiveness but may also pose rehabilitation safety risks. Finally, during the development of rehabilitation plans, therapists typically need to manually compare and analyze large amounts of user data and rehabilitation course items. The human-computer interaction process is cumbersome, with low automation, time-consuming, labor-intensive, and prone to human error, making it difficult to achieve efficient and scalable rehabilitation course recommendation services.

[0003] Therefore, existing methods for recommending pelvic floor muscle rehabilitation courses suffer from problems such as strong subjectivity, insufficient data utilization, low personalization, and low efficiency. There is an urgent need for a technical solution that can automatically, accurately, and efficiently process multi-source heterogeneous data and provide users with personalized pelvic floor muscle rehabilitation course recommendations to overcome the above-mentioned shortcomings of existing technologies. Summary of the Invention

[0004] The purpose of this application is to overcome the above-mentioned technical problems and provide an intelligent course recommendation method and system for rehabilitation training, which can accurately match the most suitable training courses for users based on their personal profiles and Glazer assessment data of pelvic floor muscles.

[0005] Firstly, this application provides an intelligent course recommendation method for rehabilitation training, employing the following technical solution: A smart curriculum recommendation method for rehabilitation training includes the following steps: Receive user profile information and raw electromyography (EMG) data, wherein the raw EMG data includes all EMG signal values ​​of the user in multiple different muscle state stages; Based on the preset Kegel assessment algorithm, the user's muscle ability is assessed in combination with the original electromyography data to obtain the user's pelvic floor muscle assessment score, including the total assessment score and the stage assessment score in each muscle state stage. Based on the preset abdominal muscle involvement correlation scoring algorithm, combined with the original electromyography data, the electromyographic signal relationship between the user's abdominal muscles and pelvic floor muscles is analyzed to evaluate the user's pelvic floor muscle training quality and obtain a force exertion score. The severity of the user's muscle dysfunction is determined based on the total assessment score. Each course in the course database is matched with the user profile information, the pelvic floor muscle assessment score, the severity, and the exertion score to obtain a matching score for each dimension. The matching scores of each dimension are weighted to obtain the total matching score. A course recommendation list is generated and output based on the total matching score and the corresponding course name.

[0006] By adopting the above technical solution, by receiving user profile information and multi-stage raw electromyography data, the Kegel assessment algorithm and abdominal muscle involvement correlation algorithm are used to quantitatively assess the user's muscle ability and training quality. Based on the assessment results, courses are matched and calculated from multiple dimensions, and then a total matching value is obtained through weighted fusion to generate a course recommendation list. This enables intelligent and personalized recommendations for pelvic floor muscle rehabilitation training, improving the accuracy of course and user matching and recommendation efficiency.

[0007] Preferably, the muscle state phases include a pre-resting phase, a fast-twitch fiber phase, a slow-twitch fiber phase, an endurance test phase, and a post-resting phase. Each muscle state stage includes several stage indicators, and each stage indicator corresponds to one electromyographic signal value; The phase indicators for both the pre-resting phase and the post-resting phase include average and variability. The phase indicators for the fast-twitch muscle fiber phase include maximum value, rise time, and recovery time. The phase indicators for the slow-twitch muscle fiber phase include average, rise time, recovery time, and variability. The phase indicators for the endurance test phase include average, variability, and the last 10 seconds ratio.

[0008] By adopting the above technical solutions, the stages of muscle condition and the stage indicators and corresponding electromyographic signal values ​​of each stage are clearly defined, and a complete and detailed pelvic floor muscle assessment system is constructed. This allows electromyographic signals to be analyzed and quantified from multiple physiological dimensions, enabling more comprehensive and detailed acquisition of users' original electromyographic data. This provides a rich and accurate data foundation for subsequent assessment of users' muscle capabilities based on preset algorithms and analysis of the electromyographic signal relationship between abdominal muscles and pelvic floor muscles, thereby improving the accuracy and personalization of rehabilitation course recommendations.

[0009] Preferably, the step of assessing the user's muscle strength based on a preset Kegel assessment algorithm and the raw electromyography data to obtain the user's pelvic floor muscle assessment score specifically includes the following steps: Based on the measured electromyographic signal values ​​of the user in each muscle state stage, the indicators are quantified by combining the preset mathematical model. Each mathematical model corresponds to a scoring type and a quantification calculation strategy. The mathematical models include a decreasing first model, an increasing second model, and a centrally symmetrical third model. Each stage indicator corresponds to a scoring type, and the quantification calculation strategy of the stage indicator is determined based on the scoring type; The rise time, recovery time, variability, and the average values ​​of the pre-resting phase and the post-resting phase correspond to the scoring type of the first model; the maximum value of the fast-twitch fiber phase, the average value of the slow-twitch fiber phase, and the average value of the endurance test phase correspond to the scoring type of the second model; and the ratio of the last 10 seconds of the endurance test phase corresponds to the scoring type of the third model.

[0010] By adopting the above technical solution, based on the changing characteristics of indicators at different stages, three mathematical models—decreasing, increasing, and centrally symmetrical—are used to quantify the indicators. This allows different types of indicators to adopt quantification strategies that match their changing patterns, avoiding scoring bias caused by a single model, improving the objectivity and accuracy of indicator scores, and making the evaluation results more consistent with the actual muscle function state.

[0011] Preferably, the electromyographic signal value of the current stage indicator is quantized according to the quantization calculation strategy, specifically including the following steps: Obtain the passing score and full score for each of the aforementioned stage indicators; For the stage indicators corresponding to the first model, the quantification calculation strategy is to normalize the electromyographic signal value based on the passing standard value and the full score standard value. When the electromyographic signal value is not greater than the full score standard value, the full score is output; otherwise, the attenuation is calculated based on the square root of the normalization deviation. For the stage indicators corresponding to the second model, the quantification operation strategy is to normalize and segment the electromyographic signal value based on the passing standard value and the full score standard value. When the electromyographic signal value is not less than the full score standard value, the full score is output. When it is less than the passing threshold, it is reduced by linear proportion. Otherwise, it is incrementally mapped according to the square root of the normalization deviation. For the stage index corresponding to the third model, the quantization calculation strategy is to calculate the absolute deviation of the electromyographic signal value relative to the preset optimal value, determine the passing deviation with the optimal value, perform normalization processing based on the absolute deviation and the passing deviation, output full score when the absolute deviation is zero, and perform symmetrical attenuation calculation based on the square root of the normalized deviation when the absolute deviation is greater than zero.

