Respiratory Rehabilitation Management System and Method for Children with Bronchial Asthma

Abstract

The present application discloses a respiratory rehabilitation management system and method for children with bronchial asthma, which uses artificial intelligence-based data information analysis and coding technology to perform physiological state time series correlation on the physiological data of asthmatic children objects, and performs embedded coding and semantic fusion on the personal information of the asthmatic children objects, so as to automatically generate a respiratory rehabilitation training plan based on the cross-modal joint representation between the multi-parameter time series correlation features of children's physical signs and the semantic fusion representation of the personal information of the target object. In this way, it is possible to reveal the trend of changes in physiological indicators over time and consider the relevance to the patient's personal information to generate a more accurate and personalized rehabilitation training plan, so as to better meet the needs of patients of different ages, thus realizing the refined and personalized management of children with bronchial asthma.

Description

Technical Field

[0001] This application relates to the field of intelligent management, and more specifically, to a respiratory rehabilitation management system and method for children with bronchial asthma. Background Art

[0002] Bronchial asthma is a chronic inflammatory disease that causes airway narrowing, swelling, and the production of extra mucus, resulting in symptoms such as difficulty breathing, coughing, wheezing, and chest tightness. Asthma in children is particularly common and affects their daily life and sleep quality. Effective respiratory rehabilitation management is crucial for controlling symptoms, reducing acute attacks, and improving the quality of life. Through personalized rehabilitation plans, children can better manage their condition, enhance lung function, reduce dependence on emergency medications, and promote long-term health.

[0003] The invention with the publication number CN117133440A discloses a method, system, storage medium, and electronic device for the management of respiratory chronic diseases and rehabilitation, which includes managing patients' personal information, formulating personalized medication and rehabilitation exercise plans, and sending medication and exercise reminders within a preset time to improve patients' medication compliance and self-management ability.

[0004] Although the above patent uses intelligent analysis technology to comprehensively evaluate the collected data to formulate personalized management plans, its main limitation is that it fails to deeply explore the temporal correlations between physiological data and the mutual influence between physiological parameters and personal information. This defect may lead to the formulation of rehabilitation plans that are not fine-grained and dynamic enough. Specifically, the above patent method only analyzes based on static or short-term data points, ignoring the trends of various physiological indicators over time, which are crucial for understanding the progression of the disease and adjusting treatment plans. In addition, the above patent method fails to fully consider how patients' personal information affects the changes in physiological data. Patients of different age groups may respond differently to treatment, and the lack of in-depth analysis of these factors may result in inaccurate rehabilitation plans, thus affecting the treatment effect and the quality of life of patients.

[0005] Therefore, an optimized respiratory rehabilitation management plan for children with bronchial asthma is desired. Summary of the Invention

[0006] To solve the above technical problems, this application is proposed.

[0007] According to one aspect of this application, a respiratory rehabilitation management system for children with bronchial asthma is provided, which includes:

[0008] A children's object information management and storage module for managing and storing the personal information of children with bronchial asthma; a respiratory rehabilitation training plan formulation module for formulating a respiratory rehabilitation training plan for the children based on the personal information of the children; a respiratory rehabilitation training reminder module for sending reminder information for respiratory rehabilitation training to the children within a set time based on the respiratory rehabilitation training plan. The respiratory rehabilitation training plan formulation module includes:

[0009] A children's object data acquisition module for obtaining the personal information of the target children with bronchial asthma from the children's object information management and storage module, and for performing real-time detection of the physiological state of the target children with bronchial asthma through the smart bracelet of the target children with bronchial asthma to obtain a time series of physiological data. The personal information includes basic information and disease information;

[0010] A physiological data encoding module for performing data regularization and physiological state extraction on the time series of the physiological data to obtain multi-parameter time series correlation features of children's physical signs;

[0011] A personal information encoding module for performing semantic embedding fusion analysis on the personal information to obtain a semantic fusion representation of the personal information of the target object;

[0012] A physiological individual information joint module for performing cross-modal gate interaction prompt guidance joint on the multi-parameter time series correlation features of children's physical signs and the semantic fusion representation of the personal information of the target object to obtain a cross-modal guidance joint encoding feature of physiological state - individual information semantics, and based on the cross-modal guidance joint encoding feature of physiological state - individual information semantics, obtaining the respiratory rehabilitation training plan for the target children with bronchial asthma.

[0013] Furthermore, the physiological data includes heart rate value, body temperature value, blood oxygen saturation, blood pressure value, and pulmonary function test value.

[0014] Furthermore, the physiological data encoding module includes:

[0015] A data regularization unit for regularizing the time series of the physiological data according to the time dimension and the physiological data sample dimension to obtain a multi-parameter data time series matrix of children's physiological signs;

[0016] A physiological state feature extraction unit for passing the multi-parameter data time series matrix of children's physiological signs through a physiological state extractor based on the Inception network to obtain a multi-parameter time series correlation feature map of children's physical signs as the multi-parameter time series correlation features of children's physical signs.

[0017] Furthermore, the personal information encoding module includes:

[0018] A semantic embedding encoding unit, which is used to perform semantic embedding encoding on the basic information and disease information in the personal information respectively to obtain a basic information semantic encoding feature vector and a disease information semantic encoding feature vector;

[0019] A personal information semantic joint unit, which is used to input the basic information semantic encoding feature vector and the disease information semantic encoding feature vector into a personal information semantic joint network to obtain a target object personal information semantic fusion representation vector as the target object personal information semantic fusion representation.

[0020] Further, the physiological individual information joint module includes:

[0021] A physiological state - individual information cross - modal joint unit, which is used to perform cross - modal gate interaction prompt guidance joint on the multi - parameter temporal correlation feature map of the child's physical signs and the target object personal information semantic fusion representation vector to obtain a physiological state - individual information semantic cross - modal guidance joint encoding feature map as the physiological state - individual information semantic cross - modal guidance joint encoding feature;

[0022] A respiratory rehabilitation training plan generation unit, which is used to obtain the respiratory rehabilitation training plan based on the physiological state - individual information semantic cross - modal guidance joint encoding feature map.

