Extracorporeal shock wave early intervention scheme evaluation method and system

Abstract

According to the extracorporeal shock wave early intervention scheme evaluation method and system provided by the invention, each intervention scheme is obtained based on the corresponding rehabilitation condition and the professional treatment scheme of the target object, and the rehabilitation condition data set is utilized to debug the treatment effect evaluation network to obtain the treatment effect evaluation network. An early intervention scheme set is evaluated through the treatment effect evaluation network, an early intervention scheme evaluation result is obtained, the early intervention scheme set is generated based on the rehabilitation condition data set, and the question and answer network is used for processing question and answer tasks of the target object. It can be seen that the to-be-treated effect evaluation network is debugged through the shock wave intervention data of the target object and the professional treatment scheme, and intervention popularity training is performed through the rehabilitation condition data set on this basis, so that the evaluation accuracy of the question-answer network in the treatment effect of the personalized patient corresponding to the shock wave intervention data can be improved.

Description

Technical Field

[0001] This application relates to the field of intervention program evaluation technology, and more specifically, to a method and system for evaluating early intervention programs using extracorporeal shock wave therapy. Background Technology

[0002] Extracorporeal shock wave therapy (ESWT) is a non-invasive physical therapy technique based on the physical properties and biomedical effects of shock waves. Its core treatment principle is based on a multi-effect synergistic effect to ultimately achieve clinical rehabilitation goals: On the one hand, as a high-energy mechanical wave, shock waves can directly act on lesion tissue through mechanical effects, loosening adhered tendons, ligaments, or calcifications and improving the local tissue mechanical environment; on the other hand, shock waves generate cavitation effects during propagation, forming microbubbles that burst instantaneously, producing local microjets and pressure changes, promoting blood circulation and nutrient exchange; in addition, shock waves can also activate cell signaling pathways through biological effects, stimulating the proliferation and differentiation of osteoblasts and fibroblasts, regulating the release of inflammatory factors (such as IL-6 and TNF-α), and inhibiting pain signal transmission, ultimately achieving tissue repair, pain relief, and functional recovery.

[0003] Leveraging its unique mechanism of action, ESWT has been widely applied in various clinical fields, including orthopedics / sports rehabilitation, urology, and cardiovascular disease treatment. It has become a key tool for improving patient prognosis, particularly in the early intervention of diseases such as osteochondral injuries, tendinitis, lumbar disc herniation, and urinary tract stones. The precision and personalized suitability of early intervention programs directly determine treatment effectiveness, rehabilitation duration, and the incidence of adverse reactions. Therefore, the scientific evaluation of early extracorporeal shock wave intervention programs has become a core requirement for clinical translation and the implementation of precision medicine.

[0004] However, existing extracorporeal shock wave intervention assessment techniques still face many technical bottlenecks that urgently need to be addressed in practical applications, severely limiting the accuracy and clinical applicability of assessments: The lack of semantic consistency in shockwave intervention data: There are significant differences in individual characteristics (such as age, disease course, lesion type, and baseline health status) among different patients. The therapeutic effect of shockwave therapy is highly dependent on the matching degree between parameters such as energy flux density (EFD), peak pressure, and number of shocks and individual characteristics. This results in a lack of unified standards for the clinical semantics (i.e., the actual impact on rehabilitation effect) of the same shockwave intervention data in different patients. The data correlation and comparability are insufficient, and it cannot directly support the accurate evaluation of personalized plans.

[0005] Insufficient correlation between rehabilitation data and treatment strategies: Existing rehabilitation records are mostly scattered symptom scores (such as VAS scores), imaging data (such as MRI and ultrasound results), or laboratory indicators. They have not formed a structured dataset that is deeply bound to shockwave intervention parameters and professional treatment plans. It is difficult to trace the inherent logical connection between "intervention strategy and rehabilitation effect", resulting in a lack of accurate data support in the assessment process and an inability to clarify the actual efficacy differences of different intervention plans.

[0006] The personalized treatment strategy adaptation mechanism is imperfect: Traditional assessment methods rely on clinical experience to formulate intervention plans, and have not established a reference personalized treatment strategy screening system based on individual patient characteristics. It is difficult to select the optimal plan that matches the target patient from a large number of candidate strategies, which is prone to "one-size-fits-all" assessment bias and cannot meet the needs of personalized treatment.

[0007] The treatment efficacy evaluation models lack specificity: Most existing evaluation models are general-purpose algorithms that are not customized by combining the specific mechanisms of action and data characteristics of shockwave therapy (such as energy flux density, shock frequency, treatment interval, etc.) and are not iteratively optimized using "intervention-rehabilitation" correlation data. This results in low accuracy and weak generalization ability of the models in evaluating personalized plans, and the evaluation results are out of touch with actual clinical needs.

[0008] Lack of personalized question-and-answer response capability: In clinical scenarios, patients and medical staff have highly personalized characteristics in their consultations on rehabilitation programs (such as "expected functional recovery 3 months after shockwave intervention with specific parameters" and "comparison of the effects of different treatment programs"). Existing technologies lack dedicated question-and-answer networks to handle such needs, making it impossible to quickly provide accurate descriptions of rehabilitation status and program evaluation suggestions, which affects the efficiency of clinical decision-making.

[0009] The assessment process has not yet been closed: In existing technologies, the steps of data set construction, rehabilitation data generation, model debugging, program evaluation, and question-and-answer response are isolated from each other and lack a systematic collaborative mechanism. As a result, the assessment results are difficult to effectively feed back into the optimization of intervention programs, and the entire process of "data-model-assessment-optimization" cannot be closed, which restricts the clinical translation value of the technology.

[0010] In summary, resolving the semantic inconsistency issue in shockwave intervention data, constructing an integrated structured dataset encompassing "intervention strategy and rehabilitation effect," establishing a targeted assessment model debugging and optimization mechanism, and simultaneously achieving personalized question-and-answer responses have become key technical challenges in improving the accuracy of early extracorporeal shockwave intervention program assessments and promoting the implementation of personalized clinical treatment. Based on this, this application proposes a method and system for assessing early extracorporeal shockwave intervention programs, aiming to overcome existing technical bottlenecks and provide clinical practice with a scientific, accurate, and efficient personalized assessment solution. Summary of the Invention

[0011] To address the technical problems existing in related technologies, this application provides a method and system for evaluating early intervention programs using extracorporeal shock wave therapy.