[0012] By adopting the above technical solutions, standardized scoring calculations can be performed on indicators of different models based on different types of mathematical models, as well as standard values, full score standards, or preset optimal values. This can uniformly convert electromyographic signals of different magnitudes and meanings into standard scores within standard ranges, ensuring consistent scoring ranges and reasonable trends. This provides reliable data support for subsequent course matching and improves the stability and comparability of the overall evaluation algorithm.

[0013] Preferably, the method further includes the following steps: The scores after quantitative calculation are used as the index scores of each stage indicator. The scores of each indicator are weighted according to the preset index weights within the corresponding muscle state stage to obtain the stage evaluation score of each muscle state stage. The user's total evaluation score is obtained by weighting the evaluation scores of each stage according to the preset stage weights, and the stage evaluation scores and the total evaluation score are combined into a pelvic floor muscle evaluation score.

[0014] By adopting the above technical solution, the quantitatively calculated score is used as the indicator score for each stage. Then, the stage assessment score is calculated according to the preset indicator weights within the muscle state stage. Finally, the total assessment score is calculated according to the preset stage weights and summarized as the pelvic floor muscle assessment score. This can comprehensively consider the impact of different stage indicators and muscle state stages on pelvic floor muscle assessment, reflect the importance of different indicators and different stages in overall function, make the assessment results more representative and scientific, and provide a reliable quantitative basis for subsequent course matching, severity judgment, and priority ranking. This solves the problem that existing rehabilitation assessments rely on subjective human judgment and have low reliability.

[0015] Preferably, the step of analyzing the electromyographic signal relationship between the user's abdominal muscles and pelvic floor muscles based on a preset abdominal muscle involvement correlation scoring algorithm and the original electromyographic data, and evaluating the user's pelvic floor muscle training quality to obtain a force exertion score, specifically includes the following steps: Based on the abdominal muscle EMG signal and pelvic floor muscle EMG signal in the original electromyography data, the ratio of the EMG signal between the abdominal muscle and pelvic floor muscle of the user, as well as the Pearson correlation coefficient between the abdominal muscle EMG signal sequence and the pelvic floor muscle EMG signal sequence, are calculated. The independent force exertion ability standard score is determined from the preset independent force exertion ability scoring mapping table based on the absolute value of the Pearson correlation coefficient, and the abdominal muscle control ability standard score is determined from the preset abdominal muscle control ability scoring mapping table based on the EMG signal ratio. The score for force exertion is obtained by weighting the independent force exertion ability standard score and the abdominal muscle control ability standard score with preset correlation weights.

[0016] By adopting the above technical solution, by calculating the EMG signal ratio of abdominal muscles and pelvic floor muscles and the Pearson correlation coefficient, and combining it with a preset mapping table to obtain scores for independent force exertion ability and abdominal muscle control ability, and then weighting and fusing them, it is possible to accurately identify problems of abdominal muscle compensation and force exertion incoordination, improve the professionalism of pelvic floor muscle training quality assessment, enhance training safety and rehabilitation effectiveness, and overcome the problems of strong subjectivity and insufficient data utilization in existing rehabilitation course recommendations.

[0017] Preferably, each course in the course database is matched with the pelvic floor muscle assessment score, the severity score, and the exertion score to obtain a matching score for each dimension, specifically including the following steps: Iterate through each course in the course database to obtain the suggested score range for each course and the dimension weights of each dimension involved in the matching. For the pelvic floor muscle assessment dimension, the course suggested score range, the stage assessment score and the total assessment score are combined to perform stage weighted matching and total assessment score matching. The scores are calculated according to the score range comparison or the preset deviation attenuation rule and then weighted and fused to obtain the pelvic floor muscle assessment matching score. Regarding the force exertion dimension, the user's muscle strength level and course difficulty are classified into different levels, and the corresponding force exertion matching score is output through a preset matching matrix. For the severity dimension, the severity level of the user's muscle dysfunction is determined based on the total assessment score. The severity level is then matched with the graded course difficulty level, and the severity matching score of the currently traversed course is assigned according to the misalignment level difference.

[0018] By adopting the above technical solutions, the matching scores are calculated from multiple dimensions such as pelvic floor muscle assessment, force exertion, and severity. By using segmented weighted matching, level matrix matching, and misaligned level difference assignment, the course recommendations can be fully matched with the user's functional level, degree of impairment, and training ability, significantly improving the rationality and relevance of the recommendations.

[0019] Preferably, each course in the course database is matched with the user profile information to obtain a matching score for each dimension, specifically including the following steps: The user profile information includes basic information, chief complaints and symptoms, and special physiological conditions. The basic information includes the user's age. For the user age dimension, the user age is matched with the preset course applicable age range, and a stepped user age matching score is output according to the deviation range obtained from the matching. For the chief complaint symptom dimension, the matching results between the course algorithm standard and the course designer standard and the chief complaint symptom are calculated based on the natural language processing algorithm, and the chief complaint symptom matching score is obtained by weighting according to the preset scoring system. For the special physiological condition dimension, based on the preset base score, points are added item by item according to the special physiological condition to obtain the special physiological condition matching score.

[0020] By adopting the above technical solutions, and by performing tiered matching, natural language processing matching, and conditional scoring matching on age, chief symptoms, and special physiological conditions, the recommendation algorithm fully integrates user basic information, symptom needs, and physiological status. This makes the recommendation results not only fit the functional data but also conform to the individual characteristics of the user, further improving the applicability and rehabilitation orientation of the recommendations.

[0021] Preferably, after obtaining the index scores for each of the aforementioned stage indicators, the method further includes the following steps: The weight type of the muscle state stage to which each stage indicator belongs is determined. The weight type includes high weight, medium weight and low weight. Different weight types correspond to different indicator evaluation levels. The indicator evaluation level corresponding to each stage indicator of the user is determined based on the indicator score. The advantages and disadvantages of each stage indicator are determined by combining the differences in weight types. The user's strengths and weaknesses are determined. A stage priority score is calculated based on the stage evaluation score and the stage weight, and the stage priority order is determined based on the stage priority score. The urgency of improvement is determined by combining the user's strengths, weaknesses, and stage priority order, generating priority tags and targeted recommendation reasons, and summarizing them into the generated course recommendation list.

[0022] By adopting the above technical solutions, and by differentiating advantages and disadvantages according to stage weights, and by combining advantages and disadvantages analysis with priority ranking, we can accurately identify users' weaknesses and the urgency of improvement. This makes the recommendation reasons no longer just simple numerical matching, but precise attribution and targeted improvement explanations based on the user's actual functional status. This makes the recommendation process explainable and clear in direction, and improves user understanding and rehabilitation compliance.