[0023] Further, the physiological state - individual information cross - modal joint unit includes:

[0024] A target object personal information semantic self - correlation encoding sub - unit, which is used to calculate the product between the target object personal information semantic fusion representation vector and the transposed vector of the target object personal information semantic fusion representation vector to obtain a target object personal information semantic self - correlation encoding matrix;

[0025] A feature decoupling sub - unit, which is used to perform feature decoupling on the multi - parameter temporal correlation feature map of the child's physical signs along the channel dimension to obtain a set of multi - parameter temporal correlation local feature matrices of the child's physical signs;

[0026] A personal information guidance encoding sub - unit, which is used to perform cross - modal prompt guidance encoding on the set of multi - parameter temporal correlation local feature matrices of the child's physical signs with the target object personal information semantic self - correlation encoding matrix as the guiding feature to obtain a set of physiological state - individual information semantic cross - modal prompt information semantic encoding matrices;

[0027] A gating decoding sub - unit, which is used to perform gating decoding on the set of physiological state - individual information semantic cross - modal prompt information semantic encoding matrices to obtain a set of physiological state - individual information cross - modal semantic interaction attention weights;

[0028] A cross-modal interaction sub-unit, which is used to perform cross-modal interaction on the set of the semantic self-correlation encoding matrix of the personal information of the target object and the set of the multi-parameter time-series correlation local feature matrices of the children's physical signs to obtain a set of cross-modal interaction local feature matrices of physiological state - individual information;

[0029] A cross-modal interaction optimization sub-unit, which is used to perform cross-modal interaction optimization on the set of the cross-modal semantic interaction attention weights of physiological state - individual information and the set of the cross-modal interaction local feature matrices of physiological state - individual information to obtain the cross-modal guidance joint encoding feature map of physiological state - individual information semantics.

[0030] Further, the personal information guidance encoding sub-unit is used for:

[0031] Performing a linear transformation on the semantic self-correlation encoding matrix of the personal information of the target object to obtain a semantic query encoding matrix of the personal information of the target object and a semantic value encoding matrix of the personal information of the target object;

[0032] Performing a converter cross-modal prompt information encoding on each of the multi-parameter time-series correlation local feature matrices of the children's physical signs in the set of the semantic query encoding matrix of the personal information of the target object, the semantic value encoding matrix of the personal information of the target object, and the set of the multi-parameter time-series correlation local feature matrices of the children's physical signs to obtain a set of semantic encoding matrices of cross-modal prompt information of physiological state - individual information.

[0033] Further, the breathing rehabilitation training plan generation unit is used for: unfolding the cross-modal guidance joint encoding feature map of physiological state - individual information semantics into a cross-modal guidance joint encoding feature vector of physiological state - individual information, and adding a prompt template at the end of the cross-modal guidance joint encoding feature vector of physiological state - individual information and then inputting it into a rehabilitation training plan generator based on AIGC to obtain the breathing rehabilitation training plan of the target child with bronchial asthma, and the prompt template is "Based on the input corpus, give a personalized rehabilitation training plan".

[0034] According to another aspect of the present application, there is provided a breathing rehabilitation management method for children with bronchial asthma, which includes:

[0035] Managing and storing the personal information of children with bronchial asthma;

[0036] Formulating a breathing rehabilitation training plan for the child object based on the personal information of the child object;

[0037] Sending a reminder message for breathing rehabilitation training to the child object based on the breathing rehabilitation training plan within a set time;

[0038] Among them, based on the personal information of the child object, a respiratory rehabilitation training plan is formulated for the child object, including:

[0039] Obtain the personal information of the target child with bronchial asthma from the child object information management and storage module, and use the smart bracelet of the target child with bronchial asthma to detect the physiological state of the target child with bronchial asthma in real time to obtain a time series of physiological data. The personal information includes basic information and disease information;

[0040] Perform data regularization and physiological state extraction on the time series of the physiological data to obtain multi-parameter time series correlation features of child signs;

[0041] Perform semantic embedding fusion analysis on the personal information to obtain a semantic fusion representation of the target object's personal information;

[0042] Perform cross-modal gate interaction prompt guidance and combination on the multi-parameter time series correlation features of child signs and the semantic fusion representation of the target object's personal information to obtain a physiological state - individual information semantic cross-modal guidance combined coding feature, and based on the physiological state - individual information semantic cross-modal guidance combined coding feature, obtain the respiratory rehabilitation training plan for the target child with bronchial asthma.

[0043] Compared with the prior art, a respiratory rehabilitation management system and method for children with bronchial asthma provided by the present application obtains the personal information (basic information and disease information) of the target child with bronchial asthma, and obtains a time series of physiological data by detecting the physiological state of the target child with bronchial asthma in real time through a smart bracelet, and uses artificial intelligence-based data information analysis and coding technology to perform time series arrangement and physiological state time series correlation on the physiological data, and perform embedding coding and semantic fusion on the personal information, so as to automatically generate a respiratory rehabilitation training plan according to the cross-modal joint representation between the multi-parameter time series correlation features of child signs and the semantic fusion representation of the target object's personal information. In this way, it is possible to reveal the trend of changes in physiological indicators over time, and consider the relevance with the patient's personal information to generate a more accurate and personalized rehabilitation training plan, so as to better meet the needs of patients of different ages, thereby realizing the refined and personalized management of children with bronchial asthma. BRIEF DESCRIPTION OF THE DRAWINGS

[0044] By describing the embodiments of the present application in more detail in conjunction with the drawings, the above and other objects, features, and advantages of the present application will become more obvious. The drawings are used to provide a further understanding of the embodiments of the present application, and constitute a part of the specification. Together with the embodiments of the present application, they are used to explain the present application and do not constitute a limitation to the present application. In the drawings, the same reference numerals generally represent the same components or steps.

[0045] Figure 1 Block diagram of a respiratory rehabilitation management system for children with bronchial asthma according to an embodiment of the present application;

[0046] Figure 2 Schematic diagram of data flow of a respiratory rehabilitation management system for children with bronchial asthma according to an embodiment of the present application;

[0047] Figure 3 Block diagram of a respiratory rehabilitation training plan formulation module in a respiratory rehabilitation management system for children with bronchial asthma according to an embodiment of the present application;

[0048] Figure 4 Flowchart of a respiratory rehabilitation management method for children with bronchial asthma according to an embodiment of the present application. Detailed implementation manners

[0049] Next, exemplary embodiments according to the present application will be described in detail with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of the present application, rather than all the embodiments of the present application. It should be understood that the present application is not limited by the exemplary embodiments described herein.

[0050] As shown in the present application and the claims, unless the context clearly indicates otherwise, words such as "a", "an", "one", and / or "the" are not specifically singular and may also include plural. Generally speaking, the terms "comprising" and "including" only indicate the inclusion of the steps and elements that have been clearly identified, and these steps and elements do not constitute an exclusive list. The method or device may also include other steps or elements.

[0051] Although the present application makes various references to certain modules in the system according to the embodiments of the present application, however, any number of different modules can be used and run on the user terminal and / or the server. The modules are only illustrative, and different aspects of the system and method can use different modules.

[0052] Flowcharts are used in the present application to illustrate the operations performed by the system according to the embodiments of the present application. It should be understood that the operations before or below do not necessarily need to be executed precisely in sequence. On the contrary, according to the need, various steps can be executed in reverse order or simultaneously. At the same time, other operations can also be added to these processes, or one or several steps can be removed from these processes.