[0012] Firstly, a method for evaluating early intervention programs using extracorporeal shock wave therapy is provided, the method comprising: A shockwave intervention data set for the target object is obtained, wherein the clinical semantics of any shockwave intervention data in the target object differ from the clinical semantics of the same shockwave intervention data in other objects. Based on the shockwave intervention data set, a rehabilitation status dataset corresponding to the target object is generated. Each rehabilitation status in the rehabilitation status dataset is matched with at least one shockwave intervention data in the shockwave intervention data set, and each intervention plan is generated based on the corresponding rehabilitation status and the target object's professional treatment plan. The treatment effect evaluation network is obtained by debugging the treatment effect evaluation network using the rehabilitation data set. The treatment effect evaluation network is obtained by optimizing at least one network coefficient in the treatment effect evaluation network. The early intervention protocol set is evaluated and processed by the treatment effect evaluation network to obtain the early intervention protocol evaluation results. The early intervention protocol set is generated based on the rehabilitation status dataset.

[0013] In this application, the process of generating a rehabilitation dataset corresponding to the target object based on the shockwave intervention data set includes: Obtain a set of professional treatment plans for the target object, as well as mining information for each professional treatment plan in the set of professional treatment plans; Referring to the key factor locality in the rehabilitation data example, a set of rehabilitation statuses matching the first shockwave intervention data is generated. The key factor locality includes at least one of key user information, key treatment information, and key rehabilitation status. The set of rehabilitation statuses contains at least one rehabilitation status. The first shockwave intervention data is any one of the shockwave intervention data sets. The artificial intelligence network is invoked to generate an intervention plan corresponding to each rehabilitation condition in the rehabilitation condition set based on the intervention plan local factors in the rehabilitation data example, the set of professional treatment plans for the target object, and the mining information of each professional treatment plan; The set of rehabilitation statuses and the corresponding intervention plans for each rehabilitation status are matched and loaded into the rehabilitation status dataset corresponding to the target object.

[0014] In this application, the rehabilitation status dataset includes X data tuples, where X is a positive integer; the method further includes: Obtain a set of reference personalized treatment strategies corresponding to the set of professional treatment plans, where each reference personalized treatment strategy in the set of reference personalized treatment strategies is used to characterize the method of using the corresponding professional treatment plan; The calling parsing network performs rationality parsing on the X data pairs based on the reference personalized treatment strategy set, the professional treatment plan set for the target object, and the mining information of each professional treatment plan, to obtain the parsing result of each data pair. The parsing result of the a-th data pair is used to characterize whether the intervention plan corresponding to the rehabilitation status in the data pair meets the selection requirements, where a is an integer of 1≤a≤X. Based on the parsing results of the X data tuples, the data tuples in the rehabilitation status dataset that do not meet the selection requirements are cleaned or optimized to obtain an updated rehabilitation status dataset.

[0015] In this application, the reference personalized treatment strategy set includes reference personalized treatment strategies corresponding to the target professional treatment plan, wherein the target professional treatment plan is any one of the professional treatment plan sets; obtaining the reference personalized treatment strategy set corresponding to the professional treatment plan set includes: Obtain feature representation information for Y undetermined personalized treatment strategies, where Y is an integer greater than 1; By selecting from the Y undetermined personalized treatment strategies through the target professional treatment plan, G undetermined personalized treatment strategies are obtained. The commonality coefficient between the feature representation information of each undetermined personalized treatment strategy and the feature representation information of the target professional treatment plan is greater than the specified value of the commonality coefficient, and G is an integer of 1≤G≤Y. Based on the first shockwave intervention data, the G target undetermined personalized treatment strategies are screened a second time to obtain the reference personalized treatment strategy corresponding to the target professional treatment plan. The byte matching difference between a random reference personalized treatment strategy corresponding to the target professional treatment plan and the first shockwave intervention data is greater than a specified difference value.

[0016] In this application, the rehabilitation dataset includes X data pairs, where the b-th data pair includes the b-th rehabilitation status and the b-th intervention plan, where X is a positive integer and b is an integer 1 ≤ b ≤ X; the process of using the rehabilitation dataset to debug the treatment effect evaluation network to obtain the treatment effect evaluation network includes: The intervention plan regression analysis is performed on the b-th rehabilitation condition by calling the treatment effect evaluation network, and the regression analysis results output by the treatment effect evaluation network are obtained. Based on the difference between the b-th intervention plan and the regression analysis results, a regression analysis quantitative indicator corresponding to the b-th data tuple is constructed. With the aim of minimizing the quantitative index of the regression analysis, the network coefficients in the treatment effect evaluation network are optimized to obtain the treatment effect evaluation network with optimized network coefficients. The treatment effect evaluation network with optimized network coefficients is used to iteratively generate the treatment effect evaluation network.

[0017] In this application, the b-th data tuple is matched with the second shock wave intervention data, and the regression analysis result includes P regression analysis bytes, the h-th regression analysis byte corresponds to the h-th byte in the b-th intervention scheme, P is a positive integer, and h is an integer 1 ≤ h ≤ P; the construction of regression analysis quantitative indicators based on the difference between the b-th intervention scheme and the regression analysis result includes: The P-byte quantitative indicators are summed to obtain the first local quantitative indicator. The h-th byte quantitative indicator is calculated based on the difference between the h-th regression analysis byte and the h-th byte in the b-th intervention plan. Based on the difference between the second shock wave intervention data and the first byte cluster in the regression analysis results, a second local quantitative index is constructed. The first byte cluster consists of Z consecutive regression analysis bytes in the regression analysis results. The first byte cluster is generated based on the second shock wave intervention data, where Z is an integer greater than 1. The first local quantification index and the second local quantification index are processed by a function to obtain the regression analysis quantification index corresponding to the b-th data tuple.

[0018] 7. The method according to claim 6, characterized in that, constructing a second local quantitative index based on the difference between the second shock wave intervention data and the first byte cluster in the regression analysis result includes: Calculate the sorting word quantization index between the Z local byte clusters in the first byte cluster whose differences are within the interval [1, Z] and the second shock wave intervention data to obtain Z-1 quantization index values; the differences of the Z local byte clusters are all different and all start from the first byte in the first byte cluster; A second local quantization index is obtained by applying a function to the Z-1 quantization index values; wherein, the weight of any quantization index value is determined based on the ratio of the difference of the local byte cluster corresponding to the quantization index value to Z.

[0019] In this application, the rehabilitation status dataset includes X data tuples, where X is a positive integer; the method further includes: The treatment effect evaluation network is invoked to perform intervention program regression analysis on each rehabilitation status in the X data tuples, and the analysis results corresponding to each data tuple are obtained. The intervention plans and corresponding analysis results of each data pair are analyzed and processed to obtain an early intervention plan set. The analysis and processing of a random data pair is used to characterize the best result among the intervention plans and corresponding analysis results of that data pair, wherein the best result is the intervention plan with the best treatment effect on the patient.