[0023] Secondly, this application provides an intelligent course recommendation system for rehabilitation training, which adopts the following technical solution: An intelligent course recommendation system for rehabilitation training includes the following modules: The data input module is used to receive the user's user profile information and raw electromyography (EMG) data, which includes all EMG signal values ​​of the user in multiple different muscle state stages. The Kegel assessment module is used to assess the user's muscle ability based on a preset Kegel assessment algorithm and the original electromyography data to obtain the user's pelvic floor muscle assessment score, including the total assessment score and the stage assessment score at each stage of the muscle state. The abdominal muscle correlation assessment module is used to analyze the electromyographic signal relationship between the user's abdominal muscles and pelvic floor muscles based on a preset abdominal muscle participation correlation scoring algorithm and the original electromyographic data, to assess the user's pelvic floor muscle training quality and obtain a force exertion score. The multidimensional intelligent matching module is used to determine the severity of the user's muscle dysfunction based on the total assessment score, and to match each course in the course database with the user profile information, the pelvic floor muscle assessment score, the severity, and the force exertion score to obtain a matching score for each dimension. The recommendation result output module is used to calculate the total matching score by weighting the matching scores of each dimension, generate a course recommendation list based on the total matching score and the corresponding course name, and output it.

[0024] By adopting the above technical solutions, a complete system is constructed through five modules: data input, Kegel assessment, abdominal muscle correlation assessment, multidimensional intelligent matching, and recommendation result output. This system achieves fully automated operation from electromyography data acquisition and functional assessment to course recommendation, improving the overall intelligence level and operational stability of the system and facilitating its implementation and application.

[0025] In summary, this application includes at least one of the following beneficial technical effects: (1) This invention achieves quantitative analysis of the user's pelvic floor muscle function status based on objective physiological signals through a multi-stage, multi-model quantitative scoring system and abdominal muscle exertion correlation analysis. It can comprehensively and accurately assess the user's pelvic floor muscle function and training quality, construct a standardized and objective physiological assessment mechanism, improve the credibility of the assessment results, and provide reliable data support for rehabilitation course recommendations. (2) This application adopts a multi-dimensional weighted matching and grade misalignment penalty mechanism to accurately match the user's physiological state, symptoms, severity and course attributes, realize the effective association and deep integration of multi-source heterogeneous data, and perform refined matching based on multi-dimensional evaluation results to avoid the problem of course difficulty not matching the user's actual ability, realize highly personalized rehabilitation course recommendations, and improve training safety, effectiveness and rehabilitation efficiency. (3) This application generates explainable recommendation reasons by determining advantages and disadvantages and prioritizing, making the recommendation process transparent and the results understandable, improving the system's intelligence level and user experience, and is conducive to the continuous optimization of long-term rehabilitation management and training programs. Attached Figure Description

[0026] Figure 1 This is a flowchart of the method in an embodiment of this application; Figure 2 This is an architecture diagram of the system in an embodiment of this application. Detailed Implementation

[0027] This application provides a method and system for recommending intelligent courses for rehabilitation training. To make the purpose, technical solution and advantages of this application clearer, the implementation method of this application will be further described in detail below.

[0028] The following describes in further detail an embodiment of an intelligent curriculum recommendation method for rehabilitation training according to this application, with reference to the accompanying drawings.

[0029] This application presents an intelligent course recommendation method for rehabilitation training, aiming to provide users with personalized pelvic floor muscle rehabilitation training programs based on scientific assessment data. The method integrates the user's basic information (age, medical history, symptoms) with Glazer professional assessment data, employing a multi-dimensional weighted matching algorithm to calculate the compatibility between the user and various courses in the course library, thereby outputting an accurate recommendation list.

[0030] The process of this application is as follows: Figure 1 As shown, it includes the following steps: S1. Receive user profile information and raw electromyography (EMG) data. The raw EMG data includes all EMG signal values ​​of the user in multiple different muscle state stages. The user profile information includes basic information, chief complaint symptoms, and special physiological conditions. In this embodiment, the basic information includes the user's age. In other possible implementations, the basic information may also include gender, height, weight, etc.

[0031] Specifically, the muscle state stages in this embodiment include the pre-resting stage, the fast-twitch fiber stage, the slow-twitch fiber stage, the endurance test stage, and the post-resting stage.

[0032] Each muscle state stage includes several stage indicators, and each stage indicator corresponds to an electromyographic signal value.

[0033] In this embodiment, the phase indicators for both the pre-resting and post-resting phases include the mean and variability; the phase indicators for the fast-twitch muscle fiber phase include the maximum value, rise time, and recovery time; the phase indicators for the slow-twitch muscle fiber phase include the mean, rise time, recovery time, and variability; the phase indicators for the endurance test phase include the mean, variability, and the last 10-second ratio. The last 10-second ratio of the endurance test is the ratio of the average electromyographic signal within 10 seconds immediately after the endurance contraction ends to the average value at the pre-resting phase, which is used to assess the fatigue resistance of slow-twitch muscle fibers.

[0034] S2. Based on the preset Kegel assessment algorithm and combined with the original electromyography data, assess the user's muscle strength to obtain the user's pelvic floor muscle assessment score, including the total assessment score and the stage assessment score for each muscle state stage. Specifically, this includes the following steps: S21. Based on the measured electromyographic signal values ​​of the user in each muscle state stage, the indicators are quantified using a preset mathematical model: Each mathematical model corresponds to a scoring type and a quantification strategy. The mathematical models include a decreasing first model, an increasing second model, and a centrally symmetrical third model.

[0035] Each stage indicator corresponds to a scoring type, and the quantitative calculation strategy for the stage indicator is determined based on the scoring type. Among them, rise time, recovery time, variability, and the average values ​​of the pre-resting and post-resting phases correspond to the scoring type of the first model; the maximum value of the fast-twitch fiber phase, the average value of the slow-twitch fiber phase, and the average value of the endurance test phase correspond to the scoring type of the second model; and the ratio of the last 10 seconds of the endurance test phase corresponds to the scoring type of the third model.

[0036] For example, the scoring criteria for the endurance test phase are configured as shown in the table below: Table 1. Endurance Test Stage Scoring Criteria Configuration Table Indicator Name Indicator weights Rating type Full marks standard Passing standard average value 65% The bigger the better 40 μV 30 μV Variability 10% The smaller the better 0 0.2 ratio of the last 10 seconds 25% Gaussian distribution (close to 1) 1.0 0.8 ~ 1.2 In one specific implementation method, the electromyographic signal value of the current stage indicator is quantized according to the quantization calculation strategy, which specifically includes the following steps:

[0037] S211. Obtain the passing standard value and the full score standard value corresponding to each stage indicator from the pre-configured scoring standard configuration table for each stage.