[0053] Next, exemplary embodiments according to the present application will be described in detail with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of the present application, rather than all the embodiments of the present application. It should be understood that the present application is not limited by the exemplary embodiments described herein.

[0054] In the technical solution of this application, a respiratory rehabilitation management system for children with bronchial asthma is proposed. Figure 1 It is a block diagram of a respiratory rehabilitation management system for children with bronchial asthma according to an embodiment of this application. Figure 2 It is a schematic diagram of data flow of a respiratory rehabilitation management system for children with bronchial asthma according to an embodiment of this application. As Figure 1 and Figure 2 shown, a respiratory rehabilitation management system 300 for children with bronchial asthma according to an embodiment of this application includes: a child object information management and storage module 310 for managing and storing personal information of children with bronchial asthma; a respiratory rehabilitation training plan formulation module 320 for formulating a respiratory rehabilitation training plan for the child object based on the personal information of the child object; and a respiratory rehabilitation training reminder module 330 for sending reminder information for respiratory rehabilitation training to the child object based on the respiratory rehabilitation training plan within a set time.

[0055] Specifically, the child object information management and storage module 310 is used to manage and store personal information of children with bronchial asthma. In one example, the system collects relevant information of children with bronchial asthma from multiple channels. This information is divided into two categories: basic information and disease information. The basic information helps to comprehensively understand the living environment and personal conditions of children, providing important background information for subsequent personalized rehabilitation plans. The disease information not only reflects the current health status but also supports the design of personalized treatment plans.

[0056] The collected information needs to go through a strict entry process to ensure its accuracy and integrity. Usually, information can be entered in various ways. On the one hand, through the user interface, parents or medical staff can manually enter the basic information and some disease information of the child. To ensure the quality of the data, necessary verification mechanisms, such as mandatory item checks and format validations, are set during the input process to ensure that all important information is recorded completely and accurately. On the other hand, for some dynamically changing data, such as physiological parameters (heart rate value, body temperature value, blood oxygen saturation, blood pressure value, etc.), they can be collected in real time through intelligent devices (such as intelligent bracelets, intelligent weighing scales, etc.) and automatically synchronized to the system database. This method not only improves the timeliness of the data but also reduces the errors caused by manual intervention. In addition, if the child has established an electronic medical record in a medical institution, the relevant data can be directly imported into the system to avoid repeated entry. At the same time, the connection with other medical systems can also be achieved through the API interface to obtain more comprehensive medical records and ensure the integrity and consistency of the information.

[0057] All the information collected needs to be properly stored in a secure and reliable database. Considering that the personal information of children with bronchial asthma involves sensitive content, strict security measures must be taken to protect this data. Encrypting the stored data using advanced encryption algorithms to prevent unauthorized access is one of the important measures. Setting up a strict permission management system where only authorized personnel can view or modify the information of specific children, and recording every operation log for accountability tracing. In addition, regularly backing up the database to ensure rapid data recovery even in case of accidents. An emergency response plan also needs to be developed to address possible data loss or damage issues, thus ensuring the security and reliability of the data.

[0058] The stored personal information is not only a static record but also an important resource. Through in-depth analysis of this data, a rehabilitation training plan most suitable for each child can be customized. Combining basic information and disease information, the system can generate detailed health records to help doctors better understand the specific situation of each child, thus making more scientific and reasonable clinical decisions. Establishing a prediction model based on historical data to early warn of possible disease changes and timely adjust treatment plans to improve treatment effects.

[0059] In particular, the breathing rehabilitation training plan formulation module 320 is used to formulate a breathing rehabilitation training plan for the child object based on the personal information of the child object. In a specific example of the present application, as Figure 3 shown, the breathing rehabilitation training plan formulation module 320 includes: a child object data acquisition module 321, which is used to obtain the personal information of the target child with bronchial asthma from the child object information management and storage module, and to obtain a time series of physiological data by real-time detecting the physiological state of the target child with bronchial asthma through the smart bracelet of the target child with bronchial asthma, and the personal information includes basic information and disease information; a physiological data encoding module 322, which is used to regularize the time series of the physiological data and extract the physiological state to obtain multi-parameter time series correlation features of child signs; a personal information encoding module 323, which is used to perform semantic embedding fusion analysis on the personal information to obtain a semantic fusion representation of the target object's personal information; a physiological individual information joint module 324, which is used to perform cross-modal gate interaction prompt-guided combination on the multi-parameter time series correlation features of child signs and the semantic fusion representation of the target object's personal information to obtain a physiological state-individual information semantic cross-modal guided combination encoding feature, and based on the physiological state-individual information semantic cross-modal guided combination encoding feature, obtain the breathing rehabilitation training plan of the target child with bronchial asthma.

[0060] Specifically, the child object data acquisition module 321 is configured to obtain the personal information of the target child with bronchial asthma from the child object information management and storage module, and to perform real-time detection on the physiological state of the target child with bronchial asthma through the smart bracelet of the target child with bronchial asthma to obtain a time queue of physiological data. The personal information includes basic information and disease information. Among them, the basic information refers to the information related to the basic identity and living background of the child object. Such information helps to comprehensively understand the child's living environment and personal situation. For example, basic identity materials such as name, gender, and age, and also includes more detailed background materials, such as family address, contact information of guardians, etc.; in addition, it also includes the child's educational background, the name of the school or kindergarten where the child is located, the class, etc., to understand the child's living environment and social interaction situation; living habits also reflect the child's living environment, which records the daily schedule, eating habits, exercise frequency, etc. These information help to evaluate whether the child's lifestyle is conducive to health management and disease prevention. The disease information refers to the medical records and health status descriptions directly related to the child's bronchial asthma condition, which are used to evaluate the current health status and support the design of personalized treatment plans, including disease course information, nursing records, test results, pulmonary function test values, medication records, previous attack situations, complication information, outcome information, special remarks, and family medical history, etc. The disease course information details the emergency treatment course information, the disease course record information of inpatients, etc., reflecting the development process and important events of the disease; the nursing records record the nursing measures received during each hospitalization and their effects, providing details during the treatment process; the test results include various laboratory test reports, such as blood analysis, sputum culture, allergen test, etc., providing objective biomarker data; by regularly performing vital capacity measurement, peak expiratory flow (PEF) measurement to obtain data such as pulmonary function test values, the pulmonary function status can be quantified; the medication records record the types, doses, usage times, and efficacy evaluations of the drugs used in the past, which can guide future treatment options; the previous attack situations record the time, inducement, severity, and treatment methods of each acute attack, summarizing past experience and lessons; the complication information can provide whether there are other concomitant diseases or other health problems caused by asthma, such as sinusitis, ear infections, etc.; the outcome information records the development trend of the disease, whether there is remission, deterioration, etc., evaluating the treatment effect; the special remarks refer to the special explanations or points that need attention by the doctor for the patient's condition, providing additional guiding suggestions; the family medical history is also an important part, which records the history of whether there are similar diseases or other genetic diseases in the immediate family members, which is very important for predicting certain genetic risk factors. By integrating the above two types of information, the system can more accurately understand the specific situation of each child, so as to provide them with more personalized and effective respiratory rehabilitation training programs.