[0020] In this application, the b-th data tuple includes the b-th rehabilitation status and the b-th intervention plan, where b is an integer of 1 ≤ b ≤ X; the process of evaluating the early intervention plan set through the treatment effect evaluation network to obtain the early intervention plan evaluation results includes: The first network is invoked to formulate an intervention plan for the b-th rehabilitation condition, and the first treatment intervention result is obtained. The second network is invoked to implement an intervention plan for the b-th rehabilitation condition, resulting in a second treatment intervention outcome. The treatment effect evaluation network is the base network of the second network. A specified quantitative index is constructed based on a first probability calculation value and a second probability calculation value. The first probability calculation value is the ratio of the first probability to the second probability, and the second probability calculation value is the ratio of the third probability to the fourth probability. The first probability is the probability of generating the optimal result of the b-th data tuple based on the result of the first treatment intervention. The second probability is the probability of generating the optimal result of the b-th data tuple based on the result of the second treatment intervention. The third probability is the probability of generating the non-optimal result of the b-th data tuple based on the result of the first treatment intervention. The fourth probability is the probability of generating the non-optimal result of the b-th data tuple based on the result of the second treatment intervention. The network coefficients in the first network are optimized using the specified quantitative indicators to obtain the debugged first network. The question-answering network is obtained based on the debugged first network.

[0021] In this application, the optimal result of the b-th data tuple includes the second byte cluster, and the non-optimal result of the b-th data tuple includes the third byte cluster. The second byte cluster and the third byte cluster are obtained based on the second shock wave intervention data matched by the b-th data tuple. The step of optimizing the network coefficients in the first network using the specified quantization index to obtain the debugged first network includes: Based on the treatment intervention results, the second byte cluster, and the third byte cluster, a ranking quantification index is constructed. The weighted quantitative index is obtained by performing function processing on the ranking quantitative index and the specified quantitative index. The network coefficients in the first network are optimized using the weighted quantization index to obtain the debugged first network.

[0022] In this application, the method further includes: Obtain the rehabilitation status of the intervention plan, wherein the rehabilitation status of the intervention plan has object representation information; If the object represented by the object representation information is the target object, then the question-answering network is invoked to generate rehabilitation status description data of the rehabilitation status of the intervention plan.

[0023] Secondly, an extracorporeal shock wave early intervention program evaluation system is provided, including a processor and a memory that communicate with each other, wherein the processor is used to read a computer program from the memory and execute it to implement the above-described method.

[0024] This application provides a method and system for evaluating early intervention programs using extracorporeal shock wave therapy. The method obtains a shock wave intervention dataset for the target subject and generates a rehabilitation dataset corresponding to the target subject based on this dataset. Each rehabilitation status in the rehabilitation dataset matches at least one shock wave intervention data point in the dataset. Each intervention program is derived based on the corresponding rehabilitation status and the target subject's professional treatment plan. The treatment effect evaluation network is then debugged using the rehabilitation dataset to obtain the treatment effect evaluation network. This network is used to evaluate the early intervention program set, which is generated based on the rehabilitation dataset. A question-answering network is used to handle the target subject's question-answering task. Therefore, by debugging the treatment effect evaluation network using the target subject's shock wave intervention data and professional treatment plan, and then training it with intervention intensity using the rehabilitation dataset, the accuracy of the question-answering network in evaluating the personalized patient treatment effect related to the shock wave intervention data can be improved. Attached Figure Description

[0025] To more clearly illustrate the technical solutions of the embodiments of this application, the accompanying drawings used in the embodiments will be briefly introduced below. It should be understood that the following drawings only show some embodiments of this application and should not be regarded as a limitation of the scope. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort.

[0026] Figure 1 This is a flowchart of an evaluation method for an early intervention program using extracorporeal shock wave therapy, provided as an embodiment of this application. Detailed Implementation

[0027] To better understand the above technical solutions, the technical solutions of this application will be described in detail below with reference to the accompanying drawings and specific embodiments. It should be understood that the embodiments of this application and the specific features in the embodiments are detailed descriptions of the technical solutions of this application, rather than limitations on the technical solutions of this application. In the absence of conflict, the embodiments of this application and the technical features in the embodiments can be combined with each other.

[0028] Please see Figure 1 This paper presents a method for evaluating early intervention programs using extracorporeal shock wave therapy, which may include the technical solutions described in steps 201-204.

[0029] S201. Obtain the shock wave intervention data set of the target object.

[0030] The core data consists of clinical shockwave intervention datasets, which typically include the following dimensions: patient baseline information: age, gender, diagnosis, disease course, comorbidities, etc.

[0031] Intervention parameter data: Energy parameters: Energy flux density (EFD), peak pressure (p) + ), pulse energy.

[0032] Operating parameters: number of impacts, frequency, treatment site, treatment interval, number of treatment courses.

[0033] Equipment parameters: Shock wave type (focused / diffused), equipment model.

[0034] Assessment indicators and data: Subjective indicators: VAS pain score, JOA score, IIEF questionnaire, etc.; Objective indicators: Imaging data (X-ray, MRI, ultrasound), laboratory indicators (IL-6, TNF-α), functional tests (gait analysis, joint range of motion); Safety indicators: incidence of adverse reactions, types of complications.

[0035] S202. Based on the shockwave intervention data set, generate a rehabilitation data set corresponding to the target object.

[0036] The recovery status dataset is a structured data combination that records the full-dimensional rehabilitation status of the target subjects (mainly human beings) after intervention measures (such as shockwave therapy, physical therapy, rehabilitation training, etc.), including functional recovery, symptom improvement, and quality of life improvement. It is mainly used to evaluate the efficacy of intervention, optimize rehabilitation programs, and explore recovery patterns. It is the core data support for rehabilitation medicine research, clinical decision-making, and the development of related devices / therapies.

[0037] This dataset can be precisely categorized according to the field of rehabilitation, with orthopedic / sports rehabilitation, which is highly related to shockwave intervention, as the core category. It also covers mainstream rehabilitation fields such as neurology, cardiopulmonary, and urology. The following is an explanation based on a general framework plus a detailed approach for core fields, taking into account the standardization, practicality, and scalability of the data, and adapting to the data support needs in scientific research, clinical practice, and patent writing.