[0038] In this embodiment, the scoring criteria configuration tables for other stages are shown in Tables 2 to 5: Table 2. Scoring Criteria Configuration for the Pre-Rest Phase Indicator Name Indicator weights Rating type Full marks standard Passing standard average value 55% The smaller the better 0 μV 4 μV Variability 45% The smaller the better 0 0.2 Table 3. Scoring Criteria for Fast-Twitch Fiber Stages Indicator Name Indicator weights Rating type Full marks standard Passing standard Maximum value 40% The bigger the better 60 μV 40 μV Ascent Time 30% The smaller the better 0.1 s 0.5 s Recovery time 30% The smaller the better 0.1 s 0.5 s Table 4. Slow-twitch muscle fiber phase scoring criteria. Indicator Name Indicator weights Rating type Full marks standard Passing standard average value 60% The bigger the better 45 μV 35 μV Ascent Time 10% The smaller the better 0.2 s 1.0 s Recovery time 10% The smaller the better 0.2 s 1.0 s Variability 20% The smaller the better 0 0.2 Table 5. Scoring Criteria for the Post-Rest Phase Indicator Name Indicator weights Rating type Full marks standard Passing standard average value 70% The smaller the better 0 μV 4 μV Variability 30% The smaller the better 0 0.2

[0039] S212. In a specific implementation, the scoring curve of the first model with decreasing characteristics is better as it is smaller, and an exponential decay function is used, with the score dropping rapidly after exceeding the passing score.

[0040] Furthermore, for the stage indicators corresponding to the first model, the quantization operation strategy is to normalize the electromyographic signal values ​​based on the passing standard value and the full score standard value. When the electromyographic signal value is not greater than the full score standard value, the full score is output; otherwise, the attenuation is calculated based on the square root of the normalization deviation.

[0041] In this embodiment, the quantization score of the first model is calculated as follows: If the electromyography signal value is less than or equal to the full score standard value, the quantitative score is 100 points. If the electromyography signal value is greater than the full score standard value, normalization processing is performed for calculation: R1 = (EMG signal value - full score standard value) / (passing standard value - full score standard value); Therefore, the quantitative score = 100 - 40 × (R1) 2 ); It should be noted that the quantitative score is limited to the range of 0-100.

[0042] S213. In a specific implementation, the scoring curve of the incremental second model is better the larger it is, using a logarithmic growth function (above the passing grade) and a linear decay function (below the passing grade). Furthermore, for the stage indicators corresponding to the second model, the quantization strategy is as follows: based on the passing standard value and the full score standard value, the electromyographic signal value is normalized and segmented and mapped. When the electromyographic signal value is not less than the full score standard value, the full score is output. When it is less than the passing threshold, it is reduced by linear proportion. Otherwise, it is incrementally mapped according to the square root of the normalization deviation.

[0043] In this embodiment, the second model quantization score is calculated as follows: If the electromyographic signal value is greater than or equal to the full score standard value, then the quantitative score is 100 points; If the passing standard value ≤ electromyographic signal value < full score standard value, normalization processing is performed for calculation: R2 = (EMG signal value - passing standard value) / (full score standard value - passing standard value); Therefore, the quantitative score = 60 + 40 × (R²) 0.5 ); If the electromyography signal value is less than the passing standard value, normalization processing is performed for calculation: R2 = Electromyographic signal value / Passing standard value; Therefore, the quantitative score = 60 × R2; similarly, the quantitative score is limited to the range of 0-100.

[0044] If the electromyography (EMG) signal value is less than the passing standard value, normalization processing is performed and the following calculation is performed: R = (passing standard value - EMG signal value) / (passing standard value - 0); S214. In a specific implementation, the centrally symmetrical third model score curve exhibits a Gaussian distribution, centered on the optimal value (1.0), and decays towards both sides.

[0045] Furthermore, for the stage indicators corresponding to the third model, the quantification operation strategy is to calculate the absolute deviation of the electromyographic signal value relative to the preset optimal value, determine the passing deviation with the optimal value, and perform normalization processing based on the absolute deviation and the passing deviation. When the absolute deviation is zero, the full score is output; when it is greater than zero, symmetrical attenuation calculation is performed based on the square root of the normalized deviation.

[0046] In this embodiment, the quantization score of the third model is calculated as follows: Absolute deviation = |EMG signal value - optimal value|; Passing deviation = |Passing threshold value - Optimal value|; If the absolute deviation is 0, the quantitative score is 100 points; If the absolute deviation is >0, normalization is performed for calculation: R3 = absolute deviation / passing deviation; Therefore, the quantitative score = 100 - 40 × (R³) 2 Similarly, the quantitative score is limited to the range of 0-100.

[0047] S22. The scores after quantitative calculation are used as the indicator scores of each stage. The scores of each indicator are weighted according to the preset indicator weights within the corresponding muscle state stage to obtain the stage evaluation score of each muscle state stage.

[0048] In this embodiment, the preset indicator weights within each stage are shown in Tables 1 to 5.

[0049] S23. The evaluation scores of each stage are weighted according to the preset stage weights to obtain the user's total evaluation score. The stage evaluation scores and the total evaluation score are combined to form the pelvic floor muscle evaluation score.

[0050] Right now, ; in, For the first The stage assessment score (0-100 points) for each stage. For the first The stage weights configured for each stage.

[0051] In this embodiment, the weight configuration for each stage is shown in Table 6 below: Phase Name Weight (Weight_i) Function Description Pre-Baseline 10% Assess muscle baseline tension and resting stability Fast-twitch fibers 30% Assess rapid contraction ability and immediate response capability. Slow-twitch muscle fibers (Tonic) 30% Assess sustained contraction capacity and muscle endurance Endurance test 20% Assess long-duration work capacity and fatigue resistance Post-Baseline 10% Assess post-training recovery ability The above steps are used to standardize the scoring of Glazer assessment data for the pelvic floor muscles. The assessment is divided into five stages, each containing several key internal indicators. The method uses a specific mathematical model to convert the measurements into scores from 0 to 100, and calculates the assessment scores for each stage and the total assessment score according to preset weights.

[0052] S3. Based on the preset abdominal muscle involvement correlation scoring algorithm and combined with the original electromyography (EMG) data, analyze the EMG signal relationship between the user's abdominal muscles and pelvic floor muscles, evaluate the user's pelvic floor muscle training quality, and obtain a force exertion score. This includes the following steps: S31. Calculate the Pearson correlation coefficient between the abdominal muscle EMG signal sequence and the pelvic floor muscle EMG signal sequence based on the original electromyography data, so as to measure the degree of linear correlation between the two.

[0053] The calculation method is as follows: ; in, This is a real-time abdominal muscle EMG signal sequence. This is a real-time pelvic floor muscle EMG signal sequence; , These are the average values ​​of the two sets of signals, respectively.