[0061] Specifically, the physiological data encoding module 322 is used to perform data regularization and physiological state extraction on the time queue of the physiological data to obtain the multi-parameter time series correlation characteristics of the child's physical signs. That is, in the technical solution of the present application, first, the time queue of the physiological data is data regularized according to the time dimension and the physiological data sample dimension to obtain the multi-parameter data time series matrix of the child's physiological signs; considering that each parameter in the time queue of the physiological data has time series characteristic information on the scale of time, and each parameter has mutual correlation on the dimension of time, therefore, in order to more intuitively display the relationship between different indicators and their changing trends over time, in the technical solution of the present application, the time queue of the physiological data is data regularized according to the time dimension and the physiological data sample dimension to obtain the multi-parameter data time series matrix of the child's physiological signs, which helps to discover the potential time series correlation patterns between the various physiological parameters, enhances the understanding of the health status of children with bronchial asthma, and lays a rich data support for the subsequent production of personalized training plans.

[0062] Next, the time series matrix of the multi-parameter data of the child's physiological signs is passed through a physiological state extractor based on the Inception network to obtain a multi-parameter time series correlation feature map of the child's physical signs as the multi-parameter time series correlation feature of the child's physical signs. Here, considering that the time series matrix of the multi-parameter data of the child's physiological signs contains complex nonlinear relationships and time dependencies, there is also time series feature correlation information of different scales, such as heart rate fluctuations in a short period of time, or blood pressure values ​​in a long period of time. Therefore, in order to further capture and mine the characteristic information of physiological states at different time scales, so as to more deeply understand and analyze the physiological state of the child, in the technical solution of the present application, the time series matrix of the multi-parameter data of the child's physiological signs is processed by a physiological state extractor based on the Inception network, so as to use convolution kernels of different sizes in the Inception network to capture the physiological characteristic information of the child at multiple scales, and obtain a multi-parameter time series correlation feature map of the child's physical signs, so that complex patterns and trends can be identified from the original physiological data, such as heart rate changes, blood oxygen saturation fluctuations, etc., and the potential correlation between these physiological indicators can be found.

[0063] Specifically, the personal information encoding module 323 is used to perform semantic embedding fusion analysis on the personal information to obtain the semantic fusion representation of the target object's personal information. That is, in the technical solution of this application, first, considering that both the basic information and the disease information in the personal information contain rich semantic feature information, including the semantic association between the key information about the training plan and the context. Based on this, in the technical solution of this application, the basic information and the disease information in the personal information are respectively subjected to semantic embedding encoding to obtain the basic information semantic encoding feature vector and the disease information semantic encoding feature vector. That is, semantic embedding not only considers the vocabulary itself but also the meaning of the vocabulary in the context, which helps the model better understand medical terms such as "asthma" and "allergy history" and their meanings in specific situations. For example, the word "asthma" may be semantically related to words such as "difficulty breathing" and "wheezing", and through embedding encoding, this relevance can be reflected in the feature vector. In particular, in a specific embodiment of this application, the basic information and the disease information can be embedded and encoded through a pre-trained embedding matrix.

[0064] Next, the basic information semantic encoding feature vector and the disease information semantic encoding feature vector are input into the personal information semantic joint network to obtain the semantic fusion representation vector of the target object's personal information as the semantic fusion representation of the target object's personal information. It should be understood that the basic information semantic encoding feature vector and the disease information semantic encoding feature vector respectively represent different aspects of the personal information of the child object. Therefore, in order to integrate information from different aspects and generate a comprehensive and high-level feature representation to better understand and describe the overall health status of the individual, in the technical solution of this application, the basic information semantic encoding feature vector and the disease information semantic encoding feature vector are input into the personal information semantic joint network to obtain the semantic fusion representation vector of the target object's personal information, which helps to capture the interaction and potential association between different information. For example, patients of a specific age group may have different responses to certain treatment methods. In particular, in a specific embodiment of this application, the semantic fusion representation vector of the target object's personal information is obtained by weighting the basic information semantic encoding feature vector and the disease information semantic encoding feature vector by position, so as to generate a personalized semantic fusion representation vector, thereby comprehensively understanding the overall health information of different children.

[0065] Specifically, the physiological individual information joint module 324 is used to perform cross-modal gate interaction prompt-guided combination on the multi-parameter time-series correlation features of the child's physical signs and the semantic fusion representation of the target object's personal information to obtain a physiological state-individual information semantic cross-modal guided combined coding feature, and based on the physiological state-individual information semantic cross-modal guided combined coding feature, obtain the respiratory rehabilitation training plan for the target child with bronchial asthma. Considering that the multi-parameter time-series correlation features of the child's physical signs express the trend of the child's physiological indicators changing over time and the relationships between them. The semantic fusion representation of the target object's personal information synthesizes the basic information of the individual to form a numerical representation that comprehensively describes the personal situation of the child. These two features not only contain their respective original feature information but also have some implicit correlation patterns between them. Therefore, by combining data from different modalities to generate a comprehensive and multi-level feature representation to capture the complementary information between different data types, thereby providing a more comprehensive description of the health status. In the technical solution of this application, cross-modal gate interaction prompt-guided combination is performed on the multi-parameter time-series correlation feature map of the child's physical signs and the semantic fusion representation vector of the target object's personal information to obtain a physiological state-individual information semantic cross-modal guided combined coding feature map. That is, through the cross-modal gate interaction prompt-guided combination mechanism, the model's understanding of the interaction between different modality data can be enhanced, and the importance weights of different modality data can be dynamically adjusted to better adapt to different scenarios and individual differences.