[0038] Intervention-related information (connecting intervention measures and clarifying the attribution basis of rehabilitation effects): record the core intervention methods that lead to changes in rehabilitation status. If it is a combined intervention, the parameters / execution status of each method should be marked. The shockwave intervention adaptation field should be bolded: intervention type (single / combined, such as "shockwave + muscle strength training"); core intervention parameters: shockwave type (focused / radial), energy flux density (EFD), number of shocks, treatment site, number of treatment courses, treatment interval; frequency / duration / intensity of rehabilitation training; mode / parameters of physical therapy; intervention execution status: actual number of times completed, whether interrupted, reason for interruption (such as adverse reactions / patient compliance).

[0039] The rehabilitation dataset consists of several data pairs, each including a rehabilitation status (hint) and an intervention plan. The rehabilitation status in a random data pair is matched with at least one shockwave intervention data in the shockwave intervention dataset. The intervention plan in a random data pair is obtained based on the corresponding rehabilitation status and the professional treatment plan of the target subject.

[0040] Next, referring to the key factors in the rehabilitation data example, a set of rehabilitation statuses matching the first shockwave intervention data is generated. The key factors include at least one of key user information, key treatment information, and key rehabilitation status; the set of rehabilitation statuses includes at least one rehabilitation status; and the first shockwave intervention data is a random shockwave intervention data point from the shockwave intervention data set. In one implementation, the first shockwave intervention data is used as input to the artificial intelligence network, and the rehabilitation status generation process is constrained by the rehabilitation data example to obtain one or more rehabilitation statuses related to the first shockwave intervention data output by the artificial intelligence network.

[0041] The artificial intelligence network is invoked to generate an intervention plan corresponding to each rehabilitation condition in the set of rehabilitation conditions based on the intervention plan factors in the rehabilitation data example, the set of professional treatment plans for the target object, and the mining information of each professional treatment plan. The intervention plan factors include at least one of the following: user information, treatment information, and logical expression. It should be understood that by incorporating the target object's professional treatment plan and the mining information of those professional treatment plans during the intervention plan generation process, the professionalism and accuracy of the intervention plan can be improved.

[0042] After obtaining the intervention plan for each rehabilitation condition in the rehabilitation condition set, the rehabilitation condition set and the intervention plan for each rehabilitation condition in the rehabilitation condition set are matched and loaded into the rehabilitation condition dataset corresponding to the target object. Following the above implementation method, data tuples for each shockwave intervention data can be generated, thereby obtaining the rehabilitation condition dataset.

[0043] In one possible implementation, the rehabilitation status dataset can be further selected (e.g., cleaned or optimized for unreasonable data tuples) to improve the descriptive accuracy of the rehabilitation status dataset. In one embodiment, the rehabilitation status dataset includes X data tuples, where X is a positive integer. A set of reference personalized treatment strategies corresponding to the set of professional treatment plans is obtained. The reference personalized treatment strategies in the set of reference personalized treatment strategies are used to characterize / describe the usage method of the corresponding professional treatment plan in the set of professional treatment plans; for example, the reference personalized treatment strategy corresponding to the target professional treatment plan may include a target sentence locality, which is an example sentence containing the target professional treatment plan. Then, the parsing network is invoked to perform rationality parsing on the X data tuples based on the set of reference personalized treatment strategies, the set of professional treatment plans for the target object, and the mining information of each professional treatment plan, to obtain the parsing results of the X data tuples; wherein, the parsing result of the a-th data tuple is used to characterize whether the intervention plan for the a-th rehabilitation status meets the selection requirements. The parsing network can be trained by parsing the training dataset to train the artificial intelligence network, where a is a positive integer not greater than X. It is important to understand that by adding a reference to personalized treatment strategies, the parsing network can more accurately determine whether each data pair is reasonable during the rationality analysis process. After obtaining the parsing results of X data pairs, based on the parsing results of the X data pairs, the data pairs in the rehabilitation status dataset that do not meet the selection requirements are cleaned or optimized to obtain an updated rehabilitation status dataset.

[0044] S203. The treatment effect evaluation network is debugged using the rehabilitation data set to obtain the treatment effect evaluation network.

[0045] The treatment effect evaluation network is obtained by optimizing at least one coefficient in the treatment effect evaluation network. The debugging is used to improve the accuracy of the intervention plan corresponding to the rehabilitation status of the target object generated by the network.

[0046] S204. The early intervention protocol set is evaluated and processed through the treatment effect evaluation network to obtain the early intervention protocol evaluation results.

[0047] The treatment effect evaluation network is used to evaluate the rehabilitation treatment effect of each early intervention program in the early intervention program set, generating a corresponding early intervention program evaluation dataset. Based on the evaluation dataset, the expected treatment effect of each early intervention program is quantitatively analyzed and graded. If this technical step is for shockwave early intervention programs, domain-specific technical features can be added to make the description more consistent with the actual technical solution. The treatment effect evaluation network is used to perform multi-dimensional rehabilitation effect evaluation on the set of early shock wave intervention programs to obtain the evaluation results corresponding to each early shock wave intervention program. Based on the evaluation results, the adaptability of energy parameters and operation parameters of each program is quantitatively analyzed.

[0048] In an alternative embodiment, the rehabilitation dataset comprises X data pairs, where X is a positive integer. The treatment effect evaluation network is invoked to perform intervention regression analysis on each rehabilitation case in the X data pairs, yielding analysis results for each data pair. This is equivalent to generating an additional intervention plan for each rehabilitation case in each data pair to compare with the original intervention plans in each data pair. The intervention plans and corresponding analysis results for each data pair are then analyzed to obtain an early intervention plan set. Analysis of a randomly selected data pair is used to characterize the best outcome among the intervention plans and corresponding analysis results for that data pair. For a given rehabilitation case, the labeled intervention plan includes an optimal outcome (i.e., the labeled intervention plan) and a non-optimal outcome (i.e., the unlabeled intervention plan). The analysis can be performed by the target object's labeling network or by referencing historical data of the target object.

[0049] In one possible implementation, the b-th data tuple includes the b-th recovery status and the b-th intervention plan, where b is an integer of 1 ≤ b ≤ X. The process of training the treatment effect evaluation network with intervention heat using the early intervention plan set includes: on the one hand, calling the first network to implement the intervention plan for the b-th recovery status, obtaining the first treatment intervention result; on the other hand, calling the second network to implement the intervention plan for the b-th recovery status, obtaining the second treatment intervention result. The base networks of both the first and second networks are treatment effect evaluation networks. The difference between the first and second networks is that the network coefficients of the second network remain constant, while the network coefficients of the first network are optimized based on specified quantitative indicators. After obtaining the results of the first and second treatment interventions, the probability of generating the optimal result of the b-th data pair based on the first treatment intervention result is calculated (denoted as the first probability), the probability of generating the optimal result of the b-th data pair based on the second treatment intervention result is calculated (denoted as the second probability), the probability of generating the non-optimal result of the b-th data pair based on the first treatment intervention result is calculated (denoted as the third probability), and the probability of generating the non-optimal result of the b-th data pair based on the second treatment intervention result is calculated (denoted as the fourth probability).