[0054] During pelvic floor muscle training, the abdominal muscles should remain relaxed while the pelvic floor muscles contract. A value close to 1 indicates highly synchronized contraction of the abdominal and pelvic floor muscles, signifying severe "synergistic abdominal muscle contraction" or "compensation," which should be avoided. A value close to 0 indicates that the two are unrelated, and the pelvic floor muscles are working independently, which is the ideal state.

[0055] S32. Based on the abdominal muscle EMG signal and pelvic floor muscle EMG signal in the original electromyography data, calculate the ratio of the EMG signal between the user's abdominal muscles and pelvic floor muscles, thereby measuring the ratio of the strength of the abdominal muscles to the strength of the pelvic floor muscles.

[0056] The calculation method is as follows: ; in, The average amplitude of the abdominal muscle EMG signal. The average amplitude of the pelvic floor muscle EMG signal.

[0057] The smaller the value, the less abdominal muscle activity is compared to pelvic floor muscle activity, indicating that the abdominal muscles are in a relaxed state. The larger the value (>1), the more the abdominal muscles are engaged than the pelvic floor muscles, which is a typical incorrect force exertion pattern.

[0058] S33. In order to convert the above-mentioned original physical quantities into readable 0-1 fractions, a piecewise mapping method is used for standardization.

[0059] S331. Determine the standard score of independent exertion ability from the preset independent exertion ability scoring mapping table based on the absolute value of the Pearson correlation coefficient. The lower the correlation, the higher the score.

[0060] In one specific implementation method, the independent force exertion capability scoring mapping table is shown in Table 7 below: Table 7 Independent Force Exertion Ability Scoring Mapping Table absolute value of the correlation coefficient (|r|) Standardized score (S_corr) Clinical significance 0.00 - 0.15 1.00 Excellent: Extremely strong independent control capability, completely independent 0.15 - 0.35 0.85 Good: Strong independent control ability 0.35 - 0.55 0.65 Medium: Some degree of correlation exists, but it is controllable. 0.55 - 0.75 0.45 Pass: Clear correlation, attention required. 0.75 - 0.90 0.25 Needs improvement: Strong muscle coordination, severe abdominal muscle compensation. >0.90 0.10 Very poor: Relying entirely on abdominal muscles for power

[0061] S332. Determine the standard score of abdominal muscle control ability from the preset abdominal muscle control ability scoring mapping table based on the EMG signal ratio. The lower the ratio, the higher the score.

[0062] In one specific implementation method, the abdominal muscle control ability scoring mapping table is shown in Table 8 below: Table 8. Abdominal Muscle Control Ability Scoring Mapping Table EMG ratio (abdominal muscles / pelvic floor muscles) Standardized score (S_emg) Clinical significance 0.00 - 0.25 1.00 Excellent: Abdominal muscles are almost silent 0.25 - 0.45 0.80 Good: Slight abdominal muscle activity 0.45 - 0.65 0.60 Moderate: Abdominal muscle involvement is average 0.65 - 0.85 0.40 Pass: High level of abdominal muscle involvement 0.85 - 1.00 0.20 Needs improvement: Severe abdominal muscle interference >1.00 0.05 Extremely poor: Abdominal muscles exert more force than pelvic floor muscles

[0063] S34. The score for force exertion is obtained by weighting the standard scores for independent force exertion ability and abdominal muscle control ability using preset correlation weights.

[0064] In one specific implementation, the weight of the independent force exertion ability standard score is set to 70%, and the weight of the abdominal muscle control ability standard score is set to 30%.

[0065] S35. Map the force exertion score to the corresponding level according to the preset scoring level system to obtain the user's muscle strength level.

[0066] S36. In another specific implementation, in addition to the overall score, each stage of training is analyzed independently to capture the abdominal muscle compensation characteristics of users under different loads.

[0067] The degree of abdominal muscle involvement at different stages typically follows the following pattern: Pre- and post-resting phases: The correlation with EMG values ​​is usually the lowest. If the value is high in this phase, it suggests that relaxation is not possible or that baseline tension is too high. Fast-twitch phase: Due to the rapid movement, abdominal muscles are easily reflexively contracted, and the correlation in this phase is usually higher than that in the resting phase.

[0068] Endurance / Slow-Motion Phase: As pelvic floor muscle fatigue increases, users may unconsciously increase abdominal muscle exertion to maintain pressure, leading to an increase in correlation and ratio over time.

[0069] The above steps aim to quantitatively assess the quality of pelvic floor muscle training. The core is to analyze the electromyographic (EMG) relationship between the abdominal and pelvic floor muscles to determine whether the user can "independently contract the pelvic floor muscles" and "reduce abdominal muscle compensation." The method combines Pearson correlation coefficient and EMG ratio to generate a comprehensive score ranging from 0% to 100%. Through phased, multi-dimensional quantitative scoring, it can accurately pinpoint the user's exertion problems, thus providing a scientific basis for developing personalized training programs.

[0070] S4. Based on the total assessment score, determine the severity of the user's muscle dysfunction. Match each course in the course database with the user profile information, pelvic floor muscle assessment score, severity, and exertion score to obtain a matching score for each dimension. The specific steps include the following: S41. Iterate through each course in the course database and obtain the suggested score range for each course and the dimensional weights of each dimension involved in the matching.

[0071] The dimension weights for each matching dimension are configured as shown in Table 9 below: Table 9. Dimension Weight Configuration Table for Each Matching Dimension Matching Dimensions Dimension weights Design Intent Evaluation score matching (Score) 30% Objective physiological indicators are used to ensure that the difficulty of the course matches the user's current muscle strength. Symptom matching 20% Responding to user complaints, we address specific issues such as urinary incontinence and pain. Force scoring match 20% The difficulty of the course is matched based on muscle contraction capacity (maximum fast-twitch muscle strength / average slow-twitch muscle strength). Age matching 10% Differentiate the physiological characteristics of different age groups (such as postpartum recovery vs. anti-aging in middle-aged and elderly people). Severity matching 10% Beginner / intermediate / advanced courses are recommended based on the severity of the muscle disorder (mild / moderate / severe). Special condition matching 10% Consider contraindications or strongly related factors such as surgical history and reproductive history.

[0072] S42. For the pelvic floor muscle assessment dimension, the pelvic floor muscle assessment matching score is obtained by combining the course-suggested score range, stage assessment score and total assessment score. The score is calculated according to the score range comparison or the preset deviation attenuation rule and then weighted and fused to obtain the pelvic floor muscle assessment matching score.

