Chronic kidney disease comprehensive management system and method based on artificial intelligence large model

Abstract

The invention discloses a chronic kidney disease comprehensive management system and method based on an artificial intelligence large model, and relates to the technical field of chronic kidney disease management, in the system, a model training module is used for training a preset artificial intelligence large model based on a labeling corpus about the chronic kidney disease to obtain an intelligent suggestion generation model; the packaging module is used for packaging the intelligent suggestion generation model into an embedded service; and the personalized suggestion generation module is used for deploying the embedded service to the terminal of each preset role, so that the terminal of each preset role generates personalized suggestions about the chronic kidney disease through the embedded service. According to the method, massive patient data can be quickly processed based on the intelligent suggestion generation model, personalized suggestions are quickly generated, the efficiency is greatly improved, and the requirements of more patients are met. Moreover, the system is easy to popularize widely, enables primary patients to enjoy high-quality services, and assists the management of chronic kidney diseases to be popularized.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of chronic kidney disease management, and particularly relates to a chronic kidney disease comprehensive management system and method based on an artificial intelligence large model. BACKGROUND

[0002] Currently, there are many problems in giving personalized suggestions for chronic kidney disease by relying on expert experience and manual judgment. On the one hand, the cultivation cycle of experts is long, and the number of experienced persons is scarce, which is difficult to meet the needs of a large number of patients, and often leads to that patients are difficult to obtain professional guidance in time. On the other hand, manual judgment is easily disturbed by subjective factors, and is easy to miss when dealing with complex conditions, which affects the accuracy of the suggestions and is low in efficiency. In addition, it is difficult for primary medical institutions to obtain support, which leads to that the suggestions are difficult to be widely promoted, and primary patients are difficult to benefit. SUMMARY

[0003] The technical problem solved by the present application is that the prior art is not enough, specifically for the technical problem that the traditional chronic kidney disease management process relies on expert experience and manual judgment, which is low in efficiency and difficult to be widely promoted to primary medical institutions. Specifically, a chronic kidney disease comprehensive management system and method based on an artificial intelligence large model are provided, as follows.

[0004] 1) In a first aspect, the present application provides a chronic kidney disease comprehensive management system based on an artificial intelligence large model, and the specific technical solutions are as follows:

[0005] The system comprises a model training module, an encapsulation module and a personalized suggestion generation module.

[0006] The model training module is configured to train a preset artificial intelligence large model based on an annotated corpus about chronic kidney disease, to obtain an intelligent suggestion generation model.

[0007] The encapsulation module is configured to encapsulate the intelligent suggestion generation model into an embeddable service.

[0008] The personalized suggestion generation module is configured to deploy the embeddable service to a terminal of each preset role, so that the terminal of each preset role generates personalized suggestions about chronic kidney disease through the embeddable service.

[0009] The chronic kidney disease comprehensive management system based on an artificial intelligence large model provided by the present application has the following beneficial effects:

[0010] Based on the intelligent suggestion generation model, massive patient data can be quickly processed, personalized suggestions can be quickly generated, efficiency can be greatly improved, and more patient needs can be met. Moreover, the intelligent suggestion generation model can learn a large number of cases and the latest research results, and comprehensively consider multi-dimensional information, so that the suggestions are more accurate. In terms of promotion, the system can be integrated into an electronic medical record system of a primary medical institution and the like, and is convenient for different preset roles to use, is easy to be widely promoted, enables primary patients to also enjoy high-quality services, and helps popularization of chronic kidney disease management.

[0011] On the basis of the above scheme, the chronic kidney disease comprehensive management system based on the artificial intelligence large model of the present application can be further improved as follows.

[0012] Further, the operation interface design module is further included, and the operation interface design module is used for: designing intelligent operation interfaces respectively adapted to each preset role;

[0013] The personalized suggestion generation module is further used for: receiving data about chronic kidney disease input by the corresponding preset role through each intelligent operation interface, and generating corresponding personalized suggestions about chronic kidney disease by using the embeddable service.

[0014] The beneficial effects of the above further scheme are: the data collection efficiency can be improved, the waste of medical resources can be reduced, the medical efficiency can be improved, and the suggestion quality can be continuously optimized through data accumulation.

[0015] Further, the guiding module is further included, and the guiding module is used for: guiding each preset role to input data about chronic kidney disease in an interactive manner.

[0016] The beneficial effects of the above further scheme are: the accuracy and completeness of data collection are improved; user participation and experience are enhanced; personalized data input is realized, the accuracy of medical suggestions is improved; data errors can be found and corrected in time; medical service efficiency and quality are improved through promoting doctor-patient communication.

[0017] Further, the model optimization module is further included, and the model optimization module is used for: optimizing the intelligent suggestion generation model according to feedback information of each preset role and / or newly obtained data about chronic kidney disease.

[0018] The beneficial effects of the above further scheme are: the adaptability of the intelligent suggestion generation model to complex conditions of chronic kidney disease is enhanced; the update of clinical practice and guidelines is reflected in time; the practicability and reliability of the intelligent suggestion generation model are improved; and the scientificity and rationality of medical decision are promoted.

[0019] Further, the method further comprises a labeled corpus obtaining module, which is configured to collect epidemiological survey data, disease prevention and control center data and actual cases about chronic kidney disease, and obtain a labeled corpus about chronic kidney disease after expert labeling.

[0020] The above further scheme has the beneficial effect of providing high-quality training data to improve the accuracy of personalized suggestions given by the intelligent suggestion generation model.

[0021] 2) In a second aspect, the present application further provides a chronic kidney disease comprehensive management method based on an artificial intelligence large model, and the specific technical solutions are as follows:

[0022] Training the preset artificial intelligence large model based on the labeled corpus about chronic kidney disease to obtain an intelligent suggestion generation model;

[0023] Encapsulating the intelligent suggestion generation model as an embeddable service;

[0024] Deploying the embeddable service to the terminal of each preset role, so that the terminal of each preset role generates personalized suggestions about chronic kidney disease through the embeddable service.

[0025] On the basis of the above scheme, the chronic kidney disease comprehensive management method based on an artificial intelligence large model of the present application can be further improved as follows.

[0026] Further, the method further comprises designing intelligent operation interfaces respectively adapted to each preset role;

[0027] Through each intelligent operation interface, receiving data about chronic kidney disease input by the corresponding preset role, and generating corresponding personalized suggestions about chronic kidney disease by using the embeddable service.

[0028] Further, the method further comprises guiding each preset role to input data about chronic kidney disease in an interactive manner.

[0029] Further, the method further comprises optimizing the intelligent suggestion generation model according to feedback information of each preset role and / or newly obtained data about chronic kidney disease.

[0030] Further, the method further comprises collecting epidemiological survey data, disease prevention and control center data and actual cases about chronic kidney disease, and obtaining a labeled corpus about chronic kidney disease after expert labeling.

[0031] 3) In a third aspect, the present application also provides an electronic device, the electronic device comprising a processor and a memory coupled to the processor, the memory storing at least one computer program, the at least one computer program being loaded and executed by the processor to enable the electronic device to implement any of the above-mentioned chronic kidney disease comprehensive management methods based on an artificial intelligence large model.

[0032] 4) In a fourth aspect, the present application also provides a computer readable storage medium, the computer readable storage medium storing a computer program, the computer program being executed by a processor to implement any of the above-mentioned chronic kidney disease comprehensive management methods based on an artificial intelligence large model.