[0050] Next, a specified quantification index is constructed based on the first and second probability calculation values; wherein the first probability calculation value is the ratio of the first probability to the second probability, and the second probability calculation value is the ratio of the third and fourth probabilities. After obtaining the specified quantification index, the network coefficients in the first network are optimized using the specified quantification index to obtain the debugged first network. The question-answering network is obtained by further optimizing the debugged first network according to the above implementation method.

[0051] In this embodiment, a shockwave intervention dataset for the target object is obtained. Based on this dataset, a rehabilitation dataset corresponding to the target object is generated. Each rehabilitation condition in the dataset matches at least one shockwave intervention data point in the dataset. Each intervention plan is derived based on the corresponding rehabilitation condition and the target object's professional treatment plan. The treatment effect evaluation network is then debugged using the rehabilitation dataset to obtain the treatment effect evaluation network. This network is used to evaluate the early intervention plan set, which is generated based on the rehabilitation dataset. A question-answering network is used to handle the target object's question-answering task. Therefore, by debugging the treatment effect evaluation network using the target object's shockwave intervention data and professional treatment plan, and then training it with intervention intensity using the rehabilitation dataset, the accuracy of the question-answering network in evaluating the personalized patient treatment effect related to shockwave intervention data can be improved.

[0052] The present application provides another method for evaluating early intervention programs of extracorporeal shock wave therapy. This method can be executed by an operator and may include the following steps S301-S310: S301, obtaining a set of shock wave intervention data for the target object.

[0053] S302. Based on the shockwave intervention data set, generate a rehabilitation data set corresponding to the target object.

[0054] In an alternative embodiment, the reference personalized treatment strategy set includes reference personalized treatment strategies corresponding to the target professional treatment plan, where the target professional treatment plan is a random professional treatment plan from the set of professional treatment plans. Feature representation information of Y pending personalized treatment strategies is obtained, where Y is an integer greater than 1. Then, the commonality coefficient between the feature representation information of the target professional treatment plan and the feature representation information of each pending personalized treatment strategy is calculated. G target pending personalized treatment strategies are selected from the Y pending personalized treatment strategies using a specified commonality coefficient value; wherein the commonality coefficient between the feature representation information of each target pending personalized treatment strategy and the feature representation information of the target professional treatment plan is greater than the specified commonality coefficient value, and G is an integer 1 ≤ G ≤ Y. After obtaining the G target pending personalized treatment strategies, the byte matching difference between each target pending personalized treatment strategy and the first shockwave intervention data is statistically analyzed. Then, a reference personalized treatment strategy corresponding to the target professional treatment plan is selected from the G target pending personalized treatment strategies using a specified difference value; wherein the byte matching difference between a random reference personalized treatment strategy corresponding to the target professional treatment plan and the first shockwave intervention data is greater than the specified difference value.

[0055] S304. Update the rehabilitation status dataset based on the reference personalized treatment strategy set to obtain the updated rehabilitation status dataset.

[0056] Among them, the reference personalized treatment strategy set refers to a collection of personalized and differentiated treatment strategies that are pre-formulated / actually implemented based on the individual characteristics (age, lesion type, disease course, baseline rehabilitation status, etc.) of different rehabilitation subjects (patients), and serves as the basis for the implementation of rehabilitation intervention.

[0057] Combining shockwave intervention scenarios: This set includes customized shockwave strategies for different patients (focus / radial type, energy flux density (EFD), number of shocks, treatment interval, number of treatment sessions) + combined rehabilitation strategies (muscle strengthening, joint mobilization, etc.), and each strategy is associated with corresponding applicable patient characteristics (e.g., "Patients with talus osteochondral lesions with a disease duration of <6 months: radial shockwave EFD 0.15mJ / mm"). 2 Once a week, for a total of 3 courses of treatment + ankle muscle strengthening training.

[0058] Rehabilitation status dataset: This refers to the previously defined structured rehabilitation status dataset, which records the baseline information of the rehabilitation subjects, subjective / objective / functional / quality of life rehabilitation assessment indicators at each time point after intervention, follow-up prognosis, and other data. It serves as the data record carrier for rehabilitation status.

[0059] Update: It is not simply "adding data", but "relational data improvement" based on treatment strategies. It includes four core actions: supplementation, correction, annotation, and fusion. These are the core technologies of this step, rather than just basic data operations.

[0060] The updated rehabilitation status dataset refers to the structured dataset that integrates "treatment strategy - rehabilitation status" after the "reference personalized treatment strategy set" is linked and merged with the original rehabilitation status dataset. Its core feature is that each rehabilitation status data can be traced back to the corresponding personalized treatment strategy.

[0061] In an alternative embodiment, a parsing network is invoked to perform rationality parsing on X data pairs based on the reference personalized treatment strategy set, the professional treatment plan set for the target object, and the mining information of each professional treatment plan, obtaining parsing results for X data pairs. The parsing result of the a-th data pair is used to characterize whether the intervention plan for the a-th rehabilitation condition meets the selection requirements. The parsing network can be trained on an artificial intelligence network by parsing a training dataset, where a is a positive integer not greater than X. After obtaining the parsing results of the X data pairs, the data pairs in the rehabilitation condition dataset that do not meet the selection requirements are cleaned or optimized based on these results to obtain an updated rehabilitation condition dataset.

[0062] This application provides an architecture for updating a rehabilitation status dataset. As shown, firstly, rehabilitation status is generated based on key factors in the rehabilitation data examples, centered around at least one shockwave intervention data point. It should be understood that the more shockwave intervention data points involved in a rehabilitation status, the more complex the rehabilitation status becomes. Increasing the complexity of the rehabilitation status allows for advanced training of the network. Next, based on the professional treatment plan (and its explanation), the intervention plan factors in the rehabilitation data examples, and the rehabilitation status, an intervention plan is generated, thus obtaining the rehabilitation status dataset.

[0063] S305. The treatment effect evaluation network is debugged using the rehabilitation data set to obtain the treatment effect evaluation network.