[0073] Specifically, the phased matching method involves comparing the user's five-stage evaluation scores with the suggested course score range. If the evaluation score falls within the suggested course score range, the single-stage matching score is full (1.0). If the evaluation score is not within the suggested course score range, the score deviation value is calculated. If the score deviation value is less than or equal to the first preset deviation threshold (10 points), the single-stage matching score is output as the first preset score (0.8). Otherwise, it is calculated using a linear decay method: single-stage matching score = max(0, 1 - score deviation value / 40). The phase matching score is obtained by weighting the single-stage matching scores of the five stages according to the stage weights of each stage.

[0074] The method for matching the total assessment score is to calculate the degree of match between the total assessment score and the suggested score range for the course: If the total assessment score is within the course's suggested score range, the total score matching score is the full score (1.0); if the assessment score is not within the course's range, the score deviation value is calculated. If the score deviation value is ≤ the second preset deviation threshold (15 points), the total score matching score is output as the second preset score (0.8); otherwise, the total score matching score = max(0, 1 - score deviation value / 50).

[0075] Finally, the pelvic floor muscle assessment matching score is calculated by weighting the stage matching score and the total matching score with their corresponding weight coefficients. In this embodiment, the weight coefficient of the stage matching score is set to 0.8, and the weight coefficient of the total matching score is set to 0.2. All of the above preset thresholds can be adjusted.

[0076] S43. For the force exertion dimension, the user's muscle strength level and course difficulty obtained in step S35 are divided into levels, and the corresponding force exertion matching score is output through the preset matching matrix.

[0077] In this embodiment, the preset course difficulty is mapped to the corresponding course difficulty score, including advanced (course difficulty score ≥ 7.5), beginner (course difficulty score < 6.5) and intermediate (6.5 ≤ course difficulty score < 7.5).

[0078] In this embodiment, the matching matrix is ​​shown in Table 10 below: Table 10 Matching Matrix for Force Application User muscle strength level Includes beginner courses Intermediate courses Advanced courses Excellent / Good (≥80) 0.5 (Too easy) 0.8 (Suitable) 1.0 (Perfect) Average / Pass (60-79) 0.8 (Suitable) 1.0 (Perfect) 0.6 (Challenging) Needs improvement (<60) 1.0 (Perfect) 0.6 (Challenging) 0.2 (Not recommended)

[0079] S44. For the severity dimension, the severity level of the user's muscle dysfunction is determined based on the total assessment score. The severity level is matched with the graded course difficulty level, and the severity matching score of the currently traversed course is assigned according to the mismatch grade difference.

[0080] That is, when traversing the course library and calculating the matching degree of each course, the user's current severity level is compared with the course difficulty level of the currently traversed courses to generate a mismatch level difference. Based on the level difference, the severity matching score between each course and the user is calculated, thereby measuring the quantitative score of the fit between the user's condition severity and the course difficulty level.

[0081] In one specific implementation method, the mapping relationship between severity and course difficulty is shown in Table 11 below: Table 11 Severity Matching Score Mapping Table Severity level Course Difficulty (Based on Keywords) Matching results Severe (<40 minutes) Beginner (below 6.5) 1.0 (Perfect) Moderate (40-69 points) Intermediate (6.5 to 7.5) 1.0 (Perfect) Mild (≥70 points) Advanced (7.5 and above) 1.0 (Perfect) In this embodiment, if the level difference is 1 level (such as severe with medium level), 0.7 points are awarded; if the level difference is 2 levels (such as severe with high level), only 0.4 points are awarded.

[0082] The purpose of the above mechanism is to match patients with severe muscle dysfunction to basic courses and those with mild dysfunction to advanced courses, ensuring that the training difficulty matches the user's current ability and avoiding training overload or poor results.

[0083] User profile information includes basic information, chief symptoms, and special physiological conditions. Basic information includes the user's age.

[0084] S45. For the user age dimension, match the user age with the preset course applicable age range, and output a stepped user age matching score according to the deviation range obtained from the matching.

[0085] In this embodiment, if the user's age is within the applicable age range of the course, the user age matching score is full (1.0). If the deviation between the two is within the first preset age threshold range (±5 years), the first preset age matching score (0.7) is output; otherwise, the second preset age matching score (0.3) is output as the user age matching score.

[0086] S46. For the chief complaint symptom dimension, the matching results between the course algorithm standard and the course designer standard and the chief complaint symptom are calculated based on the natural language processing algorithm, and the chief complaint symptom matching score is obtained by weighting according to the preset scoring system.

[0087] In this embodiment, the chief complaint symptoms are matched with the adaptive symptoms automatically generated by the course algorithm and the adaptive symptoms pre-marked by the course designer by text matching or semantic matching to obtain the matching results. If a match is found, the score is full (1); otherwise, the preset symptom matching score (0.5) is output as the matching result.

[0088] In this embodiment, the weight of the score matching the chief complaint symptom with the course algorithm symptom is 30%, and the weight of the score matching the course designer's labeled adaptive symptom is 70%. The weighted calculation yields the chief complaint symptom matching score.

[0089] S47. For the special physiological condition dimension, based on the preset base score, points are added item by item according to the special physiological condition to obtain the special physiological condition matching score.

[0090] The special physiological conditions in this embodiment include manual laborers, surgical history, menopause, medical history, and reproductive history. Each item is worth 0.1 points, and the base score is 0.5.

[0091] S5. According to the dimension weight, calculate the total matching score (0~1.0) by weighting the matching scores of each dimension. Generate a course recommendation list based on the total matching score and the corresponding course name and output it.

[0092] The total matching score is related to the recommendation priority. In this embodiment, courses with a total matching score ≥ 0.8 are highly recommended; courses with a total matching score < 0.8 are recommended; courses with a total matching score < 0.4 are considered; and courses with a total matching score < 0.4 are not recommended.

[0093] The course recommendation list includes the course name and the priority tags mentioned above, as well as intelligently generated recommendation reasons, using the following method: Determine the weight type of the muscle state stage to which each stage indicator belongs. The weight types include high weight, medium weight, and low weight.

[0094] In this embodiment, the fast-twitch and slow-twitch muscle fiber phases are given high weight, the endurance test phase is given medium weight, and the pre- and post-resting phases are given low weight.

[0095] Different weighting types correspond to different indicator evaluation levels, that is, different scoring level classification precision. For example, high weighting is divided into 7 levels, medium weighting into 5 levels, and low weighting into 3 levels. At the same time, different indicator evaluation levels of each indicator also correspond to different score ranges.

[0096] Based on the indicator scores, determine the indicator evaluation level corresponding to each stage of the user's indicators, and combine the differences in weight types to determine the advantages and disadvantages of each stage of the indicators, thereby identifying the user's strengths and weaknesses.

