A comprehensive management system and method for postoperative care of orthopedic patients

Abstract

The present application relates to the technical field of postoperative care for orthopedic patients, and specifically discloses a comprehensive management system and method for postoperative care for orthopedic patients, which obtains user questions through digital humans, and introduces data processing and semantic understanding algorithms based on artificial intelligence and natural language processing technology into digital humans to perform semantic analysis on user questions, and then searches for multiple answer records related to the semantics of user questions in the knowledge base, thereby generating dynamic aggregation representation information of the semantic feature distribution of answer records, and then generating question reply text based on this information. In this way, in the process of postoperative care and management of orthopedic patients, digital humans can automatically obtain and reply to user questions. This intelligent question-and-answer system helps to reduce the workload of medical staff during postoperative care of orthopedic patients, improve the efficiency of doctor-patient communication, and thus help to establish a more intelligent comprehensive management system for postoperative care of orthopedic patients.

Description

Technical Field

[0001] The present application relates to the technical field of postoperative care for orthopedic patients, and more specifically, to a comprehensive management system and method for postoperative care for orthopedic patients. Background Art

[0002] With the continuous advancement of medical technology, especially in the field of orthopedics, the importance of postoperative care has become increasingly prominent. Postoperative care is not only related to the patient's recovery speed and quality of life, but also directly affects the surgical effect and the incidence of complications. Therefore, it is particularly important to establish an efficient and comprehensive postoperative care management system.

[0003] After orthopedic surgery, patients usually need to go through a long recovery period. During this process, individual differences in patients, postoperative complications, rehabilitation progress and other factors require nursing staff to provide personalized care plans. However, traditional postoperative care management methods for orthopedic patients often rely on paper records or simple electronic document management. This management method results in scattered storage of patient information, which is difficult to integrate and retrieve, affecting the information integrity and continuity when managing postoperative orthopedic patients. In addition, traditional nursing management methods are often standardized and difficult to adjust according to the specific circumstances of each patient, and cannot meet the individual needs of orthopedic patients after surgery.

[0004] Therefore, an optimized integrated management system for postoperative care of orthopedic patients is desired. Summary of the invention

[0005] The present application provides an orthopedic patient postoperative care comprehensive management system and method, which can automatically obtain and respond to users' questions through digital humans during the process of postoperative care and management of orthopedic patients. This intelligent question-and-answer system helps to reduce the workload of medical staff during postoperative care of orthopedic patients and improve the efficiency of doctor-patient communication, thereby helping to establish a more intelligent orthopedic patient postoperative care comprehensive management system.

[0006] In the first aspect, a comprehensive management system for postoperative care of orthopedic patients is provided, comprising:

[0007] Personal information management module, used to manage basic patient information and surgical information;

[0008] Health record management module, used to manage preoperative examination reports, surgical records and postoperative recovery tracking;

[0009] A rehabilitation plan formulation and tracking module is used to customize a personalized rehabilitation training plan according to the patient's condition and regularly evaluate the rehabilitation training plan;

[0010] Medication reminder and guidance module, used to set medication reminders and provide detailed medication instructions;

[0011] Nutritional advice module to recommend a list of foods that will aid bone recovery;

[0012] Intelligent consultation question-and-answer module, used to answer user questions based on digital humans;

[0013] The intelligent consultation question-answering module includes:

[0014] A user question acquisition unit, used to acquire user questions;

[0015] A user question semantic embedding coding unit, used for performing semantic embedding coding on the user question to obtain a user question semantic embedding coding vector;

[0016] An answer record matching encoding unit, used to match multiple answer records from a knowledge base based on the user question semantic embedding encoding vector, and semantically encode the multiple answer records respectively to obtain a set of answer record semantic encoding vectors;

[0017] A feature distribution semantic dynamic aggregation processing unit, used for performing feature distribution semantic dynamic aggregation on the set of answer record semantic coding vectors to obtain a semantically significant aggregation representation of the answer record, comprising: a static energy factor calculation subunit, used for calculating the static energy factor of each answer record semantic coding vector in the set of answer record semantic coding vectors to obtain a set of answer record semantic static energy factors; a feature significant semantic aggregation processing subunit, used for performing feature significant semantic aggregation on the set of answer record semantic coding vectors based on the set of answer record semantic static energy factors to obtain the semantically significant aggregation representation of the answer record;

[0018] A question reply text generating unit is used to generate a question reply text based on the semantically significant aggregation representation of the answer record.

[0019] In a possible implementation, the answer record matching encoding unit is used to:

[0020] Extracting a set of template question-answer records from the knowledge base;

[0021] Semantically encoding multiple template questions in the set of template question-answer records to obtain a set of template question semantic embedding encoding vectors;

[0022] Calculating the similarity between each template question semantic embedding coding vector in the set of template question semantic embedding coding vectors and the user question semantic embedding coding vector to obtain a set of question semantic similarities;

[0023] Each question semantic similarity in the set of question semantic similarities is compared with a predetermined threshold to obtain the multiple answer records.

[0024] In a possible implementation, the static energy factor calculation subunit is used to:

[0025] Calculating the mean and variance of the answer record semantic encoding vector to obtain the answer record semantic mean and the answer record semantic variance;

[0026] Subtracting the answer record semantic encoding vector from the answer record semantic mean by position, and calculating the fourth power of each position of the subtracted feature vector to obtain the answer record semantic difference modulation vector;

[0027] Calculating the expected value of the answer record semantic difference modulation vector to obtain the answer record semantic expected value;

[0028] The answer record semantic expectation value is divided by the square of the answer record semantic variance, and the value obtained by the division is input into a sigmoid function to obtain the answer record semantic static energy factor.

[0029] In a possible implementation, the feature-significant semantic aggregation processing subunit includes:

[0030] The initial center vector of the semantic clustering determines a secondary subunit, which is used to select the answer record semantic encoding vector corresponding to the largest answer record semantic static energy factor in the set of answer record semantic static energy factors as the initial center vector of the answer record semantic clustering;

[0031] A dynamic aggregation energy factor calculation secondary subunit is used to calculate the dynamic aggregation energy factor of each answer record semantic coding vector in the set of answer record semantic coding vectors based on the spatial span between each answer record semantic coding vector in the set of answer record semantic coding vectors and the initial center vector of the answer record semantic clustering, as well as the static energy factor of each answer record semantic coding vector and the static energy factor of the initial center vector of the answer record semantic clustering, so as to obtain a set of answer record semantic dynamic aggregation energy factors;

[0032] A gated mask processing secondary subunit, used for inputting the set of answer record semantic dynamic aggregation energy factors into a gated mask unit to obtain a set of answer record semantic dynamic aggregation weight factors;

[0033] The weighted and processed secondary sub-unit is used to calculate the weighted sum of the set of answer record semantic encoding vectors using the set of answer record semantic dynamic aggregation weight factors as weighted weights to obtain the answer record semantic significant aggregation representation vector as the answer record semantic significant aggregation representation.