[0033] It should be noted that the technical solutions of the second to fourth aspects of the present application and the corresponding possible implementation manners have the beneficial effects mentioned above for the first aspect and its corresponding possible implementation manners, which will not be repeated here. BRIEF DESCRIPTION OF DRAWINGS

[0034] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the drawings needed in the description of the embodiments of the present application will be briefly introduced as follows:

[0035] Figure 1 FIG. 1 is a structural schematic diagram of a chronic kidney disease comprehensive management system based on an artificial intelligence large model according to an embodiment of the present application;

[0036] Figure 2 FIG. 2 is a flowchart of a chronic kidney disease comprehensive management method based on an artificial intelligence large model according to an embodiment of the present application;

[0037] Figure 3 FIG. 3 is a structural schematic diagram of an electronic device according to an embodiment of the present application. DETAILED DESCRIPTION

[0038] The principles and features of the present application will be described below, and the examples are only used to explain the present application and not to limit the scope of the present application.

[0039] The technical solutions of the present application and how the technical solutions of the present application solve the above-mentioned technical problems will be described in detail below with specific embodiments. The following specific embodiments can be combined with each other, and the same or similar concepts or processes can not be described again in some embodiments. The embodiments of the present application will be described below with reference to the drawings.

[0040] As shown in FIG. 1, a chronic kidney disease comprehensive management system based on an artificial intelligence large model according to an embodiment of the present application comprises a model training module, an encapsulation module and a personalized suggestion generation module; Figure 1

[0041] ​The model training module is configured to train the preset artificial intelligence large model based on the annotated corpus about chronic kidney disease to obtain an intelligent suggestion generation model, in particular:

[0042] In S11, the annotated corpus about chronic kidney disease is parsed, and each piece of clinical information associated with chronic kidney disease and the corresponding personalized suggestion are respectively converted into an input word sequence and an output word sequence to generate a time step aligned sample pair.

[0043] Based on the annotated corpus of chronic kidney disease, the clinical information associated with chronic kidney disease and the personalized suggestion are first parsed. The clinical information is the input word sequence, and the personalized suggestion is the output word sequence. Through technical processing, a time step aligned sample pair is generated. This process involves accurate extraction, conversion and alignment of data. Finally, effective sample pairs are formed for model training to facilitate subsequent deep learning tasks based on these sample pairs, and to realize effective utilization and analysis of chronic kidney disease related data.

[0044] In S12, the input word sequence is used as a condition, and the preset artificial intelligence large model is set to predict the probability distribution of the output word sequence word by word according to the time step. The cross-entropy loss function is used to quantify the prediction bias.

[0045] First, the input word sequence is used as a condition, and the preset artificial intelligence large model receives each word in the input word sequence in turn. At each time step, the model predicts the probability distribution of the next word in the corresponding output word sequence based on the current input word and the information of the previous time step. Then, the cross-entropy loss function is used to quantify the deviation between the probability distribution predicted by the model and the probability distribution of the true output word sequence. The cross-entropy loss function calculates the difference between the two probability distributions to obtain a loss value. This loss value can reflect the accuracy of the model prediction. The smaller the loss value, the more accurate the prediction. Based on this loss value, the model can perform back propagation and parameter update to optimize the prediction performance, thereby continuously improving the prediction ability of the output word sequence during the training process.

[0046] In S13, based on the self-attention mechanism, the context vector representation of the output word sequence is iteratively calculated, and the gradient descent method is used to adjust the parameters of the preset artificial intelligence large model to learn the mapping relationship from the clinical information to the personalized suggestion.

[0047] First, based on the self-attention mechanism, the model iteratively calculates the context vector representation of the output word sequence. The self-attention mechanism allows the model to dynamically calculate attention weights between words at different positions, capturing the dependency between words regardless of their position in the sequence. This helps the model better understand the context and long-distance dependencies. Then, using gradient descent, the error signal is propagated back through the network to adjust the parameters of the pre-trained AI large model. Gradient descent calculates the gradient of the loss function with respect to the model parameters to guide the update direction of the model parameters to minimize the prediction error. During backpropagation, error signals are passed from the output layer to the input layer layer by layer to adjust the parameters of each layer.

[0048] The purpose of this process is to learn the mapping relationship from clinical information to personalized recommendations. Through continuous iteration and adjustment, the model can gradually optimize its internal representation, so that given the clinical information, it can more accurately predict the corresponding personalized recommendations. This learning process enables the model to capture complex patterns and associations in clinical data, providing valuable personalized recommendations in practical applications.

[0049] S14、When the training loss reaches the preset threshold or the verification set recommendation generation accuracy meets the clinical requirements, stop training and output the intelligent recommendation generation model.

[0050] Among them, the annotated corpus related to chronic kidney disease contains structured or semi-structured text data associated with diagnosis, treatment or management recommendations.

[0051] Among them, the pre-set intelligent recommendation generation model can be: DeepSeek R1 model, GPT-3 model, GPT-4 model or ZhiShen large model, etc.

[0052] In another implementation, the pre-set intelligent recommendation generation model includes a medical entity perception embedding layer, an entity perception position encoding layer, a multi-specialist clinical attention layer, a clinical decision routing layer, a cross-modal fusion decoder, and a medical safety constraint output layer. Specifically:

[0053] 1) The medical entity perception embedding layer includes a word embedding sub-layer and an entity embedding sub-layer. The word embedding sub-layer converts each word in the input text into a high-dimensional vector. Then, the entity embedding sub-layer identifies medical entities (such as disease names, drug names, test indicators, etc.) in the text and converts their corresponding entity information into vectors. Finally, the word vectors and entity vectors are concatenated or fused to form embedding vectors containing medical entity information.

[0054] The word embedding sub-layer includes a linear layer, an activation function layer, and a normalization layer. The linear layer of the word embedding sub-layer uses learnable weights to capture the semantic information of the words, laying the foundation for subsequent processing. The activation function layer of the word embedding sub-layer then performs a nonlinear transformation on the word embedding vectors, which enables the model to learn more complex relationships between words. The normalization layer of the word embedding sub-layer ensures that the numerical scale of the word embedding vectors is appropriate, which helps to speed up model training and improve stability. The entity embedding sub-layer relies on a pre-trained entity recognition model (which can be trained using a neural network) to scan the text and identify medical entities such as disease names, drug names, and test indicators. The linear layer of the entity embedding sub-layer maps each identified entity index to an entity embedding space, generating entity vectors. The entity activation function layer of the entity embedding sub-layer introduces nonlinearity again, mining potential relationships between entities. The normalization layer of the entity embedding sub-layer ensures that the scale of the entity vectors is appropriate. Finally, the word vectors and entity vectors are integrated through simple concatenation or more complex fusion methods (such as weighted summation), forming embedding vectors that contain medical entity information. Such embedding vectors not only retain the semantics of the words but also enhance the perception of key entities in the medical field, providing rich feature information for subsequent model processing.

[0055] 2) The entity-aware position encoding layer includes a position embedding sub-layer. The position embedding sub-layer adds position information to each embedding vector, enabling the model to perceive the order of words in the text. Position information is usually represented by a sine, cosine function, or a learnable position embedding vector. The position embedding vector is added to the entity-aware embedding vector to obtain an entity-aware embedding vector containing position information.

[0056] The position embedding sub-layer includes a linear layer, an activation function layer, and a normalization layer. After receiving the embedding vector sequence (including the result of the fusion of word vectors and entity vectors) from the entity embedding sub-layer, the position embedding sub-layer maps the position information of the words to a vector space with the same dimension as the embedding vector through the linear layer. Position information can usually be represented by a fixed function (such as a sine function or a cosine function) or a learnable position embedding vector. The activation function layer performs a nonlinear transformation on the generated position embedding vector, enhancing the interaction between position information and embedding vectors. The normalization layer normalizes the position embedding vector to ensure that its numerical range is consistent with that of the entity-aware embedding vector. Finally, the normalized position embedding vector is added to the entity-aware embedding vector to obtain an entity-aware embedding vector containing position information, enabling the model to perceive the order of words in the text.