[0097] For high-weighted indicators, if the indicator score falls within the range of Exceptional, Excellent, or Very_Good, it is considered an advantage; if it falls within the range of Fair, Poor, or Very_Poor, it is considered a disadvantage. For indicators with medium / low weight, if the indicator score falls into the Excellent level (medium weight) or the Good level (low weight), it is considered an advantage; if it falls into the Poor or Very Poor level, it is considered a disadvantage.

[0098] Different stages have varying degrees of importance within the overall functionality. Scores alone cannot distinguish between critical and secondary weaknesses. Therefore, by assigning weighted levels to determine strengths and weaknesses, we can accurately identify weaknesses in high-weight stages, ensuring that training focuses on improving core functionalities. Strengths are maintained and reinforced, while weaknesses are prioritized for repair and enhancement. The results of these strength and weakness assessments directly serve as the core basis for personalized training plans, course recommendations, and priority ranking, ensuring that recommendations are not merely generic matches but specifically designed to strengthen weak areas.

[0099] A stage priority score is calculated based on the stage evaluation score and stage weight, and the stage priority order is determined based on the stage priority score. Priority score = stage weight × (100 − stage evaluation score); A higher priority score indicates a more important and weaker stage, and training improvements should be prioritized for that stage, thus determining the order of improvement for each stage. Prioritize weak areas with "high weight and low score" (such as fast-twitch muscle dysfunction) rather than simply focusing on the lowest-scoring items.

[0100] By combining the user's strengths, weaknesses, and stage priorities, the urgency of improvement is determined, priority tags and targeted recommendation reasons are generated, and then summarized into the generated course recommendation list.

[0101] The system can determine a user's strengths and weaknesses based on their strengths and weaknesses, determine the urgency of improvement based on priority ranking, match corresponding courses with weaknesses and high-priority stages, and generate personalized recommendation reasons that include improvement goals, targeted basis, and adaptation instructions based on course suitability attributes.

[0102] By identifying user strengths and weaknesses, prioritizing scores to pinpoint areas for improvement, and then using multi-dimensional matching to accurately recommend suitable courses, the entire process of assessment, analysis, and recommendation is made intelligent and personalized, thereby improving the relevance, effectiveness, and user compliance of rehabilitation training.

[0103] Based on the same inventive concept described above, this application also discloses an intelligent course recommendation system for rehabilitation training, with the following architecture: Figure 2 As shown, it includes the following modules: The data input module is used to receive the user's user profile information and raw electromyography (EMG) data. The raw EMG data includes all EMG signal values ​​of the user in multiple different muscle state stages. The Kegel assessment module is used to assess the user's muscle capabilities based on a preset Kegel assessment algorithm and raw electromyography data to obtain the user's pelvic floor muscle assessment score, including the total assessment score and the stage assessment score at each muscle state stage. The abdominal muscle correlation assessment module is used to analyze the electromyographic signal relationship between the user's abdominal muscles and pelvic floor muscles based on the preset abdominal muscle participation correlation scoring algorithm and the raw electromyographic data, to evaluate the user's pelvic floor muscle training quality and obtain a score for force exertion. The multidimensional intelligent matching module is used to determine the severity of a user's muscle dysfunction based on the total assessment score. It matches each course in the course database with the user profile information, pelvic floor muscle assessment score, severity, and exertion score to obtain a matching score for each dimension. The recommendation results output module is used to calculate the total matching score by weighting the matching scores of each dimension, and generate and output a course recommendation list based on the total matching score and the corresponding course name.

[0104] Those skilled in the art will understand that all or part of the steps of the above embodiments can be implemented by hardware, or by a program instructing related hardware. The program can be stored in a computer-readable storage medium, such as a USB flash drive, a portable hard drive, a read-only memory (ROM), a random access memory (RAM), a magnetic disk, or an optical disk, and other media capable of storing program code.

[0105] The above description is merely an optional embodiment of this application and is not intended to limit this application. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the protection scope of this application.

Claims

1. A smart curriculum recommendation method for rehabilitation training, characterized in that, Includes the following steps: Receive user profile information and raw electromyography (EMG) data, wherein the raw EMG data includes all EMG signal values ​​of the user in multiple different muscle state stages; Based on the preset Kegel assessment algorithm, the user's muscle ability is assessed in combination with the original electromyography data to obtain the user's pelvic floor muscle assessment score, including the total assessment score and the stage assessment score in each muscle state stage. Based on the preset abdominal muscle involvement correlation scoring algorithm, combined with the original electromyography data, the electromyographic signal relationship between the user's abdominal muscles and pelvic floor muscles is analyzed to evaluate the user's pelvic floor muscle training quality and obtain a force exertion score. The severity of the user's muscle dysfunction is determined based on the total assessment score. Each course in the course database is matched with the user profile information, the pelvic floor muscle assessment score, the severity, and the exertion score to obtain a matching score for each dimension. The matching scores of each dimension are weighted to obtain the total matching score. A course recommendation list is generated and output based on the total matching score and the corresponding course name.

2. The intelligent curriculum recommendation method for rehabilitation training according to claim 1, characterized in that: The muscle state phases include the pre-resting phase, fast-twitch fiber phase, slow-twitch fiber phase, endurance test phase, and post-resting phase. Each muscle state stage includes several stage indicators, and each stage indicator corresponds to one electromyographic signal value; The phase indicators for both the pre-resting phase and the post-resting phase include average and variability. The phase indicators for the fast-twitch muscle fiber phase include maximum value, rise time, and recovery time. The phase indicators for the slow-twitch muscle fiber phase include average, rise time, recovery time, and variability. The phase indicators for the endurance test phase include average, variability, and the last 10 seconds ratio.

3. The intelligent curriculum recommendation method for rehabilitation training according to claim 2, characterized in that, The process of assessing the user's muscle strength based on a preset Kegel assessment algorithm and the raw electromyography data to obtain the user's pelvic floor muscle assessment score specifically includes the following steps: Based on the measured electromyographic signal values ​​of the user in each muscle state stage, the indicators are quantified by combining the preset mathematical model. Each mathematical model corresponds to a scoring type and a quantification calculation strategy. The mathematical models include a decreasing first model, an increasing second model, and a centrally symmetrical third model. Each stage indicator corresponds to a scoring type, and the quantification calculation strategy of the stage indicator is determined based on the scoring type; The rise time, recovery time, variability, and the average values ​​of the pre-resting phase and the post-resting phase correspond to the scoring type of the first model; the maximum value of the fast-twitch fiber phase, the average value of the slow-twitch fiber phase, and the average value of the endurance test phase correspond to the scoring type of the second model; and the ratio of the last 10 seconds of the endurance test phase corresponds to the scoring type of the third model.