[0034] In a possible implementation, the dynamic aggregation energy factor calculation secondary subunit is used to:

[0035] Multiplying the static energy factor of the answer record semantic encoding vector and the static energy factor of the answer record semantic clustering initial center vector by a first weighting parameter to obtain a first answer record semantic dynamic aggregation energy factor;

[0036] Multiplying the square of the spatial span between the answer record semantic encoding vector and the answer record semantic clustering initial center vector by a second weighting parameter to obtain a second answer record semantic dynamic aggregation energy factor;

[0037] The first answer record semantic dynamic aggregation energy factor is divided by the second answer record semantic dynamic aggregation energy factor to obtain the answer record semantic dynamic aggregation energy factor.

[0038] In a possible implementation, the gated mask processing secondary subunit is used to:

[0039] Normalizing the set of answer record semantic dynamic aggregation energy factors to obtain a set of normalized answer record semantic dynamic aggregation energy factors;

[0040] Masking the set of normalized answer record semantic dynamic aggregation energy factors to obtain the set of answer record semantic dynamic aggregation weight factors;

[0041] Among them, in response to each normalized answer record semantic dynamic aggregation energy factor in the set of normalized answer record semantic dynamic aggregation energy factors being greater than a predetermined threshold, the normalized answer record semantic dynamic aggregation energy factors greater than the predetermined threshold are set to the original value, and the rest are set to zero to obtain the set of answer record semantic dynamic aggregation weight factors.

[0042] In a possible implementation, normalizing the set of answer record semantic dynamic aggregation energy factors to obtain a set of normalized answer record semantic dynamic aggregation energy factors includes:

[0043] Calculate an exponential function value with the natural constant e as the base and the negative of each answer record semantic dynamic aggregation energy factor in the set of answer record semantic dynamic aggregation energy factors as the exponent to obtain a set of answer record semantic dynamic aggregation energy exponential factors;

[0044] After adding the set of answer record semantic dynamic aggregation energy index factors and the constant one positionally, the inverse of each answer record semantic dynamic aggregation energy index modulation factor in the set of answer record semantic dynamic aggregation energy index modulation factors is calculated to obtain the set of normalized answer record semantic dynamic aggregation energy factors.

[0045] In a possible implementation, the question reply text generating unit is used to: input the semantically significant aggregation representation vector of the answer record into an answer generator based on a large language model to obtain the question reply text.

[0046] Secondly, a comprehensive management method for postoperative care of orthopedic patients is provided, including:

[0047] Manage patients' basic information and surgical information;

[0048] Manage preoperative examination reports, surgical records and postoperative recovery tracking;

[0049] Customize individualized rehabilitation training plans according to the patient's condition and regularly evaluate the rehabilitation training plans;

[0050] Set medication reminders and provide detailed medication instructions;

[0051] A list of foods recommended to aid bone recovery;

[0052] Answer user questions based on digital humans;

[0053] Among them, answering user questions based on digital humans includes:

[0054] Get user questions;

[0055] Performing semantic embedding coding on the user question to obtain a user question semantic embedding coding vector;

[0056] Based on the user question semantic embedding coding vector, multiple answer records are matched from the knowledge base, and the multiple answer records are semantically encoded respectively to obtain a set of answer record semantic encoding vectors;

[0057] The set of answer record semantic coding vectors is subjected to feature distribution semantic dynamic aggregation to obtain answer record semantic significant aggregation representation, including: calculating the static energy factor of each answer record semantic coding vector in the set of answer record semantic coding vectors to obtain a set of answer record semantic static energy factors; based on the set of answer record semantic static energy factors, the set of answer record semantic coding vectors is subjected to feature significant semantic aggregation to obtain the answer record semantic significant aggregation representation;

[0058] Based on the semantically significant aggregate representation of the answer record, a question reply text is generated.

[0059] The present application provides a comprehensive management system and method for postoperative care of orthopedic patients, which obtains user questions through digital humans, and introduces data processing and semantic understanding algorithms based on artificial intelligence and natural language processing technology into the digital humans to perform semantic analysis on user questions, and then searches for multiple answer records related to the semantics of the user questions in the knowledge base, thereby generating dynamic aggregation representation information of the semantic feature distribution of the answer records, and then using this information to generate a question reply text. In this way, in the process of postoperative care and management of orthopedic patients, digital humans can automatically obtain and reply to user questions. This intelligent question-and-answer system helps to reduce the workload of medical staff during postoperative care of orthopedic patients and improve the efficiency of doctor-patient communication, thereby helping to establish a more intelligent comprehensive management system for postoperative care of orthopedic patients. BRIEF DESCRIPTION OF THE DRAWINGS

[0060] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the drawings of the embodiments of the present application are briefly introduced below. Obviously, the drawings described below only relate to some embodiments of the present application, and are not intended to limit the present application.

[0061] Figure 1 This is a schematic block diagram of the comprehensive management system for postoperative care of orthopedic patients according to an embodiment of the present application.

[0062] Figure 2 This is a schematic block diagram of the intelligent consultation question and answer module in the comprehensive management system for postoperative care of orthopedic patients in an embodiment of the present application.

[0063] Figure 3 This is a schematic diagram of the data flow of the intelligent consultation question and answer module in the comprehensive management system for postoperative care of orthopedic patients in an embodiment of the present application.

[0064] Figure 4 This is a schematic block diagram of a feature distribution semantic dynamic aggregation processing unit in the comprehensive management system for postoperative care of orthopedic patients according to an embodiment of the present application.

[0065] Figure 5 This is a schematic flowchart of the comprehensive management method for postoperative care of orthopedic patients according to an embodiment of the present application. DETAILED DESCRIPTION

[0066] The technical solutions in the embodiments of the present application will be described clearly and completely below in conjunction with the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present application, not all embodiments. Based on the embodiments of the present application, all other embodiments obtained by ordinary technicians in this field without creative work also fall within the scope of protection of the present application.

[0067] The recovery phase after orthopedic surgery is a critical period during which patients may face various challenges, including individual recovery differences and possible complications. These factors require the nursing team to provide more personalized and flexible care strategies. However, in previous practices, information management for postoperative care of orthopedic patients mostly relied on traditional paper records or basic electronic document systems. This approach leads to fragmented storage of patient-related information, which makes it difficult to integrate and retrieve information, and thus affects the quality and efficiency of care. At the same time, traditional nursing models are usually carried out according to unified standards and lack the ability to make adjustments based on the specific circumstances of each patient, which obviously cannot fully meet the needs of modern orthopedic postoperative care.

[0068] In view of the above technical problems, a comprehensive postoperative care management system for orthopedic patients is proposed, which integrates multiple functional modules. Figure 1 Schematic diagram of the comprehensive management system for postoperative care of orthopedic patients according to the embodiment of the present application. Figure 1 As shown, the orthopedic patient postoperative care integrated management system 100 includes: a personal information management module 110, which is used to manage the patient's basic information and surgical information; a health record management module 120, which is used to manage preoperative examination reports, surgical records and postoperative recovery tracking; a rehabilitation plan formulation and tracking module 130, which is used to customize personalized rehabilitation training plans according to the patient's condition and regularly evaluate the rehabilitation training plans; a medication reminder and guidance module 140, which is used to set medication reminders and provide detailed medication instructions; a nutritional advice module 150, which is used to recommend a list of foods that help bone recovery; and an intelligent consultation question and answer module 160, which is used to answer user questions based on a digital person.