[0057] 3) The multi-expert clinical attention layer consists of multiple attention heads and a mixture-of-experts sublayer. These heads process the input vector in parallel, each focusing on a different feature dimension. The mixture-of-experts sublayer synthesizes the outputs of each attention head, simulating the fusion of opinions from different experts. In this way, the model is able to capture clinical information in the text from multiple perspectives.

[0058] Each attention head consists of a linear layer and an activation function layer. The linear layer maps the input vector to query, key, and value vectors. The linear layer learns feature representations of different dimensions. The activation function layer applies nonlinear transformations to these vectors, enhancing the model's ability to express features. Multiple attention heads process in parallel, each focusing on a different feature dimension of the input vector, thereby capturing text information from multiple perspectives.

[0059] The expert mixture sublayer includes a linear layer, an activation function layer, and a normalization layer. The expert mixture sublayer receives the output vectors of multiple attention heads. The linear layer of the expert mixture sublayer performs a linear transformation and weighted summation on the output of each attention head, calculates the fusion weight of each attention head, and determines the importance of each attention head in the overall information extraction. Next, the activation function layer performs a nonlinear transformation on the fused vector to enhance its expressive power, enabling the model to better capture clinical information in the text from multiple perspectives. Finally, the normalization layer normalizes the transformed vector to ensure the numerical stability and consistency of the output vector and the accuracy of subsequent processing.

[0060] 4) The clinical decision routing layer includes a routing algorithm sublayer and a decision sublayer. The routing algorithm sublayer assigns feature vectors to different decision branches based on their content. The decision sublayer performs specific clinical decision logic within each branch. This helps the model select the most appropriate decision path based on different clinical scenarios.

[0061] The routing algorithm sublayer consists of a linear layer and an activation function layer. After receiving a feature vector, the linear layer performs a linear transformation on the input feature vector to extract key features for decision branch selection. The activation function layer (such as Sigmoid or Softmax) maps the transformed features to the selection probabilities of different branches. Based on the output probability distribution, the decision branch to which the feature vector is assigned is determined.

[0062] The decision sublayer consists of a linear layer, a fully connected layer, and an activation function layer. After receiving the input feature vector, the linear layer performs a linear transformation on the input to extract key features. Based on the clinical decision logic, the fully connected layer maps the input features to the decision space and learns the mapping between features and decisions. The activation function layer converts the output into a probability distribution of the decision outcome and selects the appropriate decision path based on this distribution.

[0063] 5) The cross-modal fusion decoder includes a cross-modal fusion sublayer and a decoding sublayer. The cross-modal fusion sublayer fuses features from different modalities (such as text, image, structured data, etc.). The decoding sublayer gradually decodes the fused feature vector to generate a representation for the subsequent output layer.

[0064] The cross-modal fusion sublayer includes a linear layer, a fully connected layer, an activation function layer, and a normalization layer. After receiving feature inputs from different modalities (such as text, image, structured data, etc.), the linear layer maps the features of each modality to a unified feature space, aligning the dimensions and scales. Then, the fully connected layer combines and transforms the mapped features to capture the correlation and interaction information between modalities. The activation function layer introduces nonlinearity, enhancing the model's ability to model complex feature relationships. The normalization layer ensures the stability of the fused feature values, improving the training stability and output reliability of the model.

[0065] The decoding sublayer includes a linear layer, a fully connected layer, and an activation function layer. After receiving the fused feature vector, the linear layer maps it to a suitable dimensional space to adjust the feature expression form. The fully connected layer captures the complex correlation between features, realizing the conversion from abstract features to specific outputs. The activation function layer introduces nonlinearity, making the decoded features better match the output requirements and generate a representation for the subsequent output layer.

[0066] 6) The medical safety constraint output layer includes a safety constraint sublayer and an output sublayer. The safety constraint sublayer conducts risk assessment and compliance checks on the generated recommendations to ensure that the output meets medical safety standards and clinical guidelines. The output sublayer presents the final personalized recommendations to the pre-set role in natural language form.

[0067] The safety constraint sublayer includes a linear layer, a fully connected layer, and an activation function layer. After receiving the generated recommendation representation vector, the linear layer maps the vector to a feature space related to medical safety rules, extracting risk and compliance features. The fully connected layer combines and judges the features based on medical safety standards and clinical guidelines. The activation function layer converts the output into risk assessment results and compliance indications, based on which the recommendation representation is adjusted to ensure that the output recommendations meet safety and compliance requirements.

[0068] The output sub-layer includes a linear layer, a fully connected layer, an activation function layer and an output layer. After the output sub-layer receives the suggestion representation vector processed by the security constraint sub-layer, the linear layer maps the vector to a dimension corresponding to the vocabulary table. The fully connected layer further processes to generate the prediction probability of each word. The activation function layer (such as Softmax) converts the prediction probability into a normalized probability distribution. The output layer selects the most likely word sequence according to the probability distribution to output the final personalized medical suggestion in natural language form, ensuring that the preset role can clearly understand the information provided by the model.

[0069] The packaging module is configured to package the intelligent suggestion generation model into an embeddable service, specifically:

[0070] S21, knowledge distillation is performed on the trained intelligent suggestion generation model to generate a lightweight model that retains the personalized suggestion generation logic of the original intelligent suggestion generation model.

[0071] First, the intelligent suggestion generation model is used as a teacher model, and the generation logic of the personalized suggestion of the intelligent suggestion generation model and the corresponding probability distribution are used as guidance information, and then a lightweight model with a simpler structure is constructed as a student model, the input of which is the same as that of the teacher model. When training the student model, not only the original supervision signal, i.e., the real personalized suggestion, is used, but also the output of the teacher model is used to optimize the difference between the output of the student model and the output of the teacher model through the KL divergence loss function, so that the student model learns the implicit knowledge of the teacher model. At the same time, the cross-entropy loss function is combined to ensure that the output of the student model is consistent with the real label. Through such joint training, the student model retains the main logic and performance of the teacher model while maintaining a small model size. Finally, the obtained lightweight model can run efficiently in a resource-limited environment while providing similar personalized suggestion generation capabilities as the original model.

[0072] S22, a medical compliance sandbox is pre-built in a target device (which can be a server, etc.), the lightweight model is deployed in the medical compliance sandbox, and an embeddable service is generated.

[0073] A medical compliance sandbox is pre-built on the target device, which has functions such as data encryption, access control, and audit tracking to ensure the security and compliance of medical data. The lightweight model is imported into the sandbox in the form of a binary file or a model library, and encapsulated using the standardized interface provided by the medical compliance sandbox. The model calling interface is connected to the operating system or the application programming interface (API) of the target device, so that the model can be accessed by the application program on the target device. At the same time, the resource limit parameters of the medical compliance sandbox are configured to ensure that the lightweight model does not excessively occupy device resources during operation. Finally, a embeddable service is output, which can be seamlessly integrated into electronic medical record systems, medical decision support systems and other medical software, providing a function module for preset roles to call lightweight models to generate personalized recommendations in a safe and compliant environment, realizing intelligent auxiliary decision support in medical scenarios.

[0074] In another implementation, the intelligent recommendation generation model is directly deployed in the medical compliance sandbox to generate an embeddable service.