[0066] Specifically, the specific process of performing cross-modal gate interaction prompt-guided combination on the multi-parameter time-series correlation features of the child's physical signs and the semantic fusion representation of the target object's personal information includes: First, perform cross-modal gate interaction prompt-guided combination on the multi-parameter time-series correlation feature map of the child's physical signs and the semantic fusion representation vector of the target object's personal information to obtain a physiological state-individual information semantic cross-modal guided combined coding feature map as the physiological state-individual information semantic cross-modal guided combined coding feature. Specifically, first, calculate the product between the semantic fusion representation vector of the target object's personal information and the transposed vector of the semantic fusion representation vector of the target object's personal information to obtain a target object personal information semantic autocorrelation coding matrix. By calculating the target object personal information semantic autocorrelation coding matrix, the relationship strength, that is, autocorrelation, between different feature elements within the same modality is revealed. The specific autocorrelation coding formula is:

[0067] ;

[0068] Wherein, is the semantic fusion representation vector of the target object's personal information, is matrix multiplication, is The transposed vector, is the semantic self - correlation encoding matrix of the target object's personal information.

[0069] Next, perform feature decoupling on the multi - parameter temporal correlation feature map of the child's physical signs along the channel dimension to obtain a set of multi - parameter temporal correlation local feature matrices of the child's physical signs. Here, performing feature decoupling on the multi - parameter temporal correlation feature map of the child's physical signs is beneficial for implementing fine - grained cross - modal interaction. That is, it improves the model's attention to the local details of the multi - parameter temporal correlation of the child's physical signs, while reducing the redundant information between features and enhancing the distinctiveness of features. Specifically, use the following feature decoupling formula to perform feature decoupling on the multi - parameter temporal correlation feature map of the child's physical signs along the channel dimension to obtain a set of multi - parameter temporal correlation local feature matrices of the child's physical signs; where the feature decoupling formula is:

[0070] ;

[0071] Among them, is the multi - parameter temporal correlation feature map of the child's physical signs, is for performing a feature decoupling operation, , and are respectively the 1st, the th, and the th multi - parameter temporal correlation local feature matrices in the set of multi - parameter temporal correlation local feature matrices of the child's physical signs.

[0072] Then, using the target object personal information semantic self - correlation encoding matrix as the guiding feature, perform cross - modal prompt - guided encoding on the set of the multi - parameter temporal correlation local feature matrices of the children's physical signs to obtain a set of physiological state - individual information semantic cross - modal prompt information semantic encoding matrices. That is, in the technical solution of this application, first, perform a linear transformation on the target object personal information semantic self - correlation encoding matrix to obtain a target object personal information semantic query encoding matrix and a target object personal information semantic value encoding matrix. That is, in a linear mapping manner, convert the target object personal information semantic self - correlation encoding matrix into a form suitable for attention mechanism calculation, where the target object personal information semantic query encoding matrix is used to find the correlation with other modal features, and the target object personal information semantic value encoding matrix preserves the important information of the original features. This conversion is the key to realizing cross - modal interaction, enabling information from different modalities to be compared and fused in a unified way. Subsequently, perform transformer cross - modal prompt information encoding on the target object personal information semantic query encoding matrix, the target object personal information semantic value encoding matrix, and each of the multi - parameter temporal correlation local feature matrices in the set of the multi - parameter temporal correlation local feature matrices of the children's physical signs respectively to obtain a set of physiological state - individual information semantic cross - modal prompt information semantic encoding matrices. In this process, the transformer structure allows the model to perform well in dealing with long - distance dependencies and can effectively capture the complex interaction patterns between different modal features. The set of physiological state - individual information semantic cross - modal prompt information semantic encoding matrices generated in this way not only retains the information of the original modal features but also integrates complementary knowledge from other modalities, improving the richness and accuracy of feature representation. More specifically, using the target object personal information semantic self - correlation encoding matrix as the guiding feature, perform cross - modal prompt - guided encoding on the set of the multi - parameter temporal correlation local feature matrices of the children's physical signs with the following cross - modal prompt - guided encoding formula to obtain a set of physiological state - individual information semantic cross - modal prompt information semantic encoding matrices; where the cross - modal prompt - guided encoding formula is:

[0073] ;

[0074] ;

[0075] ;

[0076] where, is the query embedding matrix, is the target object personal information semantic query encoding matrix, is the value embedding matrix, is the target object personal information semantic value encoding matrix, is is the transposed matrix of is the scale of , that is, the width of the matrix multiplied by the height of the matrix is a function is and is the semantic cross-modal cue information semantic encoding matrix between and .

[0077] Furthermore, gated decoding is performed on the set of the physiological state-individual information semantic cross-modal cue information semantic encoding matrices to obtain a set of physiological state-individual information cross-modal semantic interaction attention weights. In a specific example, the set of the cross-modal cue information semantic encoding matrices is input into an information gating unit based on a decoder for gated decoding to obtain a set of cross-modal semantic interaction attention weights. In particular, the information gating unit is responsible for adjusting the importance of different modal features, and determines which information is more critical by calculating the semantic interaction attention weights to ensure the effectiveness and rationality of the final fusion result. Further, cross-modal interaction is performed on the target object personal information semantic self-correlation encoding matrix and the set of the child physical sign multi-parameter time-series associated local feature matrices to obtain a set of physiological state-individual information cross-modal interaction local feature matrices. In a specific example, the target object personal information semantic self-correlation encoding matrix and the set of the child physical sign multi-parameter time-series associated local feature matrices are input into a cross-modal interaction unit to capture complementary information of the two modalities at the same spatial position, thereby enhancing the spatial consistency of the feature representation to obtain a set of physiological state-individual information cross-modal interaction local feature matrices. Finally, cross-modal interaction optimization is performed on the set of the physiological state-individual information cross-modal semantic interaction attention weights and the set of the physiological state-individual information cross-modal interaction local feature matrices to perform weighted fusion of different modal features by combining the attention mechanism and the interaction features to obtain the physiological state-individual information semantic cross-modal guided joint encoding feature map. More specifically, the following cross-modal interaction optimization formula is used to perform cross-modal interaction optimization on the set of the physiological state-individual information cross-modal semantic interaction attention weights and the set of the physiological state-individual information cross-modal interaction local feature matrices to obtain the physiological state-individual information semantic cross-modal guided joint encoding feature map; where, the cross-modal interaction optimization formula is:

[0078] ;

[0079] ;

[0080] where is the target object personal information semantic self-correlation encoding matrix and and are respectively the 1st, th, and th local feature matrices of the multi-parameter time series correlation of children's physical signs in the set of local feature matrices, is the decoding weight matrix, is the decoder, , , are respectively corresponding cross-modal semantic interaction attention weights of physiological state - individual information, is element-wise multiplication by position, is feature coupling along the channel dimension, is the physiological state - individual information semantic cross-modal guided joint encoding feature map.