[0069] In the above-mentioned comprehensive management system for postoperative care of orthopedic patients, the personal information management module 110 is used to manage basic patient information and surgical information. It should be understood that the personal information management module is mainly responsible for collecting, organizing, storing and updating the patient's personal information and information related to the surgery. The role of this module is to ensure that the detailed information of each patient is properly preserved, convenient for medical staff to check at any time, and also convenient for personalized management of the patient's postoperative care. Optionally, in one embodiment of the present application, the main functions of the personal information management module include but are not limited to the following points: 1. Patient basic information management: including basic identity information such as name, gender, age, contact information, emergency contact, etc. This information is very important in emergency situations, such as when it is necessary to quickly notify family members or emergency contacts. 2. Surgical information management: record the patient's surgery type, surgery date, surgeon, anesthesia method, etc. This information is crucial to understanding the patient's treatment process and postoperative care needs. 3. Medical history record: save the patient's past medical history, allergy history, family genetic history, etc., so that doctors can refer to it when formulating treatment plans.

[0070] In the above-mentioned comprehensive management system for postoperative care of orthopedic patients, the health record management module 120 is used to manage preoperative examination reports, surgical records and postoperative recovery tracking. It should be understood that the main responsibility of the health record management module is to comprehensively manage the patient's preoperative examination reports, surgical records and postoperative recovery. The design purpose of this module is to ensure that all important information related to the patient's health status can be systematically recorded and convenient for medical staff to view and use at any time to support more effective clinical decision-making and postoperative care. Optionally, in one embodiment of the present application, the functions of the health record management module include but are not limited to the following aspects: 1. Preoperative examination report management: Before the patient undergoes surgery, a series of examinations are usually required to assess whether the patient's physical condition is suitable for surgery. These examinations may include blood tests, imaging examinations (such as X-rays, CT scans, MRI) and other special examination results. The health record management module can store these examination reports to ensure that they are easy to access, so that doctors can assess surgical risks and formulate surgical plans. 2. Surgical record management: The surgical record contains key information about the surgical process, such as the time of surgery, the anesthesia method used, the surgical operation steps, the problems encountered during the operation and the solutions. These records are very important for postoperative care, and can help doctors understand the specific circumstances of the surgery, so as to better formulate postoperative care plans. 3. Postoperative recovery tracking: Postoperative recovery tracking involves monitoring the patient's recovery process, including but not limited to wound healing, pain control, rehabilitation training progress, etc. This module can record the patient's recovery data, so that medical staff can adjust the treatment plan according to the actual situation and ensure the patient's smooth recovery.

[0071] In the above-mentioned comprehensive management system for postoperative care of orthopedic patients, the rehabilitation plan formulation and tracking module 130 is used to customize a personalized rehabilitation training plan according to the patient's condition and regularly evaluate the rehabilitation training plan. It should be understood that the main responsibility of the rehabilitation plan formulation and tracking module is to formulate a personalized rehabilitation training plan according to the individual situation of the patient and regularly evaluate the effects of these plans. Such a module is essential to ensure that the patient can recover safely and effectively after surgery. Optionally, in one embodiment of the present application, the functions of the rehabilitation plan formulation and tracking module include but are not limited to the following aspects: 1. Personalized rehabilitation training plan: Based on the specific situation of each patient (including type of surgery, physical condition, age, personal habits, etc.), a rehabilitation training plan that meets the actual needs of the patient is designed. This ensures that the training program is both targeted and adaptable to the patient's specific conditions, thereby improving the rehabilitation effect. 2. Regularly evaluate rehabilitation progress: Through continuous monitoring of the implementation of the rehabilitation training plan, potential problems or obstacles can be discovered in a timely manner, and the rehabilitation strategy can be adjusted accordingly. Regular evaluation helps to ensure that the patient recovers according to the expected progress and can quickly respond to any deviation from the target. 3. Promote doctor-patient communication: The rehabilitation plan formulation and tracking module enables medical staff to better understand the patient's recovery status, while also allowing patients to understand what steps they should follow to promote recovery. This transparency helps to enhance trust between doctors and patients, thereby improving the overall care experience.

[0072] In a specific embodiment of the present application, assuming that a patient undergoes knee replacement surgery, the rehabilitation plan formulation and tracking module will play the following roles: 1. Rehabilitation training plan formulation: In the early stage of surgery, a plan including ice compresses, leg elevation and simple toe movements is formulated to reduce swelling and promote blood circulation. As the recovery progresses, more activities are gradually added, such as sitting, standing, walking exercises, etc., aimed at enhancing muscle strength and joint flexibility. Later stages may include balance training, gait correction exercises, and enhanced exercise to help patients regain independence in daily life. 2. Rehabilitation training plan tracking: During the first month after surgery, a rehabilitation assessment is conducted at least once a week, including observing the patient's mobility and listening to the patient's feedback on the degree of pain. According to the results of each assessment, the intensity and type of training are adjusted in a timely manner, such as increasing the frequency of certain activities or reducing other activities to ensure the effectiveness of the rehabilitation plan. Use this module to record the results of each assessment, including the patient's subjective feelings, objective measurement indicators (such as steps, pain scores, etc.), and any areas that need adjustment.

[0073] In the above-mentioned comprehensive management system for postoperative care of orthopedic patients, the medication reminder and guidance module 140 is used to set a reminder for taking medication and provide detailed medication instructions. It should be understood that the medication reminder and guidance module is mainly used to help patients take medication correctly and on time, and provide detailed medication instructions. This module plays a vital role in the patient's postoperative recovery process, because it helps to ensure that the patient uses the medication correctly according to the doctor's instructions, thereby promoting recovery and preventing complications. Optionally, in one embodiment of the present application, the functions of the medication reminder and guidance module include but are not limited to the following aspects: 1. Medication reminder: The module can set a timed reminder according to the doctor's prescription to ensure that the patient does not forget to take the medication. The reminder method can be in the form of a notification of a mobile phone application, a text message, a phone call, or a prompt tone on a dedicated device. 2. Detailed medication instructions: In addition to reminding patients when to take medication, the module will also provide a detailed medication guide, including information such as the name of the drug, dosage, method of taking (such as before or after meals), medication interval, and possible side effects. This helps patients correctly understand how to use the medication and reduce the possibility of misuse.

[0074] In the above-mentioned comprehensive management system for postoperative care of orthopedic patients, the nutrition advice module 150 is used to recommend a list of foods that are helpful for bone recovery. It should be understood that the role of the nutrition advice module 150 is to provide patients with dietary advice that is beneficial to bone recovery. The module analyzes the individual conditions of the patient (such as age, weight, type of surgery, etc.) to provide a personalized food list to promote the patient's recovery process. Reasonable matching of nutrients has a positive effect on bone repair, muscle recovery, wound healing and overall health. Optionally, in one embodiment of the present application, the functions of the nutrition advice module include but are not limited to the following aspects: 1. Personalized food list: According to the specific conditions of the patient, such as the type of surgery, physical condition and personal preferences, a food list rich in protein, vitamin D, calcium and other nutrients that are beneficial to bone health is generated. This helps to accelerate bone healing and muscle reconstruction. 2. Educate patients: By providing detailed dietary advice, help patients understand which foods are beneficial to bone recovery and encourage them to take these nutrients in their daily diet, thereby improving their eating habits.