[0075] The personalized recommendation generation module is configured to deploy the embeddable service to each terminal of a preset role, so that each terminal of a preset role generates personalized recommendations for chronic kidney disease through the embeddable service, specifically:

[0076] S31, bind the calling permission of the lightweight model with the clinical role authentication through the access control engine embedded in the medical compliance sandbox, and activate the corresponding clinical knowledge subgraph in the lightweight model based on different preset roles to generate a role-specific service package.

[0077] The access control engine embedded in the medical compliance sandbox binds the calling permission of the lightweight model with the clinical role authentication. When different preset roles (such as doctors, nurses, and patients) request to call the model, the access control engine verifies their identity and permission. After identity verification, the engine activates the clinical knowledge subgraph corresponding to the role in the lightweight model. The clinical knowledge subgraph is a knowledge module designed specifically for a particular role in the model, containing clinical information and logic related to the role. After activation, the model will generate role-specific personalized recommendations and services based on the subgraph. Finally, the system integrates these personalized services into a role-specific service package and provides it to the corresponding role to meet its specific clinical needs.

[0078] S32, bind the target device ID, role permission level, and access key through digital certificates on each terminal of a preset role, so that different preset roles call the corresponding role-specific service package through the access key to generate personalized recommendations for chronic kidney disease, specifically:

[0079] The access control engine embedded in the medical compliance sandbox is bound with the invocation permission of the lightweight model, and is associated with the clinical role authentication. When a preset role requests to invoke the lightweight model, the access control engine first verifies the identity information of the preset role and matches it with the preset clinical role. According to the matching result, the access control engine activates the corresponding clinical knowledge subgraph in the lightweight model. The lightweight model is internally divided into multiple clinical knowledge subgraphs, each of which contains the knowledge and logic required by a specific clinical role. These clinical knowledge subgraphs may cover clinical knowledge of different departments or different professional fields. Activating a specific subgraph means that the model will only use the knowledge in this subgraph to generate suggestions, thereby ensuring that the output suggestions are highly relevant to the clinical role and needs of the preset role. Finally, the role-specific service package generated in this way is a model service that contains the knowledge and logic required by a specific clinical role. This service package can be embedded in a medical information system to provide personalized suggestions and decision support for preset roles of different clinical roles. This process ensures the accuracy and relevance of medical recommendations, while ensuring data security and compliance through the access control engine.

[0080] The personalized suggestions include at least one of diet management suggestions, psychological management suggestions, sleep management suggestions, and treatment plans.

[0081] The data in the annotated corpus further includes disease symptoms, diagnosis results, treatment plans, diet management suggestions, psychological management suggestions, sleep management suggestions, and the like. The intelligent suggestion generation model trained based on the annotated corpus can output at least one of diet management suggestions, psychological management suggestions, sleep management suggestions, and treatment plans.

[0082] Optionally, in the above technical solution, an operation interface design module is further included, which is configured to design intelligent operation interfaces respectively adapted to each preset role, specifically:

[0083] 1) When the preset role is a doctor, the clinical guidelines and electronic medical record operation flow are parsed, the diagnosis decision tree and test index priority graph are extracted, and the test data drilling module and treatment plan comparison tool are integrated. The medical ontology visualization component is loaded, the current patient medical history timeline is automatically expanded, and the abnormal index associated personalized suggestions are highlighted, thereby obtaining the intelligent operation interface of the doctor.

[0084] In generating an intelligent operation interface for the doctor role, first parse the clinical guidelines and electronic medical record operation flow, extract the diagnosis decision tree and test index priority graph features, and provide diagnostic support for doctors. At the same time, integrate the test data drilling module to allow doctors to deeply view test data details; add a treatment plan comparison tool to help doctors compare the pros and cons of different treatment plans. In addition, load the medical ontology visualization component to visually display medical concepts and their relationships. The system automatically expands the current patient medical history timeline, clearly presents the disease development context, and highlights personalized recommendations related to abnormal indicators to help doctors quickly capture key information. Finally, integrate the above functions to form an intelligent operation interface dedicated to doctors, improving medical decision-making efficiency.

[0085] Among them, the test index priority graph is determined according to clinical guidelines and expert experience to determine the priority of test indexes, guiding doctors to focus on key indicators; the test data drilling module allows doctors to interactively and deeply view and analyze test data to obtain more detailed diagnostic information; the treatment plan comparison tool provides comparison and analysis of multiple treatment plans to support doctors to make better treatment decisions. The medical ontology visualization component displays medical concepts and their relationships in an intuitive graphical manner to help doctors better understand and apply medical knowledge.

[0086] 2) When the preset role is a patient, analyze patient education materials and follow-up data, extract medication reminders, etc.; embed medication adherence trackers and emergency symptom red alert modules; load life-like interaction components; fold complex medical jargon and dynamically generate personalized suggestion cards. Thus, the patient's intelligent operation interface is obtained.

[0087] When the preset role is a patient, first analyze patient education materials and follow-up data to extract key medication reminder information to help patients take medication on time. Then embed the medication adherence tracker to monitor the patient's medication in real time and improve treatment effectiveness. At the same time, load the emergency symptom red alert module to issue an alarm in a timely manner for dangerous symptoms that may occur to ensure patient safety. To facilitate patient understanding, the system loads life-like interaction components to communicate with patients using simple and understandable language and interface. For complex medical jargon, the system will fold and simplify information presentation. According to the patient's condition and historical data, the system dynamically generates personalized suggestion cards to provide targeted health management guidance for patients, and finally forms an intelligent operation interface dedicated to patients.

[0088] Among them, the patient education material is the education material provided to the patient about the disease knowledge, treatment process and health management, helping the patient to better understand the disease and treatment plan. The follow-up data refers to the data generated by the patient during the regular follow-up after treatment, including physical indicators, symptom changes, etc., which are used to monitor the recovery of the patient. The medication adherence tracker is a tool or functional module for monitoring and recording whether the patient takes medicine correctly according to the doctor's advice, so as to improve the treatment effect and the health status of the patient. The red early warning module for emergency symptoms is used to: when potential dangerous symptoms are monitored, an emergency alarm is sent to remind the patient to seek medical treatment in time. The life-like interaction component is a component designed for the interface and function of simplifying the interaction between the patient and the system, using simple and easy-to-understand language and intuitive operation mode, so that the patient can easily obtain and understand the information. The personalized advice card refers to the advice information dynamically generated according to the personal data and condition of the patient, providing targeted health management guidance to help the patient better self-manage.

[0089] The personalized advice generation module is also used to: through each intelligent operation interface, receive the data about chronic kidney disease input by the corresponding preset role, and generate corresponding personalized advice about chronic kidney disease by using the embeddable service.

[0090] When the preset role is a doctor or a patient, the corresponding intelligent operation interface receives the chronic kidney disease related data input by the role. The doctor inputs diagnosis information, test indicators, etc., and the patient inputs symptoms, medication feedback, etc. The interface processes the data in a structured manner and sends it to the embeddable service. The service uses a lightweight model, combined with a clinical knowledge subgraph, to analyze the input data and generate personalized advice. The doctor's advice may involve diagnosis process optimization, treatment plan adjustment, etc.; the patient's advice may include medication reminders, lifestyle adjustments, etc. Finally, the personalized advice is returned to the intelligent operation interface for the preset role to view and reference, realizing precise guidance for chronic kidney disease.