[0064] Among them, the rehabilitation status dataset: This is the updated rehabilitation status dataset with the "treatment strategy-rehabilitation effect" correlation feature after structured organization and quality control verification (i.e. the "dataset updated based on the set of personalized treatment strategies" in the previous step), which is the core data foundation for model debugging.

[0065] In the context of shockwave scenarios: This dataset contains fully labeled structured data including patient demographics / clinical baselines, shockwave intervention parameters (EFD, number of shocks, treatment duration, etc.), combined rehabilitation strategies, rehabilitation assessment indicators at each time point (VAS, JOA, joint range of motion, etc.), and follow-up prognosis. It must meet the model training requirements of "features that can be extracted and labels that can be quantified".

[0066] Network for evaluating treatment effectiveness: This refers to a prototype network model that has completed its basic architecture but has not been trained with actual rehabilitation data, has initial parameters, and does not yet have the ability to accurately evaluate the treatment effect. It is the target for debugging.

[0067] In the context of rehabilitation assessment: the network is usually a machine learning / deep learning architecture (such as random forest, CNN, Transformer or lightweight neural network). The input layer is adapted to the feature dimension of "patient baseline + treatment strategy", and the output layer is adapted to the prediction dimension of "rehabilitation effect indicators (pain score, functional recovery rate, prognosis achievement rate, etc.)". Only the architecture design is completed without actual data iteration.

[0068] Debugging: This is not a single "model training" process, but a combined model optimization process for rehabilitation assessment scenarios**, which includes seven core steps: data preprocessing, dataset partitioning, model training, hyperparameter tuning, model validation, generalization testing, and iterative optimization**. This is the core technology of this step and is different from basic model training operations.

[0069] The treatment effect evaluation network obtained refers to the mature rehabilitation effect evaluation network with "precision, stability and generalization" obtained after the above-mentioned whole process of debugging. It is the final product of the debugging.

[0070] Key features: It can accept "new patient baseline + proposed intervention plan (such as shock wave parameter strategy)" as input, and quickly output the corresponding expected rehabilitation effect assessment results (such as pain relief rate, functional recovery score, long-term recurrence rate prediction, etc.), and the assessment error is within the clinically acceptable range.

[0071] In an alternative embodiment, the debugging process includes: calling the treatment effect evaluation network to perform an intervention program regression analysis on the b-th rehabilitation condition, obtaining the regression analysis result output by the treatment effect evaluation network, and then constructing a regression analysis quantitative index corresponding to the b-th data tuple based on the difference between the b-th intervention program and the regression analysis result. In one embodiment, the b-th data tuple is matched with the second shock wave intervention data, and the regression analysis result includes P regression analysis bytes, where the h-th regression analysis byte corresponds to the h-th byte in the b-th intervention program, P is a positive integer, and h is an integer 1 ≤ h ≤ P.

[0072] On the other hand, based on the difference between the first byte cluster in the second shock wave intervention data and the regression analysis results, a second local quantitative index is constructed. The first byte cluster consists of Z consecutive regression analysis bytes in the regression analysis results. The first byte cluster is generated based on the second shock wave intervention data, where Z is an integer greater than 1. In one implementation, the ranking word quantitative index between the Z local byte clusters in the first byte cluster with differences within the interval [1, Z] and the second shock wave intervention data is calculated respectively, resulting in Z-1 quantitative index values. The differences among the Z local byte clusters are inconsistent, and each of the Z local byte clusters is a consecutive byte cluster starting from the first byte in the first byte cluster. After obtaining the Z-1 quantitative index values, a function is applied to the Z-1 quantitative index values ​​to obtain the second local quantitative index. The weight of any quantitative index value is determined based on the ratio of the difference of the local byte cluster corresponding to that quantitative index value to Z. Specifically, it can be expressed as: After obtaining the first and second local quantitative indices, the first and second local quantitative indices are processed by functions to obtain the regression analysis quantitative indices corresponding to the b-th data tuple.

[0073] S306. Generate a set of early intervention protocols through a treatment efficacy evaluation network.

[0074] S307. The early intervention protocol set is evaluated and processed through the treatment effect evaluation network to obtain the early intervention protocol evaluation results.

[0075] In an alternative embodiment, a first network and a second network are generated based on a treatment effect evaluation network. The early intervention protocol set is then processed through the first and second networks respectively to obtain the first treatment intervention result and the second treatment intervention result. Next, a specified quantitative indicator is constructed based on the first and second treatment intervention results, which can be specifically represented as: Where represents the input rehabilitation status, represents the optimal outcome, represents the non-optimal outcome, represents the first network, represents the second network, represents the probability that the treatment intervention of the first network will produce the optimal outcome (i.e., the first probability), similarly, represents the probability that the treatment intervention of the second network will produce the optimal outcome (i.e., the second probability), represents the probability that the treatment intervention of the first network will produce the non-optimal outcome (i.e., the third probability), and represents the probability that the treatment intervention of the second network will produce the non-optimal outcome (i.e., the fourth probability).

[0076] In one possible implementation, the optimal result of the b-th data tuple includes the second byte cluster, and the non-optimal result of the b-th data tuple includes the third byte cluster. The second and third byte clusters are obtained based on the second shockwave intervention data matched by the b-th data tuple. A ranking quantification index can also be constructed based on the treatment intervention results, the second byte cluster, and the third byte cluster. The specific implementation is similar to the construction method of the second local quantification index described above, and the ranking quantification index can be expressed as: This application provides a network training process for a rehabilitation state. First, a treatment effect evaluation network is obtained by debugging a treatment effect evaluation network using a rehabilitation status dataset. Then, the treatment effect evaluation network is trained using an early intervention program set to obtain a question-and-answer network. The intervention training process includes generating a first network and a second network from the treatment effect evaluation network, keeping the network coefficients of the second network unchanged, and optimizing the network coefficients of the first network using the specified quantitative indicators (and ranking quantitative indicators) to obtain the question-and-answer network.

[0077] S308. Obtain the rehabilitation status of the intervention plan.

[0078] In an alternative embodiment, the rehabilitation status of the intervention program has object representation information; for example, the objects involved in the rehabilitation status are explicitly stated in the key pre-declaration. The objects involved in the rehabilitation status of the intervention program can be determined directly based on the object representation information, and step S309 or step S310 can be performed.

[0079] In one possible implementation, if the rehabilitation status of the intervention plan does not carry object representation information, the neighborhood of the rehabilitation status of the intervention plan can be parsed by keywords and context (such as historical rehabilitation status) to obtain the objects involved in the rehabilitation status of the intervention plan, and then proceed to step S309 or step S310.