[0075] In the above-mentioned comprehensive management system for postoperative care of orthopedic patients, the intelligent consultation question-and-answer module 160 is used to answer user questions based on digital humans. It should be understood that the main function of the intelligent consultation question-and-answer module is to automatically obtain and answer user questions through digital humans. This module integrates artificial intelligence and natural language processing technology, can understand user questions and find relevant answers from the knowledge base, and then generate replies in the form of text. This automated question-and-answer system helps to improve the efficiency of doctor-patient communication and reduce the workload of medical staff.

[0076] In particular, in the intelligent consultation question-and-answer module, in order to understand the user's questions and find answer records related to the user's questions, the technical concept of the present application is to obtain user questions through digital humans, and introduce data processing and semantic understanding algorithms based on artificial intelligence and natural language processing technology into digital humans to perform semantic analysis on user questions, and then find multiple answer records related to the semantics of user questions in the knowledge base, so as to generate dynamic aggregation representation information of the semantic feature distribution of answer records, and then generate question reply text with this information. In this way, in the process of postoperative care and management of orthopedic patients, digital humans can automatically obtain and reply to user questions. This intelligent question-and-answer system helps to reduce the workload of medical staff during postoperative care of orthopedic patients and improve the efficiency of doctor-patient communication, thereby helping to establish a more intelligent comprehensive management system for postoperative care of orthopedic patients.

[0077] Figure 2 This is a schematic block diagram of the intelligent consultation question and answer module in the comprehensive management system for postoperative care of orthopedic patients in an embodiment of the present application. Figure 3 This is a schematic diagram of the data flow of the intelligent consultation question-and-answer module in the comprehensive management system for postoperative care of orthopedic patients in the embodiment of the present application. Figure 2 and Figure 3 As shown, the intelligent consultation question and answer module 160 includes: a user question acquisition unit 161, which is used to acquire user questions; a user question semantic embedding coding unit 162, which is used to perform semantic embedding coding on the user questions to obtain user question semantic embedding coding vectors; an answer record matching coding unit 163, which is used to match multiple answer records from the knowledge base based on the user question semantic embedding coding vector, and semantically encode the multiple answer records respectively to obtain a set of answer record semantic coding vectors; a feature distribution semantic dynamic aggregation processing unit 164, which is used to perform feature distribution semantic dynamic aggregation on the set of answer record semantic coding vectors to obtain a semantically significant aggregation representation of answer records; a question reply text generation unit 165, which is used to generate a question reply text based on the semantically significant aggregation representation of the answer record.

[0078] In the above-mentioned comprehensive management system for postoperative care of orthopedic patients, the user question acquisition unit 161 is used to acquire user questions. Optionally, in one embodiment of the present application, the user question acquisition method includes the following: 1. Chat interface: The user enters the question through the chat window, such as through an application on a computer or smartphone. The user types the question in the input box and submits it. 2. Voice recognition: The user enters the question by voice, collects the sound through the built-in microphone or external device, and then converts it into text through voice recognition technology.

[0079] In the above-mentioned comprehensive management system for postoperative care of orthopedic patients, the user question semantic embedding coding unit 162 is used to perform semantic embedding coding on the user question to obtain a user question semantic embedding coding vector. It should be understood that by performing semantic embedding coding on the user question, the text form of the user's question can be converted into a computer-processable vector form, and the semantic embedding coding features of the user's question can be captured. Here, the semantic embedding coding features of the user's question can better reflect the essential meaning of the user's question, rather than just the literal semantics of the text. This ensures that even if the user's question words are different but the meaning is similar, the system can correctly understand the user's intention.

[0080] In the above-mentioned comprehensive management system for postoperative care of orthopedic patients, the answer record matching coding unit 163 is used to match multiple answer records from the knowledge base based on the semantic embedding coding vector of the user's question, and semantically encode the multiple answer records respectively to obtain a set of answer record semantic coding vectors. In other words, using the semantics of the question asked by the user as a basis, multiple answer records related to the semantics of the user's question are matched from the knowledge base. Matching multiple answer records from the knowledge base is to increase the flexibility and coverage of the system. Since there may be many different interpretations or intentions of user questions, screening and matching multiple answers from a large knowledge base can avoid directly returning irrelevant or inaccurate answers, and improve the comprehensiveness and accuracy of the answers finally generated later.

[0081] Furthermore, the plurality of answer records are semantically encoded respectively to obtain a set of answer record semantic encoding vectors. By semantically encoding different answer records matched from the knowledge base, the semantic encoding features of these answer records can be captured, thereby forming a semantic feature representation of each answer record.

[0082] Optionally, in one embodiment of the present application, the answer record matching encoding unit 163 is used to: extract a set of template question-answer records from the knowledge base; perform semantic encoding on multiple template questions in the set of template question-answer records to obtain a set of template question semantic embedding encoding vectors; calculate the similarity between each template question semantic embedding encoding vector in the set of template question semantic embedding encoding vectors and the user question semantic embedding encoding vector to obtain a set of question semantic similarities; compare each question semantic similarity in the set of question semantic similarities with a predetermined threshold to obtain the multiple answer records.

[0083] In the above-mentioned comprehensive management system for postoperative care of orthopedic patients, the feature distribution semantic dynamic aggregation processing unit 164 is used to perform feature distribution semantic dynamic aggregation on the set of answer record semantic coding vectors to obtain a significant aggregation representation of answer record semantics. It should be understood that each answer record semantic coding vector in the set of answer record semantic coding vectors respectively contains each answer record semantic feature related to the user's question semantics retrieved and generated from the knowledge base, and each answer record semantic feature is a semantic particle. When participating in the subsequent semantic feature aggregation process to obtain a more sufficient and accurate answer record semantic aggregation representation, these different answer record semantic particles will have different importance and contribution to the subsequent reply text generation task. Based on this, in the technical solution of the present application, the set of answer record semantic coding vectors is further subjected to feature distribution semantic dynamic aggregation to obtain a significant aggregation representation of answer record semantics. In particular, here, the process of feature distribution semantic dynamic aggregation is modeled based on the interaction between each answer record semantic particle through a concept similar to a force field. This force field not only considers the direct spatial relationship between the semantics of each answer record (such as Euclidean distance), but also combines the properties of each answer record semantics itself (such as static energy factor) and the overall semantic structure of the entire answer record semantic set. In this way, the importance of each answer record semantic feature in the semantic aggregation process can be dynamically adjusted, and the information can be integrated into a more compact and representative answer record semantic aggregation representation according to a specific aggregation strategy, that is, the answer record semantic saliency aggregation representation vector.