4. The intelligent curriculum recommendation method for rehabilitation training according to claim 3, characterized in that, The electromyographic signal value of the current stage indicator is quantized according to the quantization calculation strategy, specifically including the following steps: Obtain the passing score and full score for each of the aforementioned stage indicators; For the stage indicators corresponding to the first model, the quantification calculation strategy is to normalize the electromyographic signal value based on the passing standard value and the full score standard value. When the electromyographic signal value is not greater than the full score standard value, the full score is output; otherwise, the attenuation is calculated based on the square root of the normalization deviation. For the stage indicators corresponding to the second model, the quantification operation strategy is to normalize and segment the electromyographic signal value based on the passing standard value and the full score standard value. When the electromyographic signal value is not less than the full score standard value, the full score is output. When it is less than the passing threshold, it is reduced by linear proportion. Otherwise, it is incrementally mapped according to the square root of the normalization deviation. For the stage index corresponding to the third model, the quantization calculation strategy is to calculate the absolute deviation of the electromyographic signal value relative to the preset optimal value, determine the passing deviation with the optimal value, perform normalization processing based on the absolute deviation and the passing deviation, output full score when the absolute deviation is zero, and perform symmetrical attenuation calculation based on the square root of the normalized deviation when the absolute deviation is greater than zero.

5. The intelligent curriculum recommendation method for rehabilitation training according to claim 3, characterized in that, It also includes the following steps: The scores after quantitative calculation are used as the index scores of each stage indicator. The scores of each indicator are weighted according to the preset index weights within the corresponding muscle state stage to obtain the stage evaluation score of each muscle state stage. The user's total evaluation score is obtained by weighting the evaluation scores of each stage according to the preset stage weights, and the stage evaluation scores and the total evaluation score are combined into a pelvic floor muscle evaluation score.

6. The intelligent curriculum recommendation method for rehabilitation training according to claim 2, characterized in that, The process involves analyzing the electromyographic signal relationship between the user's abdominal muscles and pelvic floor muscles based on a preset abdominal muscle involvement correlation scoring algorithm and the raw electromyographic data. This evaluation assesses the quality of the user's pelvic floor muscle training and yields a score on muscle activation. Specifically, the process includes the following steps: Based on the abdominal muscle EMG signal and pelvic floor muscle EMG signal in the original electromyography data, the ratio of the EMG signal between the abdominal muscle and pelvic floor muscle of the user, as well as the Pearson correlation coefficient between the abdominal muscle EMG signal sequence and the pelvic floor muscle EMG signal sequence, are calculated. The independent force exertion ability standard score is determined from the preset independent force exertion ability scoring mapping table based on the absolute value of the Pearson correlation coefficient, and the abdominal muscle control ability standard score is determined from the preset abdominal muscle control ability scoring mapping table based on the EMG signal ratio. The score for force exertion is obtained by weighting the independent force exertion ability standard score and the abdominal muscle control ability standard score with preset correlation weights.

7. The intelligent curriculum recommendation method for rehabilitation training according to claim 2, characterized in that, Each course in the course database is matched with the pelvic floor muscle assessment score, the severity score, and the exertion score to obtain a matching score for each dimension. The specific steps include the following: Iterate through each course in the course database to obtain the suggested score range for each course and the dimension weights of each dimension involved in the matching. For the pelvic floor muscle assessment dimension, the course suggested score range, the stage assessment score and the total assessment score are combined to perform stage weighted matching and total assessment score matching. The scores are calculated according to the score range comparison or the preset deviation attenuation rule and then weighted and fused to obtain the pelvic floor muscle assessment matching score. Regarding the force exertion dimension, the user's muscle strength level and course difficulty are classified into different levels, and the corresponding force exertion matching score is output through a preset matching matrix. For the severity dimension, the severity level of the user's muscle dysfunction is determined based on the total assessment score. The severity level is then matched with the graded course difficulty level, and the severity matching score of the currently traversed course is assigned according to the misalignment level difference.

8. The intelligent curriculum recommendation method for rehabilitation training according to claim 7, characterized in that, The matching scores for each course in the course database and the user profile information are obtained by matching each course with the user profile information. The specific steps include the following: The user profile information includes basic information, chief complaints and symptoms, and special physiological conditions. The basic information includes the user's age. For the user age dimension, the user age is matched with the preset course applicable age range, and a stepped user age matching score is output according to the deviation range obtained from the matching. For the chief complaint symptom dimension, the matching results between the course algorithm standard and the course designer standard and the chief complaint symptom are calculated based on the natural language processing algorithm, and the chief complaint symptom matching score is obtained by weighting according to the preset scoring system. For the special physiological condition dimension, based on the preset base score, points are added item by item according to the special physiological condition to obtain the special physiological condition matching score.

9. The intelligent curriculum recommendation method for rehabilitation training according to claim 5, characterized in that, After obtaining the scores of each of the aforementioned stage indicators, the following steps are also included: The weight type of the muscle state stage to which each stage indicator belongs is determined. The weight type includes high weight, medium weight and low weight. Different weight types correspond to different indicator evaluation levels. The indicator evaluation level corresponding to each stage indicator of the user is determined based on the indicator score. The advantages and disadvantages of each stage indicator are determined by combining the differences in weight types. The user's strengths and weaknesses are determined. A stage priority score is calculated based on the stage evaluation score and the stage weight, and the stage priority order is determined based on the stage priority score. The urgency of improvement is determined by combining the user's strengths, weaknesses, and stage priority order, generating priority tags and targeted recommendation reasons, and summarizing them into the generated course recommendation list.

10. An intelligent course recommendation system for rehabilitation training, characterized in that, Includes the following modules: The data input module is used to receive the user's user profile information and raw electromyography (EMG) data, which includes all EMG signal values ​​of the user in multiple different muscle state stages. The Kegel assessment module is used to assess the user's muscle ability based on a preset Kegel assessment algorithm and the original electromyography data to obtain the user's pelvic floor muscle assessment score, including the total assessment score and the stage assessment score at each stage of the muscle state. The abdominal muscle correlation assessment module is used to analyze the electromyographic signal relationship between the user's abdominal muscles and pelvic floor muscles based on a preset abdominal muscle participation correlation scoring algorithm and the original electromyographic data, to assess the user's pelvic floor muscle training quality and obtain a force exertion score. The multidimensional intelligent matching module is used to determine the severity of the user's muscle dysfunction based on the total assessment score, and to match each course in the course database with the user profile information, the pelvic floor muscle assessment score, the severity, and the force exertion score to obtain a matching score for each dimension. The recommendation result output module is used to calculate the total matching score by weighting the matching scores of each dimension, generate a course recommendation list based on the total matching score and the corresponding course name, and output it.

Images