[0081] Subsequently, based on the physiological state - individual information semantic cross-modal guided joint encoding feature map, the respiratory rehabilitation training plan is obtained. That is, in the technical solution of this application, the physiological state - individual information semantic cross-modal guided joint encoding feature map is unfolded into a physiological state - individual information semantic cross-modal guided joint encoding feature vector, and after adding a prompt template to the tail of the physiological state - individual information semantic cross-modal guided joint encoding feature vector, it is input into a rehabilitation training plan generator based on AIGC to obtain the respiratory rehabilitation training plan for the target child with bronchial asthma. The prompt template is "Based on the input corpus, give a personalized rehabilitation training plan". That is, the physiological state - individual information semantic cross-modal guided joint encoding feature map obtained by cross-modal guided combination of the multi-parameter time series correlation feature map of children's physical signs and the semantic fusion representation vector of the personal information of the target object is used for generation processing, so as to automatically generate a respiratory rehabilitation training plan. In this way, the trend of physiological indicators changing over time can be revealed, and the relevance with the patient's personal information can be considered to generate a more accurate and personalized rehabilitation training plan, so as to better meet the needs of patients of different ages, thus realizing the refined and personalized management of children with bronchial asthma.

[0082] In particular, during the process of generating a personalized rehabilitation training plan, it is first necessary to expand the above-mentioned physiological state - individual information semantic cross-modal guided joint encoding feature map into a feature vector, and add a prompt template at its tail: "Based on the input corpus, give a personalized rehabilitation training plan". Next, this information with context guidance is fed into an AIGC-based rehabilitation training plan generator. The generator utilizes machine learning models (such as variational autoencoders [VAE], generative adversarial networks [GAN], or more advanced language models like the Transformer architecture) to generate a specific rehabilitation training plan according to the encoded feature vector and the prompt template. Specifically, the generator simulates the effects of different treatment plans and selects the most optimized path. It also takes into account various factors, such as daily activity suggestions (recommending low-intensity exercises suitable for asthmatic children), breathing exercises (teaching how to perform effective breathing techniques), environmental control (guiding families to reduce allergen exposure), medication management (specifying the medication schedule), and psychological support (providing psychological counseling or joining a mutual support group), etc., to ensure that the generated rehabilitation training plan is both scientific and user-friendly.

[0083] In one example, since the multi-parameter temporal correlation feature map of the child's physical signs and the semantic fusion representation vector of the target object's personal information respectively represent the temporal-sample cross-correlation features of physiological data and the embedded semantic joint encoding features of basic information and disease information, after cross-modal prompt gate-guided interactive optimization encoding, the physiological state - individual information semantic cross-modal guided joint encoding feature vector after expansion of the physiological state - individual information semantic cross-modal guided joint encoding feature map will also have differences in the interactive fusion distribution space structure due to cross-modal feature prompt guiding differences, affecting the generative convergence consistency, and thus affecting the quality of the respiratory rehabilitation training plan obtained by the AIGC-based rehabilitation training plan generator.

[0084] Therefore, when the physiological state - individual information semantic cross-modal guided joint encoding feature vector is input into the AIGC-based rehabilitation training plan generator after adding the prompt template at its tail to obtain a respiratory rehabilitation training plan, the physiological state - individual information semantic cross-modal guided joint encoding feature vector is optimized, including the following steps:

[0085] Arrange the respective feature values of the physiological state - individual information semantic cross-modal guided joint encoding feature vector in ascending order to obtain a physiological state - individual information semantic cross-modal guided sequential encoding feature vector;

[0086] In response to the th feature value of the physiological state - individual information semantic cross-modal guided sequential encoding feature vector and the The absolute value of the difference between the eigenvalues is less than or equal to the distance difference hyperparameter , calculate the weighted sum of the -th eigenvalue and the -th eigenvalue as the optimized -th eigenvalue , where and are the -th eigenvalue and the -th eigenvalue of the physiological state-individual information semantic cross-modal guided joint sequential coding feature vector respectively, is the optimized -th eigenvalue, and represent different weight parameters;

[0087] In response to the absolute value of the difference between the -th eigenvalue and the -th eigenvalue of the physiological state-individual information semantic cross-modal guided joint sequential coding feature vector being greater than the distance difference hyperparameter , calculate the square root of the sum of the squares of all eigenvalues of the physiological state-individual information semantic cross-modal guided joint coding feature vector, and multiply the square root by 2 and then divide by the square of the length of the physiological state-individual information semantic cross-modal guided joint coding feature vector to obtain the physiological state-individual information semantic cross-modal guided joint coding space primitive value ;

[0088] After multiplying the physiological state-individual information semantic cross-modal guided joint coding space primitive value by the -th eigenvalue, calculate the weighted subtraction between the product and the -th eigenvalue as the optimized -th eigenvalue , where and represent different weight parameters.

[0089] Combine the optimized -th eigenvalue of the physiological state-individual information semantic cross-modal guided joint sequential coding feature vector to obtain the optimized physiological state-individual information semantic cross-modal guided joint coding feature vector, where the first eigenvalue of the physiological state-individual information semantic cross-modal guided joint sequential coding feature vector remains unchanged.

[0090] In this way, for the problem of insufficient global interaction fusion distribution representation ability caused by long distances exceeding the predetermined local distribution interval threshold in the feature set of the physiological state-individual information semantic cross-modal guided joint encoding feature vector under the predetermined eigenvalue order distribution, the high-dimensional feature space primitive representation based on self-inner product fusion of the physiological state-individual information semantic cross-modal guided joint encoding feature vector is used to capture the complex structure of the global network interaction of its eigenvalues. Thus, by simulating the potential primitive of the high-dimensional feature space based on scale, the interaction fusion distribution relationship between the eigenvalues of the physiological state-individual information semantic cross-modal guided joint encoding feature vector is reconstructed, so as to realize the encoding reconstruction of the true sequence distribution behavior of the physiological state-individual information semantic cross-modal guided joint encoding feature vector at long distances, improve the interaction fusion expression effect of the physiological state-individual information semantic cross-modal guided joint encoding feature vector, improve the convergence consistency of generative regression, and improve the quality of the respiratory rehabilitation training plan obtained by the AIGC-based rehabilitation training plan generator.

[0091] Specifically, the respiratory rehabilitation training reminder module 330 is used to send reminder information for respiratory rehabilitation training to the child object based on the respiratory rehabilitation training plan within a set time. In one example, in order to achieve timely and accurate reminders, the system adopts a variety of information transmission mechanisms. First, the system will trigger reminder tasks at specified time points according to a pre-set schedule. These time points are carefully arranged based on the professional advice of doctors and the specific conditions of the child patients to ensure that each reminder can achieve the best effect. For example, if a certain child patient needs to perform breathing exercises every morning, the system will send a reminder at a fixed time every morning. This timed reminder helps to form regular living habits and improve the persistence and effectiveness of rehabilitation training.