[0091] Optionally, in the above technical solution, it also includes a guidance module, which is used to guide each preset role to input data about chronic kidney disease in an interactive manner, specifically:

[0092] S101, based on the clinical guidelines for chronic kidney disease, abstracting test indicators, symptom descriptions and medication records into medical ontology nodes, and generating a dynamically expandable problem decision tree;

[0093] Based on the clinical guidelines for chronic kidney disease, the key elements such as test indicators, symptom descriptions, and medication records are abstracted as medical ontology nodes. This requires in-depth analysis of clinical guidelines to identify core concepts and their relationships. Then, using data structures such as graph databases or tree structures, these medical ontology nodes are connected to form a preliminary decision tree. The construction of the decision tree needs to follow the logical flow in the clinical guidelines, ensuring that each node represents a clinical decision point and the edges represent the decision logic. Then, through machine learning algorithms or expert system rules, the decision tree has dynamic scalability. This means that the decision tree can automatically adjust its structure, add new nodes or modify the connection between existing nodes based on new input data or updates to the clinical guidelines. At the same time, the decision tree should be able to handle the specific circumstances of different patients and provide personalized decision support. Finally, this dynamically scalable decision tree is integrated into the medical information system, allowing it to receive and process patient data in real time and provide support for clinical decision-making. This process needs to seamlessly integrate with existing medical data systems and ensure data security and compliance. Through this technical implementation, the clinical guidelines can be effectively transformed into practical decision support tools to help doctors make more accurate decisions in the diagnosis and treatment of chronic kidney disease.

[0094] S102、According to the current input content of the preset role, real-time traversal of the problem decision tree is performed, and a natural language question conforming to the preset role's cognitive level is generated through an NLG engine, wherein professional terms are automatically replaced by preset colloquial expressions;

[0095] When the preset role inputs information about chronic kidney disease, the system real-time traverses the pre-constructed problem decision tree. According to the content input by the preset role, the decision tree determines the current node and the key information to be asked next. At the same time, the system uses the NLG (Natural Language Generation) engine to generate natural language questions according to the preset role's cognitive level (such as professional doctors or ordinary patients). The NLG engine has a built-in mapping library of professional term colloquial expressions, which automatically replaces the professional terms with preset colloquial expressions when detecting professional terms. The generated question not only conforms to the understanding ability of the preset role, but also guides the provision of effective information, thereby assisting the system to make more accurate diagnoses or suggestion generation.

[0096] S103, calling the medical rule verification engine to perform logical checking on the text, voice or form data input by the preset role, and real-time feedback of visual warning;

[0097] Upon receiving the text, voice, or form data input by the pre-set role, the data is immediately transmitted to the medical rule verification engine. The engine is built-in with medical logic and data verification rules to perform logical verification on the input data. For text and voice data, the data is first structured by the voice recognition and natural language processing modules, and then verified according to medical logic together with form data. Once data inconsistencies or logical errors are found, the engine triggers the visual alert module to provide real-time feedback to the pre-set role through interface pop-ups, highlight displays, or icon prompts, ensuring the accuracy and logic of the input data and the reliability of the subsequent processing flow.

[0098] S104, dynamically mapping the verified data to the medical ontology nodes to generate a structured patient data graph, and automatically triggering the next round of question generation based on the graph integrity.

[0099] Upon receiving the data input by the pre-set role, the medical rule verification engine is first invoked to perform logical verification on the data. The verified data is dynamically mapped to the pre-defined medical ontology nodes. This mapping process is achieved through natural language processing technology, which converts unstructured data into structured form and matches it with standard terminology and concepts in the medical ontology library. Then, using graph database technology, the mapped data nodes are connected to generate a structured data graph for the patient. Each node represents a medical concept (such as symptoms, test indicators, medication records), and the edges represent the relationships between them. As more data is input, the graph dynamically expands and updates to reflect the patient's complete medical condition. By continuously evaluating the integrity of the graph, it checks for missing critical information. When the integrity reaches a pre-set threshold, the system automatically triggers the next round of question generation, using the natural language generation engine to ask follow-up questions to the pre-set role to supplement the missing parts of the graph. This process ensures the comprehensiveness and accuracy of the data, providing a solid foundation for subsequent diagnosis and treatment recommendations.

[0100] The preset role can be a community doctor, a specialist or a patient. When the preset role is a community doctor, the input data about chronic kidney disease includes basic information (name, gender, age and nationality) of the patient, medical history, symptom records (whether the patient has edema, urine volume changes, fatigue, nausea and vomiting, etc.), physical signs (measurement of vital signs such as blood pressure, heart rate and respiration, and physical examination results such as height, weight and BMI), and test results (blood test, urine test, kidney function, electrolyte test, etc.). When the preset role is a specialist, the input data about chronic kidney disease includes detailed medical history of the patient, symptom assessment results (detailed inquiry of patient symptoms such as nocturia, lower back pain, hematuria, etc., compared with and supplemented by the symptoms recorded by the community doctor), physical signs, kidney function and damage indicators (more accurate kidney function assessment indicators such as cystatin C and endogenous creatinine clearance rate, and early indicators of kidney damage such as urinary microalbumin and urinary albumin / creatinine ratio), autoimmune-related test results (anti-nuclear antibody, anti-double-stranded DNA antibody, anti-glomerular basement membrane antibody, etc. to rule out autoimmune-related kidney diseases), and kidney pathology test results; when the preset role is a patient, the input data about chronic kidney disease includes basic information (name, gender, age and nationality), symptoms (the patient needs to describe his symptoms in detail, such as the location and degree of edema, changes in urine volume, fatigue, nausea, vomiting and other discomforts), dietary habits (daily diet, including food types, water intake, salt intake, etc. Dietary habits are closely related to the development of chronic kidney disease), exercise (daily exercise frequency, intensity and type, which helps doctors assess the patient's overall health), medication adherence (whether the patient takes medication on time, misses medication or stops medication on his own), treatment response (whether adverse reactions such as dizziness and nausea occur after medication, and whether symptoms improve), and self-monitoring data (home blood pressure monitoring data and urine routine monitoring data).

[0101] Optionally, in the above technical solution, a model optimization module is further included, which is configured to optimize the intelligent suggestion generation model according to the feedback information of each preset role and / or newly obtained data about chronic kidney disease, specifically:

[0102] S401, align the preset role feedback information and the newly obtained clinical data to the medical ontology node through a medical entity recognition engine to generate incremental knowledge triples;

[0103] After receiving the preset role feedback information and newly acquired clinical data, the medical entity recognition engine is started, medical entities in the text such as diseases, symptoms, and test indicators are recognized, and are aligned with medical ontology nodes. After alignment, the system analyzes the relationships and attributes between entities, and combines the entities, relationships, and attributes into incremental knowledge triplets in accordance with the preset medical knowledge graph structure. The triplet form can be entity-relation-attribute. Subsequently, the system integrates these triplets into the existing medical knowledge graph, realizing incremental updating of knowledge. This process ensures the dynamic nature and timeliness of the knowledge graph, enabling it to continuously expand and enrich as new data is added.

[0104] S402, verify the logical consistency of the incremental triplets based on the clinical guideline knowledge base, and automatically assign optimization weights to conflicting data according to the authority of the data source;

[0105] The system verifies the logical consistency of the incremental triplets based on the clinical guideline knowledge base. First, the newly added triplets are compared with the existing knowledge in the knowledge base to check for logical conflicts or inconsistencies. For the conflicting data found, the system assigns weights according to the authority of the data source. Higher authority data sources (such as official clinical guidelines and authoritative medical journals) are given higher weights, while lower authority data sources are given lower weights. The optimized results will replace the original conflicting data through a weighted algorithm to ensure the accuracy and consistency of the knowledge base. This process is achieved through an automated process, ensuring the efficiency and reliability of knowledge updating.