[0080] S309. If the rehabilitation status of the intervention plan belongs to the target object, then call the question-answering network to generate rehabilitation status description data of the intervention plan.

[0081] S310. Call the question-answering network to generate rehabilitation status description data of the rehabilitation status of the intervention plan based on the target knowledge base.

[0082] The target knowledge base is matched with the subjects involved in the rehabilitation status of the intervention program. In an alternative embodiment, the intervention program is based on the rehabilitation status of inconsistent subjects using the same question-and-answer network.

[0083] In this embodiment, a shockwave intervention dataset for the target object is obtained. Based on this dataset, a rehabilitation dataset corresponding to the target object is generated. Each rehabilitation condition in the dataset matches at least one shockwave intervention data point in the dataset. Each intervention plan is derived based on the corresponding rehabilitation condition and the target object's professional treatment plan. The treatment effect evaluation network is then debugged using the rehabilitation dataset to obtain the treatment effect evaluation network. This network is used to evaluate the early intervention plan set, which is generated based on the rehabilitation dataset. A question-answering network is used to handle the target object's question-answering task. Therefore, by debugging the treatment effect evaluation network using the target object's shockwave intervention data and professional treatment plan, and then training it with intervention intensity using the rehabilitation dataset, the accuracy of the question-answering network in evaluating the personalized patient treatment effect related to shockwave intervention data can be improved.

[0084] Based on the above, an extracorporeal shock wave early intervention program evaluation system is shown, including a processor and a memory that communicate with each other. The processor is used to read a computer program from the memory and execute it to implement the above-described method.

[0085] Based on the above, a computer-readable storage medium is also provided, on which a computer program stored implements the above method during runtime.

[0086] In summary, based on the above scheme, a shockwave intervention dataset for the target subjects is obtained. Based on this dataset, a rehabilitation dataset corresponding to the target subjects is generated. Each rehabilitation condition in the rehabilitation dataset matches at least one shockwave intervention data point in the dataset. Each intervention plan is derived based on the corresponding rehabilitation condition and the target subject's professional treatment plan. The treatment effect evaluation network is then debugged using the rehabilitation dataset to obtain the treatment effect evaluation network. This network is used to evaluate the early intervention plan set, which is generated based on the rehabilitation dataset. A question-answering network is used to handle the target subjects' question-answering tasks. Therefore, by debugging the treatment effect evaluation network using the target subjects' shockwave intervention data and professional treatment plans, and then training it with intervention intensity using the rehabilitation dataset, the accuracy of the question-answering network in evaluating the personalized treatment effects of patients involving shockwave intervention data can be improved.

[0087] It should be understood that the systems and modules described above can be implemented in various ways. For example, in some embodiments, the systems and modules can be implemented by hardware, software, or a combination of both. The hardware portion can be implemented using dedicated logic; the software portion can be stored in memory and executed by an appropriate instruction execution system, such as a microprocessor or dedicated-design hardware. Those skilled in the art will understand that the methods and systems described above can be implemented using computer-executable instructions and / or included in processor control code, for example, such code provided on a carrier medium such as a disk, CD, or DVD-ROM, a programmable memory such as read-only memory (firmware), or a data carrier such as an optical or electronic signal carrier. The systems and modules of this application can be implemented not only by hardware circuits such as very large-scale integrated circuits or gate arrays, semiconductors such as logic chips, transistors, or programmable hardware devices such as field-programmable gate arrays, programmable logic devices, etc., but also by software executed by various types of processors, or by a combination of the aforementioned hardware circuits and software (e.g., firmware).

[0088] It should be noted that different embodiments may produce different beneficial effects. In different embodiments, the beneficial effects may be any one or a combination of the above, or any other possible beneficial effects.

Claims

1. A method for evaluating early intervention programs using extracorporeal shock wave therapy, characterized in that, The method includes: A shockwave intervention data set for the target object is obtained, wherein the clinical semantics of any shockwave intervention data in the target object differ from the clinical semantics of the same shockwave intervention data in other objects. Based on the shockwave intervention data set, a rehabilitation status dataset corresponding to the target object is generated. Each rehabilitation status in the rehabilitation status dataset is matched with at least one shockwave intervention data in the shockwave intervention data set, and each intervention plan is generated based on the corresponding rehabilitation status and the target object's professional treatment plan. The treatment effect evaluation network is obtained by debugging the treatment effect evaluation network using the rehabilitation data set. The treatment effect evaluation network is obtained by optimizing at least one network coefficient in the treatment effect evaluation network. The early intervention protocol set is evaluated and processed by the treatment effect evaluation network to obtain the early intervention protocol evaluation results. The early intervention protocol set is generated based on the rehabilitation status dataset.

2. The method according to claim 1, characterized in that, The process of generating a rehabilitation dataset corresponding to the target object based on the shockwave intervention data set includes: Obtain a set of professional treatment plans for the target object, as well as mining information for each professional treatment plan in the set of professional treatment plans; Referring to the key factor locality in the rehabilitation data example, a set of rehabilitation statuses matching the first shockwave intervention data is generated. The key factor locality includes at least one of key user information, key treatment information, and key rehabilitation status. The set of rehabilitation statuses contains at least one rehabilitation status. The first shockwave intervention data is any one of the shockwave intervention data sets. The artificial intelligence network is invoked to generate an intervention plan corresponding to each rehabilitation condition in the rehabilitation condition set based on the intervention plan local factors in the rehabilitation data example, the set of professional treatment plans for the target object, and the mining information of each professional treatment plan; The set of rehabilitation statuses and the corresponding intervention plans for each rehabilitation status are matched and loaded into the rehabilitation status dataset corresponding to the target object.

3. The method according to claim 2, characterized in that, The rehabilitation dataset includes X data tuples, where X is a positive integer; the method further includes: Obtain a set of reference personalized treatment strategies corresponding to the set of professional treatment plans, where each reference personalized treatment strategy in the set of reference personalized treatment strategies is used to characterize the method of using the corresponding professional treatment plan; The calling parsing network performs rationality parsing on the X data pairs based on the reference personalized treatment strategy set, the professional treatment plan set for the target object, and the mining information of each professional treatment plan, to obtain the parsing result of each data pair. The parsing result of the a-th data pair is used to characterize whether the intervention plan corresponding to the rehabilitation status in the data pair meets the selection requirements, where a is an integer of 1≤a≤X. Based on the parsing results of the X data tuples, the data tuples in the rehabilitation status dataset that do not meet the selection requirements are cleaned or optimized to obtain an updated rehabilitation status dataset.