[0084] Optionally, in one embodiment of the present application, Figure 4 This is a schematic block diagram of a feature distribution semantic dynamic aggregation processing unit in the comprehensive management system for postoperative care of orthopedic patients according to an embodiment of the present application. Figure 4 As shown, the feature distribution semantic dynamic aggregation processing unit 164 is used to perform feature distribution semantic dynamic aggregation on the set of answer record semantic coding vectors to obtain a significant aggregation representation of answer record semantics, including: a static energy factor calculation subunit 1641, used to calculate the static energy factor of each answer record semantic coding vector in the set of answer record semantic coding vectors to obtain a set of answer record semantic static energy factors; a feature significant semantic aggregation processing subunit 1642, used to perform feature significant semantic aggregation on the set of answer record semantic coding vectors based on the set of answer record semantic static energy factors to obtain the significant aggregation representation of answer record semantics;

[0085] Optionally, in one embodiment of the present application, the static energy factor calculation subunit 1641 is used to: calculate the mean and variance of the answer record semantic coding vector to obtain the answer record semantic mean and the answer record semantic variance; subtract the answer record semantic coding vector from the answer record semantic mean by position, and calculate the fourth power of each position of the subtracted feature vector to obtain the answer record semantic difference modulation vector; calculate the expected value of the answer record semantic difference modulation vector to obtain the answer record semantic expectation value; divide the answer record semantic expectation value by the square of the answer record semantic variance, and input the divided value into the sigmoid function to obtain the answer record semantic static energy factor.

[0086] Optionally, in one embodiment of the present application, the feature-significant semantic aggregation processing subunit 1642 includes: a semantic clustering initial center vector determination secondary subunit, which is used to select the answer record semantic coding vector corresponding to the largest answer record semantic static energy factor in the set of answer record semantic static energy factors as the answer record semantic clustering initial center vector; a dynamic aggregation energy factor calculation secondary subunit, which is used to calculate the dynamic aggregation energy factor based on the spatial span between each answer record semantic coding vector in the set of answer record semantic coding vectors and the answer record semantic clustering initial center vector, as well as the static energy factor of each answer record semantic coding vector and the answer record semantic clustering initial center vector. The static energy factor of the initial center vector is used to calculate the dynamic aggregation energy factor of each answer record semantic coding vector in the set of answer record semantic coding vectors to obtain a set of answer record semantic dynamic aggregation energy factors; a gated mask processing secondary sub-unit is used to input the set of answer record semantic dynamic aggregation energy factors into the gated mask unit to obtain a set of answer record semantic dynamic aggregation weight factors; a weighted sum processing secondary sub-unit is used to use the set of answer record semantic dynamic aggregation weight factors as weighted weights, calculate the weighted sum of the set of answer record semantic coding vectors to obtain an answer record semantic significant aggregation representation vector as the answer record semantic significant aggregation representation.

[0087] Optionally, in one embodiment of the present application, the dynamic aggregation energy factor calculation secondary sub-unit is used to: multiply the static energy factor of the answer record semantic coding vector and the static energy factor of the answer record semantic clustering initial center vector by a first weighting parameter to obtain a first answer record semantic dynamic aggregation energy factor; multiply the square of the spatial span between the answer record semantic coding vector and the answer record semantic clustering initial center vector by a second weighting parameter to obtain a second answer record semantic dynamic aggregation energy factor; divide the first answer record semantic dynamic aggregation energy factor by the second answer record semantic dynamic aggregation energy factor to obtain the answer record semantic dynamic aggregation energy factor.

[0088] Optionally, in one embodiment of the present application, the gated mask processing secondary sub-unit is used to: normalize the set of answer record semantic dynamic aggregation energy factors to obtain a set of normalized answer record semantic dynamic aggregation energy factors; mask the set of normalized answer record semantic dynamic aggregation energy factors to obtain a set of answer record semantic dynamic aggregation weight factors; wherein, in response to each normalized answer record semantic dynamic aggregation energy factor in the set of normalized answer record semantic dynamic aggregation energy factors being greater than a predetermined threshold, the normalized answer record semantic dynamic aggregation energy factors greater than the predetermined threshold are set to the original value, and the rest are set to zero, so as to obtain the set of answer record semantic dynamic aggregation weight factors.

[0089] Optionally, in one embodiment of the present application, the set of answer record semantic dynamic aggregation energy factors is normalized to obtain a set of normalized answer record semantic dynamic aggregation energy factors, including: calculating an exponential function value with the natural constant e as the base and the negative of each answer record semantic dynamic aggregation energy factor in the set of answer record semantic dynamic aggregation energy factors as the exponent to obtain a set of answer record semantic dynamic aggregation energy index factors; after adding the set of answer record semantic dynamic aggregation energy index factors to the constant one by position, calculating the reciprocal of each answer record semantic dynamic aggregation energy index modulation factor in the obtained set of answer record semantic dynamic aggregation energy index modulation factors to obtain the set of normalized answer record semantic dynamic aggregation energy factors.

[0090] In summary, in the embodiment of the present application, the set of semantic encoding vectors of the answer record is subjected to feature distribution semantic dynamic aggregation using the following semantic dynamic aggregation formula to obtain the semantic significant aggregation representation of the answer record;

[0091] Among them, the semantic dynamic aggregation formula is:

[0092] ;

[0093] ;

[0094] ;

[0095] ;

[0096] ;

[0097] ;

[0098] ;

[0099] ;

[0100] ;

[0101] in, a set of semantic encoding vectors for recording the answer, are the first, second, ..., and third in the set of semantic encoding vectors of the answer record. ,..., answer record semantic encoding vector, It is The answer records the feature value of each position in the semantic encoding vector. To calculate the expected value, and They are The mean and variance of yes function, yes The corresponding answer records the semantic static energy factor, To select the maximum value corresponding to value, For the Answers record semantic static energy factor, is the maximum matching value, is the initial center vector of the answer record semantic clustering, It is answer record semantic encoding vector, yes The corresponding answer records the semantic static energy factor, express and The space span between and are the first weighting parameter and the second weighting parameter respectively, for The corresponding answer records the semantic dynamic aggregation energy factor, yes The corresponding normalized answer records the semantic dynamic aggregation energy factor, For masking, is a predetermined threshold, yes The corresponding answer record semantic dynamic aggregation weight factor, recording the number of vectors in the set of semantic encoding vectors for the answer, is a semantically significant aggregate representation vector of the answer record.