[0106] S403, extract the parameter subgraph related to the incremental triplets from the intelligent suggestion generation model, and use the weighted incremental data to perform low-rank adaptive training (LoRA) to update the subgraph parameters;

[0107] The parameter subgraph related to the incremental triplets is extracted from the intelligent suggestion generation model. The weighted incremental data is input into the subgraph as training samples. Through low-rank adaptive training (LoRA), only the low-rank parameters related to the incremental triplets are updated while keeping most of the original model parameters unchanged. According to the characteristics of the incremental data, these low-rank parameters are automatically adjusted to optimize the model's adaptability to new knowledge. Finally, the updated subgraph parameters are integrated back into the original intelligent suggestion generation model, improving the model's representation ability for incremental knowledge and realizing efficient updating of the intelligent suggestion generation model.

[0108] S404, integrate the optimized subgraph parameters into the original intelligent suggestion generation model, test the key clinical indicators (suggestion compliance rate, logical error rate) in a sandbox environment, and when the indicators improve by ≥5%, deploy a new intelligent suggestion generation model.

[0109] The system integrates the optimized subgraph parameters into the original intelligent suggestion generation model, which involves updating and merging model parameters. Next, the system conducts a series of rigorous tests on the updated model in a sandbox environment, focusing on key clinical indicators such as suggestion compliance rate (i.e., the consistency of model-generated suggestions with clinical guidelines) and logical error rate (i.e., logical inconsistencies or errors in model suggestions). During the testing phase, the system uses a representative set of clinical cases and datasets to simulate actual usage scenarios to evaluate the performance of the model. When the test results show that the key clinical indicators have improved by ≥5% compared to the baseline model, the system triggers the deployment process to deploy the new intelligent suggestion generation model to the production environment. This deployment process needs to ensure the stability and reliability of the new model while minimizing interference with existing services. The system conducts final verification to ensure that the deployed model can consistently provide high-quality suggestions in actual applications. When all verification steps are completed, the new model will be officially put into use, providing a more accurate and reliable intelligent suggestion generation service for pre-set roles.

[0110] Optionally, in the above technical solution, further comprising a labeled corpus acquisition module, the labeled corpus acquisition module is used to collect epidemiological survey data, disease prevention and control center data and actual cases about chronic kidney disease, and obtain the labeled corpus about chronic kidney disease after expert labeling.

[0111] Epidemiological survey data, disease prevention and control center data and actual cases of chronic kidney disease are collected to ensure the comprehensiveness and accuracy of the data. Then, medical experts are organized to label the collected data, and the labeling content includes key information such as disease symptoms, diagnosis results and treatment plans. Experts follow uniform labeling guidelines to ensure the consistency and reliability of labeling. The labeled data is reviewed and verified to remove errors and inconsistencies, and finally integrated into a structured labeled corpus to provide high-quality data support for subsequent intelligent suggestion generation model training.

[0112] Through another embodiment, a chronic kidney disease comprehensive management system based on an artificial intelligence large model is described as follows:

[0113] At present, due to the shortage of specialist physicians, the limited diagnosis and treatment ability of primary care physicians, the low early identification rate of patients and the poor management efficiency. Although the existing general large language model has language understanding ability, it lacks targeted guidance in the field of kidney disease and cannot meet the needs of fine and individualized diagnosis and treatment. Traditional kidney disease diagnosis and treatment relies on expert experience and manual judgment, which is low in efficiency and difficult to be widely promoted to primary medical institutions.

[0114] The application provides a chronic kidney disease comprehensive management system based on an artificial intelligence large model. The system is based on a professional knowledge graph, clinical guidelines and real world data, and realizes intelligent management of the whole process from early screening, accurate diagnosis, individualized treatment to follow-up of chronic diseases through three-end (community doctor end, specialist doctor end and patient end) application. Through a "1+30+300" three-level collaborative network architecture, the application can realize the sinking of high-quality medical resources, and promote the realization of accurate diagnosis and treatment and intelligent decision-making in primary medical institutions, specifically including:

[0115] 1) The "Zhilu" large model is used as a preset artificial intelligence large model. The "Zhilu" large model structure is constructed by integrating the latest chronic kidney disease diagnosis and treatment guidelines and real world data, and a set of intelligent decision support system conforming to the prevention and treatment needs of chronic kidney disease, i.e. a chronic kidney disease comprehensive management system based on an artificial intelligence large model.

[0116] 2) Community doctor end: providing evidence-based decision support for primary doctors to help develop standard kidney disease management programs, including drug use, diagnosis and treatment process, etc., to improve the quality and efficiency of primary medical services. At this time, the preset role is a community doctor, and the community doctor end refers to a client provided for community doctors to call an intelligent suggestion generation model.

[0117] 3) Specialist doctor end: through the interface with the hospital HIS system, the patient's kidney disease related information is automatically extracted to assist the specialist doctor in accurate diagnosis and treatment, and the prescription and treatment plan can also be optimized. At this time, the preset role is a specialist doctor, and the specialist doctor end refers to a client provided for specialist doctors to call an intelligent suggestion generation model.

[0118] 4) Patient end: providing health education and self-management tools for patients, intelligently interpreting disease indicators, and improving the self-management ability and health awareness of patients. At this time, the preset role is a patient, and the patient end refers to a client provided for specialist doctors to call an intelligent suggestion generation model.

[0119] 5) Three-level collaborative network: through the cooperation of "1 national center + 30 top hospitals + 300 county medical institutions", a full-coverage chronic kidney disease management system is constructed to realize effective linkage from third-grade class-A hospitals to primary hospitals and promote the sinking of medical resources.

[0120] The beneficial effects of the application are as follows:

[0121] 1) Data-driven and intelligent decision-making: by integrating medical big data and the latest kidney disease prevention and treatment guidelines, an efficient and intelligent decision support platform is formed to improve the scientificity and accuracy of chronic kidney disease diagnosis and treatment.

[0122] 2) Multi-terminal collaboration to enhance primary medical capabilities: Ports are designed for different medical roles (i.e. different preset roles, including community doctors, specialists and patients), making multi-party collaboration possible and promoting the sinking of high-quality medical resources to the grassroots level.

[0123] 3) Full-cycle management and precise diagnosis and treatment: Through intelligent means, full-cycle coverage from early disease detection to chronic stage management is achieved, promoting precise diagnosis and treatment and long-term management of patients with chronic kidney disease.

[0124] 4) Realize intelligent management of chronic kidney disease throughout the entire cycle, across multiple roles, and in multiple scenarios; significantly improve the specialized diagnosis and treatment capabilities of primary care physicians; alleviate the pressure on tertiary hospitals and promote the downward flow of high-quality medical resources; enhance the self-management capabilities of chronic kidney disease patients and improve disease outcomes; and build a closed loop of disease knowledge with good scalability and replication and promotion value.

[0125] 5) This invention can significantly improve the diagnosis and treatment level of primary medical institutions and optimize the allocation of medical resources in the promotion of chronic kidney disease prevention and treatment at the grassroots level. Empowered by intelligent technology, it can be widely promoted in the future, helping to promote the development of chronic kidney disease prevention and treatment.

[0126] 6) This invention deeply integrates artificial intelligence with chronic disease management to build a new medical ecosystem that integrates intelligent decision-making, resource collaboration and patient education, breaking through the diagnosis and treatment bottlenecks in the traditional medical system. It has high innovation, practicality and promotion value, and can effectively improve the prevention and treatment of chronic kidney disease and improve the quality of life of patients.

[0127] like Figure 2 As shown, a comprehensive management method for chronic kidney disease based on an artificial intelligence large model according to an embodiment of the present invention includes the following steps:

[0128] S1. Based on the annotated corpus about chronic kidney disease, the preset artificial intelligence model is trained to obtain an intelligent suggestion generation model;

[0129] S2, encapsulate the intelligent suggestion generation model into an embeddable service;

[0130] S3. Deploy the embeddable service to the terminal of each preset role, so that the terminal of each preset role generates personalized recommendations about chronic kidney disease through the embeddable service.