4. The method according to claim 3, characterized in that, The set of reference personalized treatment strategies includes reference personalized treatment strategies corresponding to the target professional treatment plan, wherein the target professional treatment plan is any one of the professional treatment plan sets. The process of obtaining the reference personalized treatment strategy set corresponding to the set of professional treatment plans includes: Obtain feature representation information for Y undetermined personalized treatment strategies, where Y is an integer greater than 1; By selecting from the Y undetermined personalized treatment strategies through the target professional treatment plan, G undetermined personalized treatment strategies are obtained. The commonality coefficient between the feature representation information of each undetermined personalized treatment strategy and the feature representation information of the target professional treatment plan is greater than the specified value of the commonality coefficient, and G is an integer of 1≤G≤Y. Based on the first shockwave intervention data, the G target undetermined personalized treatment strategies are screened a second time to obtain the reference personalized treatment strategy corresponding to the target professional treatment plan. The byte matching difference between a random reference personalized treatment strategy corresponding to the target professional treatment plan and the first shockwave intervention data is greater than a specified difference value.

5. The method according to claim 1, characterized in that, The rehabilitation dataset includes X data tuples, where the b-th data tuple includes the b-th rehabilitation status and the b-th intervention plan, where X is a positive integer and b is an integer 1 ≤ b ≤ X; The process of debugging the treatment effect evaluation network using the rehabilitation data set to obtain the treatment effect evaluation network includes: The intervention plan regression analysis is performed on the b-th rehabilitation condition by calling the treatment effect evaluation network, and the regression analysis results output by the treatment effect evaluation network are obtained. Based on the difference between the b-th intervention plan and the regression analysis results, a regression analysis quantitative indicator corresponding to the b-th data tuple is constructed. With the aim of minimizing the quantitative index of the regression analysis, the network coefficients in the treatment effect evaluation network are optimized to obtain the treatment effect evaluation network with optimized network coefficients. The treatment effect evaluation network with optimized network coefficients is used to iteratively generate the treatment effect evaluation network.

6. The method according to claim 5, characterized in that, The b-th data tuple is matched with the second shock wave intervention data. The regression analysis result includes P regression analysis bytes, where the h-th regression analysis byte corresponds to the h-th byte in the b-th intervention scheme. P is a positive integer, and h is an integer 1 ≤ h ≤ P. The regression analysis quantitative index is constructed based on the difference between the b-th intervention scheme and the regression analysis result, including: The P-byte quantitative indicators are summed to obtain the first local quantitative indicator. The h-th byte quantitative indicator is calculated based on the difference between the h-th regression analysis byte and the h-th byte in the b-th intervention plan. Based on the difference between the second shock wave intervention data and the first byte cluster in the regression analysis results, a second local quantitative index is constructed. The first byte cluster consists of Z consecutive regression analysis bytes in the regression analysis results. The first byte cluster is generated based on the second shock wave intervention data, where Z is an integer greater than 1. The first local quantification index and the second local quantification index are processed by a function to obtain the regression analysis quantification index corresponding to the b-th data tuple. The construction of a second local quantitative index based on the difference between the second shock wave intervention data and the first byte cluster in the regression analysis results includes: Calculate the sorting word quantization index between the Z local byte clusters in the first byte cluster whose differences are within the interval [1, Z] and the second shock wave intervention data to obtain Z-1 quantization index values; the differences of the Z local byte clusters are all different and all start from the first byte in the first byte cluster; A second local quantization index is obtained by applying a function to the Z-1 quantization index values; wherein, the weight of any quantization index value is determined based on the ratio of the difference of the local byte cluster corresponding to the quantization index value to Z.

7. The method according to claim 1, characterized in that, The rehabilitation dataset includes X data tuples, where X is a positive integer; the method further includes: The treatment effect evaluation network is invoked to perform intervention program regression analysis on each rehabilitation status in the X data tuples, and the analysis results corresponding to each data tuple are obtained. The intervention plans and corresponding analysis results of each data pair are analyzed and processed to obtain an early intervention plan set. The analysis and processing of a random data pair is used to characterize the best result among the intervention plans and corresponding analysis results of that data pair, wherein the best result is the intervention plan with the best treatment effect on the patient.

8. The method according to claim 8, characterized in that, The b-th data tuple includes the b-th rehabilitation status and the b-th intervention plan, where b is an integer 1 ≤ b ≤ X; the process of evaluating the early intervention plan set through the treatment effect evaluation network to obtain the early intervention plan evaluation results includes: The first network is invoked to formulate an intervention plan for the b-th rehabilitation condition, and the first treatment intervention result is obtained. The second network is invoked to implement an intervention plan for the b-th rehabilitation condition, resulting in a second treatment intervention outcome. The treatment effect evaluation network is the base network of the second network. A specified quantitative index is constructed based on a first probability calculation value and a second probability calculation value. The first probability calculation value is the ratio of the first probability to the second probability, and the second probability calculation value is the ratio of the third probability to the fourth probability. The first probability is the probability of generating the optimal result of the b-th data tuple based on the result of the first treatment intervention. The second probability is the probability of generating the optimal result of the b-th data tuple based on the result of the second treatment intervention. The third probability is the probability of generating the non-optimal result of the b-th data tuple based on the result of the first treatment intervention. The fourth probability is the probability of generating the non-optimal result of the b-th data tuple based on the result of the second treatment intervention. The network coefficients in the first network are optimized by the specified quantitative index to obtain the debugged first network. The question-answering network is obtained based on the debugged first network. The optimal result of the b-th data tuple includes the second byte cluster, and the non-optimal result of the b-th data tuple includes the third byte cluster. The second byte cluster and the third byte cluster are obtained based on the second shock wave intervention data matched by the b-th data tuple. The step of optimizing the network coefficients in the first network using the specified quantization index to obtain the debugged first network includes: Based on the treatment intervention results, the second byte cluster, and the third byte cluster, a ranking quantification index is constructed. The weighted quantitative index is obtained by performing function processing on the ranking quantitative index and the specified quantitative index. The network coefficients in the first network are optimized using the weighted quantization index to obtain the debugged first network.

9. The method according to claim 1, characterized in that, The method further includes: Obtain the rehabilitation status of the intervention plan, wherein the rehabilitation status of the intervention plan has object representation information; If the object represented by the object representation information is the target object, then the question-answering network is invoked to generate rehabilitation status description data of the rehabilitation status of the intervention plan.

10. An evaluation system for early intervention programs using extracorporeal shock wave therapy, characterized in that, The method includes a processor and a memory that communicate with each other, the processor being configured to read a computer program from the memory and execute it to implement the method of any one of claims 1-9.

Images