[0092] The reminder information can be sent to users through a variety of channels, including but not limited to text messages, emails, push notifications of mobile applications, etc. For children who are younger or do not have the ability to operate intelligent devices independently, the system can also directly contact parents or guardians by voice call. In addition, considering the actual needs of different families, the system allows personalized setting of reminder methods. For example, some families may prefer to receive messages using WeChat or other instant messaging tools. This multi-channel information transmission mechanism ensures that parents or children can receive reminder information in a timely manner regardless of their location, thus avoiding missing important rehabilitation training sessions.

[0093] In addition to sending reminders according to a preset schedule, the system also has the ability to make intelligent adjustments. It can dynamically evaluate the current physical state of the child based on real-time monitored physiological data (such as heart rate value, body temperature value, blood oxygen saturation, blood pressure value, etc.), and flexibly adjust the reminder content accordingly. For example, when it detects that the child's heart rate has increased abnormally, the system can temporarily add an extra reminder for breathing relaxation exercises to help the child return to a normal state as soon as possible. This flexibility makes the rehabilitation training more in line with the actual needs, improving its effectiveness and pertinence. At the same time, the system can also appropriately adjust the reminder strategy according to factors such as weather changes and holidays to ensure that the reminder information is conveyed to the user at the most appropriate time, thereby further enhancing the user's response rate.

[0094] The system encourages users to actively provide feedback, forming a two-way interaction process. After each reminder is received, the parent or child can inform the system whether the corresponding training item has been completed through simple operations (such as clicking the confirmation button). If it is found that there are unfinished situations, the system will automatically record them and send a second reminder in due course; for those families who often forget to perform specific tasks, the system will also actively push some practical tips to help them better develop good habits. This closed-loop management mechanism not only promotes the rehabilitation process of individual patients, but also accumulates valuable experience and technical reserves for the entire field of respiratory chronic disease and rehabilitation management.

[0095] In order to continuously improve the performance and service quality of the reminder system, the background will conduct a comprehensive data analysis of all sent reminder information. On the one hand, it counts the success rate and response rate of various types of reminders, identifies which time periods are most likely to be overlooked, and then optimizes the overall schedule. For example, through data analysis, it is found that the success rate of reminders during weekends or holidays is relatively low, and the system can consider adjusting to a more intensive or more attractive form to attract users' attention. On the other hand, combined with the big data analysis results accumulated through long-term tracking, it explores more factors that affect the reminder effect, such as the impact of weather changes and holidays on the reminder success rate. Through the study of these factors, the reminder strategy can be further refined and improved to make each reminder more accurate and effective.

[0096] In addition, data analysis is also used to continuously improve the rehabilitation training plan itself. Through comparative studies of a large number of cases, it is found which types of reminder combinations are most helpful in improving compliance and rehabilitation effects, so as to provide a reference basis for subsequent similar cases. For example, certain types of reminder combinations may significantly improve the child's participation and rehabilitation effects, while others may not be effective. By continuously optimizing the reminder strategy and rehabilitation training plan, the system can customize the most suitable rehabilitation path for each child, not only enhancing the treatment effect, but also greatly improving the quality of life of them and their families.

[0097] As described above, the bronchial asthma children's respiratory rehabilitation management system 300 according to the embodiments of the present application can be implemented in various wireless terminals, such as a server with a bronchial asthma children's respiratory rehabilitation management algorithm. In a possible implementation manner, the bronchial asthma children's respiratory rehabilitation management system 300 according to the embodiments of the present application can be integrated into the wireless terminal as a software module and / or a hardware module. For example, the bronchial asthma children's respiratory rehabilitation management system 300 can be a software module in the operating system of the wireless terminal, or can be an application program developed for the wireless terminal; of course, the bronchial asthma children's respiratory rehabilitation management system 300 can also be one of the many hardware modules of the wireless terminal.

[0098] Alternatively, in another example, the bronchial asthma children's respiratory rehabilitation management system 300 and the wireless terminal can also be separate devices, and the bronchial asthma children's respiratory rehabilitation management system 300 can be connected to the wireless terminal through a wired and / or wireless network, and transmit interaction information in accordance with a predefined data format.

[0099] Furthermore, a bronchial asthma children's respiratory rehabilitation management method is also provided.

[0100] Figure 4 It is a flowchart of the bronchial asthma children's respiratory rehabilitation management method according to the embodiments of the present application. As Figure 4 shown, the bronchial asthma children's respiratory rehabilitation management method according to the embodiments of the present application includes the steps of: S1, managing and storing the personal information of the bronchial asthma children object; S2, formulating a respiratory rehabilitation training plan for the children object based on the personal information of the children object; S3, within a set time, sending a reminder message for respiratory rehabilitation training to the children object based on the respiratory rehabilitation training plan.

[0101] In summary, the respiratory rehabilitation management method for children with bronchial asthma according to the embodiments of the present application is elucidated. It obtains the personal information (basic information and disease information) of the target child with bronchial asthma, and obtains the time series of physiological data obtained by real-time detection of the physiological state of the target child with bronchial asthma through a smart bracelet. Then, it uses artificial intelligence-based data information analysis and coding technology to perform time series arrangement and physiological state time series association on the physiological data, and perform embedded coding and semantic fusion on the personal information. Thus, a respiratory rehabilitation training plan is automatically generated based on the cross-modal joint representation between the multi-parameter time series association characteristics of children's physical signs and the semantic fusion representation of the personal information of the target object. In this way, it is possible to reveal the trend of changes in physiological indicators over time, consider the relevance with the patient's personal information, generate a more accurate and personalized rehabilitation training plan, better meet the needs of patients of different ages, and thus achieve refined and personalized management of children with bronchial asthma.

[0102] The various embodiments of the present disclosure have been described above. The above description is exemplary and not exhaustive, and is not limited to the disclosed embodiments. Many modifications and variations are obvious to those of ordinary skill in the art in the technical field without departing from the scope and spirit of the described embodiments. The choice of terms used herein is intended to best explain the principles of the embodiments, practical applications, or improvements to the technology in the market, or to enable other ordinary skill in the art in the technical field to understand the embodiments disclosed herein.