[0131] Optionally, in the above technical solution, the following is further included:

[0132] Design an intelligent operation interface that is adapted to each preset role;

[0133] Through each intelligent operation interface, data about chronic kidney disease input by the corresponding preset role is received, and a corresponding personalized suggestion about chronic kidney disease is generated by using the embeddable service.

[0134] Optionally, in the above technical solution, it further includes guiding each preset role to input data about chronic kidney disease in an interactive manner.

[0135] Optionally, in the above technical solution, it further includes optimizing the intelligent suggestion generation model according to feedback information of each preset role and / or newly obtained data about chronic kidney disease.

[0136] Optionally, in the above technical solution, it further includes collecting epidemiological survey data, disease prevention and control center data, and actual cases about chronic kidney disease, and obtaining an annotated corpus about chronic kidney disease after expert annotation.

[0137] In another embodiment, the following steps are included:

[0138] S1001, data collection: collect structured and unstructured medical data including chronic kidney disease epidemiological survey data, disease prevention and control center data, and actual cases of chronic disease management, and obtain an annotated corpus about chronic kidney disease after expert annotation (also referred to as a chronic kidney disease specialist expert annotated corpus).

[0139] S1002, model training: train a preset artificial intelligence large model (also referred to as a specialist large model) using the annotated corpus about chronic kidney disease, and optimize the generation quality and reasoning ability.

[0140] S1003, deployment of intelligent suggestion generation model: through API interfacing with a hospital information system (HIS), the intelligent suggestion generation model is encapsulated as an embeddable service and deployed on different role terminals to realize end-side local reasoning or cloud-side calling.

[0141] S1004, system building: through deep learning and natural language processing technology, the condition of a chronic kidney disease patient is automatically analyzed, that is, through the intelligent suggestion generation model, the condition of a chronic kidney disease patient is automatically analyzed, and real-time decision support is provided by combining the latest medical literature and guidelines to help doctors make scientific and personalized diagnosis and treatment decisions (personalized suggestions include), and a full-cycle chronic kidney disease data support system, i.e., a chronic kidney disease comprehensive management system based on an artificial intelligence large model, is built.

[0142] S1005, multi-role and multi-scene design: according to different medical scenarios and role requirements, an adaptive intelligent operation interface is designed to enable doctors and patients to conveniently and quickly obtain the required information and make decisions.

[0143] S1006, knowledge updating mechanism: a periodic synchronization mechanism is designed to access the latest guidelines and expert consensus at home and abroad; an incremental learning algorithm is used to continuously optimize the intelligent suggestion generation model and improve the diagnosis and treatment decision-making ability.

[0144] It should be noted that the beneficial effects of the chronic kidney disease comprehensive management method based on the artificial intelligence large model provided in the above embodiments are the same as those of the chronic kidney disease comprehensive management system based on the artificial intelligence large model, and will not be repeated here. Moreover, the method and system embodiments provided in the above embodiments belong to the same concept, and the specific implementation process is detailed in the system embodiment, which will not be repeated here.

[0145] In the above embodiments, although the steps are numbered S1, S2, etc., it is only a specific embodiment given by the present application, and those skilled in the art can adjust the execution order of S1, S2, etc. according to the actual situation, which is within the protection scope of the present application. It can be understood that in some embodiments, some or all of the above embodiments can be included.

[0146] Among them, the chronic kidney disease comprehensive management system based on the artificial intelligence large model of the present application can be a computer program (including program code) running in a computer device, for example, the chronic kidney disease comprehensive management system based on the artificial intelligence large model of the present application is an application software, which can be used to execute the corresponding steps in the chronic kidney disease comprehensive management method based on the artificial intelligence large model of the present application.

[0147] In some embodiments, the chronic kidney disease comprehensive management system based on the artificial intelligence large model of the present application can be implemented in a combination of software and hardware, for example, the chronic kidney disease comprehensive management system based on the artificial intelligence large model of the present application can be a hardware decoding processor form processor which is programmed to execute the chronic kidney disease comprehensive management method based on the artificial intelligence large model of the present application, for example, the hardware decoding processor form processor can use one or more application specific integrated circuits (ASICs), DSPs, programmable logic devices (PLDs), complex programmable logic devices (CPLDs), field programmable gate arrays (FPGAs) or other electronic components.

[0148] The modules involved in the embodiments of the present application can be implemented in the form of software or in the form of hardware. In some cases, the names of the modules do not constitute a limitation on the modules themselves.

[0149] An electronic device according to an embodiment of the present application includes a memory, a processor, and a computer program stored in the memory and executable on the processor, and the processor implements any of the above-mentioned chronic kidney disease comprehensive management methods based on the artificial intelligence large model when executing the computer program. That is, the electronic device according to an embodiment of the present application can include, but is not limited to, a processor and a memory; the memory is configured to store a computer program; and the processor is configured to execute the chronic kidney disease comprehensive management method based on the artificial intelligence large model shown in any of the embodiments of the present application by invoking the computer program.

[0150] In an optional embodiment, an electronic device is provided, as shown in Figure 3 Figure 3 The electronic device 4000 shown in the optional embodiment includes a processor 4001 and a memory 4003. The processor 4001 and the memory 4003 are connected, for example, through a bus 4002. Optionally, the electronic device 4000 can also include a transceiver 4004, which can be used for data interaction, such as data transmission and / or data reception, between the electronic device and other electronic devices. It should be noted that in actual applications, the transceiver 4004 is not limited to one, and the structure of the electronic device 4000 does not constitute a limitation on the embodiments of the present application.

[0151] The processor 4001 can be a CPU (Central Processing Unit, central processor), a general-purpose processor, a DSP (Digital Signal Processor, data signal processor), an ASIC (Application Specific Integrated Circuit, application specific integrated circuit), an FPGA (Field Programmable Gate Array, field programmable gate array) or other programmable logic devices, transistor logic devices, hardware components or any combination thereof. It can implement or execute various exemplary logical blocks, modules and circuits described in combination with the present disclosure. The processor 4001 can also be a combination of computing functions, such as one or more microprocessor combinations, combinations of DSP and microprocessor, etc.

[0152] ​The bus 4002 can include a path that transmits information between the above-described components. The bus 4002 can be a PCI (Peripheral Component Interconnect) bus or an EISA (Extended Industry Standard Architecture) bus, or the like. The bus 4002 can be divided into an address bus, a data bus, a control bus, or the like. For convenience of representation, Figure 3 The bus 4002 is represented by only one thick line, but it does not mean that there is only one bus or one type of bus.

[0153] The memory 4003 can be a ROM (Read Only Memory) or other type of static storage device that can store static information and instructions, a RAM (Random Access Memory) or other type of dynamic storage device that can store information and instructions, an EEPROM (Electrically Erasable Programmable Read Only Memory), a CD-ROM (Compact Disc Read Only Memory) or other optical disk storage, a magnetic disk storage medium or other magnetic storage device, or any other medium that can be used to carry or store desired program code in the form of instructions or data structures and that can be accessed by a computer, but is not limited thereto.

[0154] The memory 4003 is used to store application code (computer program) for executing the scheme of the present application, and is controlled by the processor 4001 to execute. The processor 4001 is used to execute the application code stored in the memory 4003 to realize the content shown in the foregoing method embodiments.