[0102] Specifically, in the process of dynamic aggregation of feature distribution semantics, first, for each vector in the set of input answer record semantic encoding vectors, its stability or importance in the feature space is evaluated based on a predefined metric function, and a numerical representation is given, namely, the answer record semantic static energy factor. In this way, the status and importance of each answer record semantic feature in its data distribution and the overall answer record semantics can be quantified, providing a basis for the subsequent selection of the initial clustering center. At the same time, by identifying those answer semantic feature points that are relatively isolated or have a high local influence, it is helpful to discover the potential data cluster core. In this way, the uniqueness and influence of each answer record semantics in the feature space can be quantified, providing a basis for the subsequent aggregation process. More specifically, in the above-mentioned static energy factor calculation process, in order to understand the distribution of the input answer record semantic encoding vector in the feature space, the mean and variance of the answer record semantic encoding vector set are calculated. The mean reflects the overall trend, while the variance reflects the degree of discreteness between data points. The answer record semantic encoding vector is subtracted element by element from the answer record semantic mean, and the fourth power of the obtained difference vector is calculated to obtain the answer record semantic difference modulation vector. This step helps to highlight features that differ greatly from the average, that is, to emphasize abnormal or significant answer record semantic features. The expected value of the answer record semantic difference modulation vector is calculated to obtain the answer record semantic expectation value. This is to further quantify the uniqueness and influence of the answer record in the feature space. The answer record semantic expectation value is divided by the square of the answer record semantic variance, and the result is input into the sigmoid function to finally obtain the answer record semantic static energy factor. The Sigmoid function is used to quantify the uniqueness and influence of each answer record in the feature space.

[0103] Next, the answer record semantic encoding vector corresponding to the largest answer record semantic static energy factor in the set of answer record semantic static energy factors is selected as the initial center vector of answer record semantic clustering. That is, a reliable starting point is determined for iterative clustering operations. This point should represent the most representative part of the answer record semantic set, which can serve as the basis for subsequent clustering to ensure the rationality of the clustering results. This enhances the clustering process's ability to capture the true semantic set structure and reduces the adverse effects caused by random initialization.

[0104] Next, based on the spatial span between the semantic encoding vectors of each answer record and the initial center vector of the semantic cluster of the answer record, as well as the static energy factor of the semantic encoding vectors of each answer record and the static energy factor of the initial center vector of the semantic cluster of the answer record, the dynamic aggregation energy factor of each answer record semantic encoding vector is further derived. In particular, the dynamic aggregation energy factor takes into account not only the semantic proximity of the answer records in terms of position, but also the importance difference of the intrinsic attributes of the semantics of the answer records presented at each point, aiming to more accurately capture the semantic interactions between answer records, so as to measure the semantic importance and contribution of each answer record semantics in the overall collection and subsequent feature aggregation process.

[0105] Next, the set of answer record semantic dynamic aggregation energy factors is input into the gated mask unit, which automatically adjusts the weight of each answer record semantic dynamic aggregation energy factor based on the learned knowledge. In this way, by nonlinearly transforming and screening the original energy factors, the model's attention to the key answer record text semantic information is enhanced, while suppressing the irrelevant parts, so that the key answer semantics related to the subsequent task is given a higher weight in the aggregation process.

[0106] Finally, the obtained set of answer record semantic dynamic aggregation weight factors is used as weighting coefficients to calculate the weighted sum of the set of answer record semantic encoding vectors, so as to combine all answer record semantic feature information, and at the same time, appropriately emphasize or weaken the feature expression according to the semantic importance of each answer, so as to generate a more concise and representative answer record semantic significant aggregation representation vector, which contains the main semantic information of the original answer record and removes redundant or irrelevant parts, thereby improving the accuracy and efficiency of digital humans in answering questions.

[0107] In the above-mentioned comprehensive management system for postoperative care of orthopedic patients, the question reply text generation unit 165 is used to generate a question reply text based on the semantically significant aggregation representation of the answer record. Optionally, in one embodiment of the present application, the question reply text generation unit 165 is used to: input the semantically significant aggregation representation vector of the answer record into an answer generator based on a large language model to obtain a question reply text. In other words, the significant aggregation feature representation between the semantics of multiple answer records is used as the basis for answer generation, thereby automatically generating a question reply text. In this way, in the process of postoperative care and management of orthopedic patients, digital humans can automatically obtain and reply to users' questions. This intelligent question and answer system helps to reduce the workload of medical staff during postoperative care of orthopedic patients and improve the efficiency of doctor-patient communication, thereby helping to establish a more intelligent comprehensive management system for postoperative care of orthopedic patients.

[0108] Preferably, the set of answer record semantic encoding vectors has differences in encoding feature distribution brought about by source semantic differences of multiple answer records after semantic encoding. In this way, during dynamic aggregation of semantic distribution fields based on the imitation field, the imitation field feature distribution under the difference in encoding feature distribution will have different field dynamic aggregation modes, causing the answer record semantically significant aggregation representation vector to have a complex aggregation space structure. Therefore, when the answer record semantically significant aggregation representation vector is input into an answer generator based on a large language model to obtain a question reply text, it is expected to improve the generative regression convergence effect of the answer record semantically significant aggregation representation vector under the complex aggregation space structure.

[0109] Therefore, the present application considers optimizing the semantically significant aggregate representation vector of the answer record when the semantically significant aggregate representation vector of the answer record is input into the answer generator based on the large language model. The optimization process specifically includes:

[0110] Determine the mean and variance of the feature set composed of the feature values ​​of the semantically significant aggregation representation vector of the answer record, and divide the mean by the variance to obtain a probability weight ,in, and denote the mean and variance respectively, represents the probability weight;

[0111] The first intermediate feature vector is obtained by multiplying the semantically significant aggregation representation vector of the answer record by the inverse of its maximum eigenvalue. ,in, represents the semantically significant aggregate representation vector of the answer record, represents the maximum eigenvalue in the semantically significant aggregation representation vector of the answer record, It means point multiplication by position. represents the first intermediate eigenvector;

[0112] The first intermediate feature vector is added with the probability weight and then the logarithm with base 2 is calculated to obtain the second intermediate feature vector ,in, represents the first intermediate eigenvector, represents the probability weight, Indicates adding by position point. represents the second intermediate eigenvector;

[0113] After subtracting the first intermediate eigenvector from the probability weight point, the inverse of each eigenvalue is calculated to obtain a third intermediate eigenvector ,in, represents the first intermediate eigenvector, represents the probability weight, It means to subtract by position point. represents the third intermediate eigenvector;

[0114] Finally, the second intermediate feature vector is added to the third intermediate feature vector to obtain the optimized answer record semantically significant aggregate representation vector ,in, represents the second intermediate eigenvector, represents the third intermediate eigenvector, Indicates adding by position point. The answer record represented by the optimized semantically salient aggregate representation vector.

[0115] That is, considering that when the weight matrix of the generative probability regression acts on the semantically significant aggregation representation vector of the answer record, the uncertainty of the weight stimulus parameter of the weight matrix caused by the generative regression offset of the semantically significant aggregation representation vector of the answer record causes the posterior probability density inference of the semantically significant aggregation representation vector of the answer record to be missing through the weight matrix, which will affect the accuracy of the probability regression result. The above optimization process simulates the mesoscale interaction structure under probability constraints between the eigenvalue scale and the eigenvector scale of the semantically significant aggregation representation vector of the answer record through the probability statistical characteristics of the semantically significant aggregation representation vector of the answer record, so as to construct a bidirectional hidden variable motif based on the mesoscale short sequence relative to the probability statistical value to perform mesoscale bidirectional migration, and perform posterior recovery based on the sequence constraint on the interactive information, so as to improve the convergence effect of the semantically significant aggregation representation vector of the answer record in the probability density domain, thereby improving the text quality of the obtained question reply text. In this way, in the process of postoperative care and management of orthopedic patients, digital humans can automatically obtain and reply to users' questions. This intelligent question-answering system helps to reduce the workload of medical staff when performing postoperative care for orthopedic patients.