Claims

1. A respiratory rehabilitation management system for children with bronchial asthma, comprising: A child object information management and storage module, used to manage and store personal information of bronchial asthma child objects; A respiratory rehabilitation training program formulation module, used to formulate a respiratory rehabilitation training program for the child subject based on the personal information of the child subject; The respiratory rehabilitation training reminder module is used to send a reminder message of respiratory rehabilitation training to the child subject within a set time based on the respiratory rehabilitation training plan; characterized in that the respiratory rehabilitation training plan formulation module includes: a child object data acquisition module, used to acquire the personal information of the target bronchial asthma child object from the child object information management and storage module, and to detect the physiological state of the target bronchial asthma child object in real time through the smart bracelet of the target bronchial asthma child object to obtain a time queue of physiological data, wherein the personal information includes basic information and disease information; A physiological data encoding module, used for performing data regularization and physiological state extraction on the time queue of the physiological data to obtain multi-parameter temporal correlation characteristics of children's physical signs; A personal information encoding module, used to perform semantic embedding fusion analysis on the personal information to obtain a semantic fusion representation of the target object's personal information; A physiological individual information combination module is used to combine the multi-parameter temporal correlation features of the child's physical signs and the semantic fusion representation of the target object's personal information through a cross-modal gate interactive prompt guidance to obtain a physiological state-individual information semantic cross-modal guidance combined coding feature, and based on the physiological state-individual information semantic cross-modal guidance combined coding feature, obtain a respiratory rehabilitation training plan for the target bronchial asthma child object; Wherein, the physiological individual information combination module includes: A physiological state-individual information cross-modal combination unit, used for performing a cross-modal gate interactive prompt-guided combination on the child's physical sign multi-parameter time series correlation feature map and the target object's personal information semantic fusion representation vector to obtain a physiological state-individual information semantic cross-modal guidance joint encoding feature map as the physiological state-individual information semantic cross-modal guidance joint encoding feature; The respiratory rehabilitation training plan generating unit is used to: expand the physiological state-individual information semantic cross-modal guided joint coding feature map into a physiological state-individual information semantic cross-modal guided joint coding feature vector, and add a prompt template to the tail of the physiological state-individual information semantic cross-modal guided joint coding feature vector and input it into the rehabilitation training plan generator based on AIGC to obtain the respiratory rehabilitation training plan of the target bronchial asthma child object, wherein the prompt template is "based on the input corpus, give a personalized rehabilitation training plan". Wherein, the physiological state-individual information cross-modal joint unit includes: The target object personal information semantic autocorrelation encoding subunit is used to calculate the product between the target object personal information semantic fusion representation vector and the transposed vector of the target object personal information semantic fusion representation vector to obtain the target object personal information semantic autocorrelation encoding matrix; A feature decoupling subunit, used for performing feature decoupling along the channel dimension on the child vital sign multi-parameter time series correlation feature map to obtain a set of child vital sign multi-parameter time series correlation local feature matrices; The personal information guidance coding subunit is used to use the target object personal information semantic autocorrelation coding matrix as a guiding feature to perform cross-modal prompt guidance coding on the set of the child's physical sign multi-parameter temporal correlation local feature matrix to obtain a set of physiological state-individual information semantic cross-modal prompt information semantic coding matrices; A gated decoding subunit, configured to perform gated decoding on the set of physiological state-individual information semantic cross-modal prompt information semantic encoding matrices to obtain a set of physiological state-individual information cross-modal semantic interaction attention weights; A cross-modal interaction subunit, used for performing cross-modal interaction on the set of the semantic autocorrelation encoding matrix of the target object's personal information and the set of the multi-parameter temporal correlation local feature matrix of the child's physical signs to obtain a set of physiological state-individual information cross-modal interaction local feature matrices; The cross-modal interaction optimization subunit is used to perform cross-modal interaction optimization on the set of attention weights of the physiological state-individual information cross-modal semantic interaction and the set of local feature matrices of the physiological state-individual information cross-modal interaction to obtain the physiological state-individual information semantic cross-modal guided joint encoding feature map.

2. The respiratory rehabilitation management system for children with bronchial asthma according to claim 1, characterized in that: The physiological data include heart rate value, body temperature value, blood oxygen saturation, blood pressure value and lung function test value.

3. The respiratory rehabilitation management system for children with bronchial asthma according to claim 2, characterized in that: The physiological data encoding module comprises: A data regularization unit, used for regularizing the time queue of the physiological data according to the time dimension and the physiological data sample dimension to obtain a time series matrix of multi-parameter data of children's physiological signs; The physiological state feature extraction unit is used to pass the child physiological sign multi-parameter data time series matrix through the physiological state extractor based on the Inception network to obtain a child physical sign multi-parameter time series correlation feature map as the child physical sign multi-parameter time series correlation feature.

4. The respiratory rehabilitation management system for children with bronchial asthma according to claim 3, characterized in that: The personal information encoding module includes: A semantic embedding coding unit, used to perform semantic embedding coding on the basic information and the disease information in the personal information respectively to obtain a basic information semantic coding feature vector and a disease information semantic coding feature vector; The personal information semantic union unit is used to input the basic information semantic encoding feature vector and the disease information semantic encoding feature vector into the personal information semantic union network to obtain the target object personal information semantic fusion representation vector as the target object personal information semantic fusion representation.

5. The respiratory rehabilitation management system for children with bronchial asthma according to claim 4, characterized in that: The personal information guidance encoding subunit is used to: Performing a linear transformation on the target object personal information semantic autocorrelation coding matrix to obtain a target object personal information semantic query coding matrix and a target object personal information semantic value coding matrix; The target object personal information semantic query encoding matrix, the target object personal information semantic value encoding matrix and each child vital sign multi-parameter temporal association local feature matrix in the set of the child vital sign multi-parameter temporal association local feature matrix are respectively subjected to converter cross-modal prompt information encoding to obtain a set of physiological state-individual information semantic cross-modal prompt information semantic encoding matrices.

6. A method for respiratory rehabilitation management of children with bronchial asthma, using the respiratory rehabilitation management system for children with bronchial asthma according to claim 1, characterized in that: include: Manage and store personal information of children with bronchial asthma; Formulate a respiratory rehabilitation training plan for the child subject based on the personal information of the child subject; Sending a respiratory rehabilitation training reminder message to the child subject based on the respiratory rehabilitation training plan within a set time; Wherein, based on the personal information of the child subject, a respiratory rehabilitation training plan is formulated for the child subject, including: Obtaining personal information of a target bronchial asthma child object from the child object information management storage module, and detecting the physiological state of the target bronchial asthma child object in real time through the smart bracelet of the target bronchial asthma child object to obtain a time queue of physiological data, wherein the personal information includes basic information and disease information; Performing data regularization and physiological state extraction on the time queue of the physiological data to obtain multi-parameter temporal correlation characteristics of children's physical signs; Performing semantic embedding fusion analysis on the personal information to obtain a semantic fusion representation of the target object's personal information; The multi-parameter temporal correlation features of the children's vital signs and the semantic fusion representation of the target object's personal information are combined through cross-modal gate interaction prompt guidance to obtain the physiological state-individual information semantic cross-modal guidance joint coding features, and based on the physiological state-individual information semantic cross-modal guidance joint coding features, the respiratory rehabilitation training plan for the target bronchial asthma child object is obtained.

Images