[0155] Among them, the electronic device can also be a terminal device, and the terminal device can be any device that can install an application, including at least one of a smart phone, a tablet computer, a notebook computer, a desktop computer, a smart speaker, a smart watch, a smart television, and a smart vehicle device.

[0156] It should be noted that, Figure 3 The electronic device shown is only an example and should not limit the functions and use range of the embodiments of the present application.

[0157] The computer readable storage medium of the embodiment of the application, the computer readable storage medium stores a computer program, and the computer program is executed by a processor to implement any one of the above chronic kidney disease comprehensive management methods based on the artificial intelligence large model.

[0158] Optionally, the computer readable storage medium can be a read-only memory (Read-Only Memory, ROM), a random access memory (Random Access Memory, RAM), a read-only compact disc (Compact Disc Read-Only Memory, CD-ROM), a magnetic tape, a floppy disk and an optical data storage device, etc.

[0159] In the exemplary embodiments, a computer program product or computer program is also provided, which includes computer instructions stored in a computer readable storage medium. The processor of the electronic device reads the computer instructions from the computer readable storage medium, and the processor executes the computer instructions to make the electronic device execute any one of the above chronic kidney disease comprehensive management methods based on the artificial intelligence large model.

[0160] Computer program code for carrying out operations of the present application can be written in any combination of one or more programming languages, including an object-oriented programming language such as Java, Smalltalk, C++ or the like and conventional procedural programming languages, such as the "C" programming language or similar programming languages. The program code can execute entirely on the user's computer, partly on the user's computer, as a stand-alone software package, partly on the user's computer and partly on a remote computer or entirely on the remote computer or server. In the latter scenario, the remote computer can be connected to the user's computer through any type of network, including a local area network (LAN) or a wide area network (WAN), or the connection can be made to an external computer (for example, through the Internet using an Internet Service Provider).

[0161] It should be understood that the flow diagrams and block diagrams in the accompanying drawings illustrate the architecture, functionality, and operation of possible implementations of various embodiments of the present application. In this regard, each block in the flow diagrams and block diagrams can represent a module, segment, or portion of code, which comprises one or more executable instructions for implementing the specified logical functions. It should also be noted that in some alternative implementations, the functions noted in the blocks can occur out of the order noted in the figures. For example, two blocks shown in succession may, in fact, be executed substantially concurrently or the blocks can sometimes be executed in the reverse order, depending upon the functionality involved. It will also be noted that each of the blocks of the block diagrams and / or flowchart illustrations, and combinations thereof, can be implemented by a dedicated hardware-based system that performs the specified functions or operations, or combinations of hardware and software.

[0162] The computer readable storage medium of embodiments of the present application can be, but is not limited to, an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any suitable combination of the foregoing. More specific examples of the computer readable storage medium can include, but are not limited to, an electrical connection having one or more wires, a portable computer diskette, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or Flash memory), an optical fiber, a portable compact disc read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing. In the present application, the computer readable storage medium can be any tangible medium that can contain or store a program for use by or in connection with an instruction execution system, apparatus, or device.

[0163] The computer readable storage medium described above can bear one or more programs, which, when executed by the electronic device, cause the electronic device to perform the method shown in the above embodiments.

[0164] The above description merely provides preferred embodiments of the present application and a principle of applied technology. It should be understood by those skilled in the art that the disclosed range of the present application is not limited to the technical solutions formed by the specific combinations of the above technical features, and should also cover other technical solutions formed by any combinations of the above technical features or equivalent features without departing from the disclosed concept. For example, the above technical features can be replaced with the technical features disclosed in the present application (but not limited to) having similar functions to form technical solutions.

[0165] It should be noted that the terms "first", "second", etc. in the specification and claims of the present application are used to distinguish similar objects, and do not represent a specific order or sequential order. The order of use of similar objects can be interchanged under appropriate circumstances, so that the embodiments of the present application described herein can be implemented in an order other than that illustrated or described.

[0166] Those skilled in the art know that the present application can be implemented as a system, a method or a computer program product, so the present application can be specifically implemented as follows: it can be a complete hardware, a complete software (including firmware, resident software, microcode, etc.), or a combination of hardware and software, which is generally referred to as "circuit", "module" or "system" herein. In addition, in some embodiments, the present application can also be implemented as a computer program product in one or more computer readable media, which contains computer readable program code.

[0167] Although the embodiments of the present application have been shown and described above, it should be understood that the above-described embodiments are exemplary and should not be construed as limiting the present application, and those skilled in the art can make changes, modifications, replacements and variations to the above-described embodiments within the scope of the present application.

Claims

1. A comprehensive management system for chronic kidney disease based on an artificial intelligence large model, characterized by: Includes model training module, encapsulation module and personalized suggestion generation module; The model training module is used to: train a preset artificial intelligence large model based on a labeled corpus about chronic kidney disease to obtain an intelligent suggestion generation model; The encapsulation module is used to: encapsulate the intelligent suggestion generation model into an embeddable service; The personalized suggestion generating module is used to deploy the embeddable service to the terminal of each preset role, so that the terminal of each preset role generates personalized suggestions on chronic kidney disease through the embeddable service.

2. The comprehensive management system for chronic kidney disease based on artificial intelligence large model according to claim 1 is characterized in that: The module also includes an operation interface design module, which is used to design an intelligent operation interface adapted to each preset role; The personalized suggestion generation module is further used to: receive data on chronic kidney disease input by the corresponding preset role through each intelligent operation interface, and use the embeddable service to generate corresponding personalized suggestions on chronic kidney disease.

3. The comprehensive management system for chronic kidney disease based on artificial intelligence large model according to claim 2 is characterized in that: The system further includes a guiding module, which is used to interactively guide each preset role to input data related to chronic kidney disease.

4. A comprehensive management system for chronic kidney disease based on an artificial intelligence large model according to any one of claims 1 to 3, characterized in that: It also includes a model optimization module, which is used to optimize the intelligent suggestion generation model based on feedback information of each preset role and / or newly acquired data on chronic kidney disease.

5. A comprehensive chronic kidney disease management system based on an artificial intelligence large model according to any one of claims 1 to 3, characterized in that: It also includes a marked corpus acquisition module, which is used to collect epidemiological survey data, disease prevention and control center data and actual cases on chronic kidney disease, and obtain the marked corpus on chronic kidney disease after expert annotation.

6. A comprehensive management method for chronic kidney disease based on an artificial intelligence large model, characterized by: include: Based on the annotated corpus about chronic kidney disease, the preset artificial intelligence model is trained to obtain an intelligent suggestion generation model; Encapsulating the intelligent suggestion generation model as an embeddable service; The embeddable service is deployed to a terminal of each preset role, so that the terminal of each preset role generates personalized advice on chronic kidney disease through the embeddable service.

7. A comprehensive management method for chronic kidney disease based on an artificial intelligence large model according to claim 6, characterized in that: Also includes: Design an intelligent operation interface that is adapted to each preset role; Through each intelligent operation interface, data on chronic kidney disease input by the corresponding preset role is received, and corresponding personalized suggestions on chronic kidney disease are generated using the embeddable service.

8. The comprehensive management method for chronic kidney disease based on artificial intelligence large model according to claim 7 is characterized in that: Also includes: Each pre-set role is interactively guided through entering data about chronic kidney disease.

9. An electronic device, characterized in that: It includes a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, it implements a comprehensive management method for chronic kidney disease based on an artificial intelligence large model as described in any one of claims 6 to 8.

10. A computer-readable storage medium, characterized in that The computer-readable storage medium stores a computer program, which, when executed by a processor, implements the comprehensive management method for chronic kidney disease based on an artificial intelligence large model as described in any one of claims 6 to 8.

Images