[0116] In summary, the comprehensive management method for postoperative care of orthopedic patients according to the embodiment of the present application is explained, which obtains user questions through digital humans, and introduces data processing and semantic understanding algorithms based on artificial intelligence and natural language processing technology into the digital humans to perform semantic analysis on user questions, and then searches for multiple answer records related to the semantics of user questions in the knowledge base, thereby generating dynamic aggregation representation information of the semantic feature distribution of answer records, and then using this information to generate question reply text. In this way, in the process of postoperative care and management of orthopedic patients, digital humans can automatically obtain and reply to user questions. This intelligent question-and-answer system helps to reduce the workload of medical staff during postoperative care of orthopedic patients and improve the efficiency of doctor-patient communication, thereby helping to establish a more intelligent comprehensive management system for postoperative care of orthopedic patients.

[0117] Figure 5 Schematic flow chart of the comprehensive management method for postoperative care of orthopedic patients according to the embodiment of the present application. Figure 5 As shown, the comprehensive management method for postoperative care of orthopedic patients includes: S1, managing basic patient information and surgical information; S2, managing preoperative examination reports, surgical records and postoperative recovery tracking; S3, customizing personalized rehabilitation training plans according to the patient's condition and regularly evaluating the rehabilitation training plans; S4, setting medication reminders and providing detailed medication instructions; S5, recommending a list of foods that help bone recovery; S6, answering user questions based on digital humans.

[0118] Here, those skilled in the art will appreciate that the specific operations of each step in the above-mentioned comprehensive management method for postoperative care of orthopedic patients have been described in the above reference. Figures 1 to 4 The invention has been described in detail in the description of the comprehensive management system for postoperative care of orthopedic patients, and therefore, its repeated description will be omitted.

[0119] The basic principle of the present invention is described above in conjunction with specific embodiments. However, it should be pointed out that the advantages, strengths, effects, etc. mentioned in the present invention are only examples and not limitations, and it cannot be considered that these advantages, strengths, effects, etc. must be possessed by each embodiment of the present invention. In addition, the specific details of the above embodiments are only for the purpose of illustration and facilitation of understanding, rather than limitation, and the above details do not limit the present invention to being implemented by adopting the above specific details.

[0120] In the above embodiments, the description of each embodiment has its own emphasis. For parts that are not described or recorded in detail in a certain embodiment, please refer to the relevant description of other embodiments. In the several embodiments provided by the present invention, it should be understood that the disclosed devices and methods can be implemented in other ways. For example, the device embodiments described above are only schematic. For example, the module division is only a logical function division, and there may be other division methods in actual implementation. The modules described as separate components may or may not be physically separated, and the components displayed as modules may or may not be physical units, that is, they may be located in one place, or they may be distributed on multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the scheme of this embodiment.

[0121] In addition, each functional module in each embodiment of the present invention may be integrated into one processing unit, each unit may exist physically separately, or two or more units may be integrated into one unit. The above-mentioned integrated unit may be implemented in the form of hardware or in the form of hardware plus software functional modules.

[0122] It will be apparent to those skilled in the art that the invention is not limited to the details of the exemplary embodiments described above and that the invention can be implemented in other specific forms without departing from the spirit or essential features of the invention. Therefore, the embodiments should be considered exemplary and non-limiting in all respects, and the scope of the invention is defined by the appended claims rather than the foregoing description, and it is intended that all variations falling within the meaning and scope of the equivalent elements of the claims be included in the invention. Any reference to a figure in a claim should not be considered as limiting the claim to which it relates.

[0123] In addition, it is obvious that the word "comprising" does not exclude other units or steps, and the singular does not exclude the plural. Multiple units stated in the device claim can also be implemented by one unit through software or hardware.

[0124] Finally, it should be noted that the above description has been given for the purpose of illustration and description. In addition, the above embodiments are only used to illustrate the technical solution of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, it should be understood by those skilled in the art that the technical solution of the present invention can be modified or replaced by equivalents without departing from the spirit and scope of the technical solution of the present invention.

Claims

1. A comprehensive management system for postoperative care of orthopedic patients, characterized in that: include: Personal information management module, used to manage basic patient information and surgical information; Health record management module, used to manage preoperative examination reports, surgical records and postoperative recovery tracking; A rehabilitation plan formulation and tracking module is used to customize a personalized rehabilitation training plan according to the patient's condition and regularly evaluate the rehabilitation training plan; Medication reminder and guidance module, used to set medication reminders and provide detailed medication instructions; Nutritional advice module to recommend a list of foods that will aid bone recovery; Intelligent consultation question-and-answer module, used to answer user questions based on digital humans; The intelligent consultation question-answering module includes: A user question acquisition unit, used to acquire user questions; A user question semantic embedding coding unit, used for performing semantic embedding coding on the user question to obtain a user question semantic embedding coding vector; An answer record matching encoding unit, used to match multiple answer records from a knowledge base based on the user question semantic embedding encoding vector, and semantically encode the multiple answer records respectively to obtain a set of answer record semantic encoding vectors; A feature distribution semantic dynamic aggregation processing unit, used for performing feature distribution semantic dynamic aggregation on the set of answer record semantic coding vectors to obtain a semantically significant aggregation representation of the answer record, comprising: a static energy factor calculation subunit, used for calculating the static energy factor of each answer record semantic coding vector in the set of answer record semantic coding vectors to obtain a set of answer record semantic static energy factors; a feature significant semantic aggregation processing subunit, used for performing feature significant semantic aggregation on the set of answer record semantic coding vectors based on the set of answer record semantic static energy factors to obtain the semantically significant aggregation representation of the answer record; A question reply text generating unit, configured to generate a question reply text based on the semantically significant aggregate representation of the answer record; Wherein, the static energy factor calculation subunit is used for: Calculating the mean and variance of the answer record semantic encoding vector to obtain the answer record semantic mean and the answer record semantic variance; Subtracting the answer record semantic encoding vector from the answer record semantic mean by position, and calculating the fourth power of each position of the subtracted feature vector to obtain the answer record semantic difference modulation vector; Calculating the expected value of the answer record semantic difference modulation vector to obtain the answer record semantic expected value; Dividing the answer record semantic expected value by the square of the answer record semantic variance, and inputting the value obtained by the division into a sigmoid function to obtain the answer record semantic static energy factor; The feature-significant semantic aggregation processing subunit includes: The initial center vector of the semantic clustering determines a secondary subunit, which is used to select the answer record semantic encoding vector corresponding to the largest answer record semantic static energy factor in the set of answer record semantic static energy factors as the initial center vector of the answer record semantic clustering; A dynamic aggregation energy factor calculation secondary subunit is used to calculate the dynamic aggregation energy factor of each answer record semantic coding vector in the set of answer record semantic coding vectors based on the spatial span between each answer record semantic coding vector in the set of answer record semantic coding vectors and the initial center vector of the answer record semantic clustering, as well as the static energy factor of each answer record semantic coding vector and the static energy factor of the initial center vector of the answer record semantic clustering, so as to obtain a set of answer record semantic dynamic aggregation energy factors; A gated mask processing secondary subunit, used for inputting the set of answer record semantic dynamic aggregation energy factors into a gated mask unit to obtain a set of answer record semantic dynamic aggregation weight factors; The weighted and processed secondary sub-unit is used to calculate the weighted sum of the set of answer record semantic encoding vectors using the set of answer record semantic dynamic aggregation weight factors as weighted weights to obtain the answer record semantic significant aggregation representation vector as the answer record semantic significant aggregation representation.

2. The comprehensive management system for postoperative care of orthopedic patients according to claim 1 is characterized in that: The answer record matching encoding unit is used to: Extracting a set of template question-answer records from the knowledge base; Semantically encoding multiple template questions in the set of template question-answer records to obtain a set of template question semantic embedding encoding vectors; Calculating the similarity between each template question semantic embedding coding vector in the set of template question semantic embedding coding vectors and the user question semantic embedding coding vector to obtain a set of question semantic similarities; Each question semantic similarity in the set of question semantic similarities is compared with a predetermined threshold to obtain the plurality of answer records.

3. The comprehensive management system for postoperative care of orthopedic patients according to claim 2 is characterized in that: The dynamic aggregation energy factor calculation secondary subunit is used to: Multiplying the static energy factor of the answer record semantic encoding vector and the static energy factor of the answer record semantic clustering initial center vector by a first weighting parameter to obtain a first answer record semantic dynamic aggregation energy factor; Multiplying the square of the spatial span between the answer record semantic encoding vector and the answer record semantic clustering initial center vector by a second weighting parameter to obtain a second answer record semantic dynamic aggregation energy factor; The first answer record semantic dynamic aggregation energy factor is divided by the second answer record semantic dynamic aggregation energy factor to obtain the answer record semantic dynamic aggregation energy factor.

4. The comprehensive management system for postoperative care of orthopedic patients according to claim 3 is characterized in that: The gated mask processing secondary subunit is used for: Normalizing the set of answer record semantic dynamic aggregation energy factors to obtain a set of normalized answer record semantic dynamic aggregation energy factors; Masking the set of normalized answer record semantic dynamic aggregation energy factors to obtain the set of answer record semantic dynamic aggregation weight factors; Among them, in response to each normalized answer record semantic dynamic aggregation energy factor in the set of normalized answer record semantic dynamic aggregation energy factors being greater than a predetermined threshold, the normalized answer record semantic dynamic aggregation energy factors greater than the predetermined threshold are set to the original value, and the rest are set to zero to obtain the set of answer record semantic dynamic aggregation weight factors.

5. The comprehensive management system for postoperative care of orthopedic patients according to claim 4 is characterized in that: The set of answer record semantic dynamic aggregation energy factors is normalized to obtain a set of normalized answer record semantic dynamic aggregation energy factors, including: Calculate an exponential function value with the natural constant e as the base and the negative of each answer record semantic dynamic aggregation energy factor in the set of answer record semantic dynamic aggregation energy factors as the exponent to obtain a set of answer record semantic dynamic aggregation energy exponential factors; After adding the set of answer record semantic dynamic aggregation energy index factors and the constant one positionally, the inverse of each answer record semantic dynamic aggregation energy index modulation factor in the set of answer record semantic dynamic aggregation energy index modulation factors is calculated to obtain the set of normalized answer record semantic dynamic aggregation energy factors.

6. The comprehensive management system for postoperative care of orthopedic patients according to claim 5 is characterized in that: The question reply text generating unit is used to: input the semantically significant aggregation representation vector of the answer record into the answer generator based on the large language model to obtain the question reply text.

7. A comprehensive management method for postoperative care of orthopedic patients, characterized in that: include: Manage patients' basic information and surgical information; Manage preoperative examination reports, surgical records and postoperative recovery tracking; Customize individualized rehabilitation training plans according to the patient's condition and regularly evaluate the rehabilitation training plans; Set medication reminders and provide detailed medication instructions; A list of foods recommended to aid bone recovery; Answer user questions based on digital humans; Among them, answering user questions based on digital humans includes: Get user questions; Performing semantic embedding coding on the user question to obtain a user question semantic embedding coding vector; Based on the user question semantic embedding coding vector, multiple answer records are matched from the knowledge base, and the multiple answer records are semantically encoded respectively to obtain a set of answer record semantic encoding vectors; The set of answer record semantic coding vectors is subjected to feature distribution semantic dynamic aggregation to obtain answer record semantic significant aggregation representation, including: calculating the static energy factor of each answer record semantic coding vector in the set of answer record semantic coding vectors to obtain a set of answer record semantic static energy factors; based on the set of answer record semantic static energy factors, the set of answer record semantic coding vectors is subjected to feature significant semantic aggregation to obtain the answer record semantic significant aggregation representation; Generate a question reply text based on the semantically significant aggregate representation of the answer record; The step of calculating the static energy factor of each answer record semantic coding vector in the set of answer record semantic coding vectors to obtain a set of answer record semantic static energy factors includes: Calculating the mean and variance of the answer record semantic encoding vector to obtain the answer record semantic mean and the answer record semantic variance; Subtracting the answer record semantic encoding vector from the answer record semantic mean by position, and calculating the fourth power of each position of the subtracted feature vector to obtain the answer record semantic difference modulation vector; Calculating the expected value of the answer record semantic difference modulation vector to obtain the answer record semantic expected value; Dividing the answer record semantic expected value by the square of the answer record semantic variance, and inputting the value obtained by the division into a sigmoid function to obtain the answer record semantic static energy factor; The method of performing feature-significant semantic aggregation on the set of answer record semantic encoding vectors based on the set of answer record semantic static energy factors to obtain the answer record semantically significant aggregated representation includes: Selecting the answer record semantic encoding vector corresponding to the largest answer record semantic static energy factor in the set of answer record semantic static energy factors as the initial center vector of answer record semantic clustering; Based on the spatial span between each answer record semantic coding vector in the set of answer record semantic coding vectors and the initial center vector of the answer record semantic clustering, as well as the static energy factor of each answer record semantic coding vector and the static energy factor of the initial center vector of the answer record semantic clustering, the dynamic aggregation energy factor of each answer record semantic coding vector in the set of answer record semantic coding vectors is calculated to obtain a set of answer record semantic dynamic aggregation energy factors; Inputting the set of answer record semantic dynamic aggregation energy factors into a gated mask unit to obtain a set of answer record semantic dynamic aggregation weight factors; Taking the set of answer record semantic dynamic aggregation weight factors as weighted weights, the weighted sum of the set of answer record semantic encoding vectors is calculated to obtain the answer record semantic significant aggregation representation vector as the answer record semantic significant aggregation representation.

Images